Tutorials › VEE Economics › Chicken Games and Strategic Brinkmanship

VEE Economics · Game Theory and Strategic Choice · Tutorial 389 of 998

Chicken Games and Strategic Brinkmanship

Use payoff matrices to identify chicken-game equilibria and calculate how often each player escalates in a mixed strategy.

Advanced 11 min read

What You'll Learn

  • Recognize the incentives that define a chicken game.
  • Find and verify its two asymmetric pure-strategy equilibria.
  • Derive each player’s mixing probability from the opponent’s payoffs.
  • Analyze mixed equilibria when players face different escalation stakes.
  • Explain how changing the payoff from escalation changes equilibrium behavior.
  • Avoid confusing a player’s mixing probability with the probability that the opponent escalates.

Introduction

In a chicken game, each of two decision-makers can stand firm or yield. Standing firm can be advantageous if the other yields, but disastrous if both stand firm. This tension creates a strategic problem: neither player wants mutual escalation, yet each may prefer to escalate rather than yield when the other appears likely to yield.

The important feature is not the story used to describe the choices. It is the pattern of incentives in the payoff matrix. A chicken game has two asymmetric pure-strategy equilibria: one player stands firm while the other yields, and vice versa. It can also have a mixed-strategy equilibrium in which each player randomizes. The analysis is aligned with the SOA VEE Economics (Validation by Educational Experience) syllabus.

Defining a Chicken Game

Call the two actions Escalate and Yield. Payoffs are ordered as (Player 1, Player 2), and a larger payoff is preferred. Consider a symmetric game with the following payoffs:

$$ \begin{array}{c|cc} & \text{Player 2: Escalate} & \text{Player 2: Yield} \\ \hline \text{Player 1: Escalate} & (-K,-K) & (G,-D) \\ \text{Player 1: Yield} & (-D,G) & (0,0) \end{array} $$

Here, \(G\) is the gain from being the only player to escalate, \(D\) is the loss from yielding to an escalator, and \(K\) is the loss to each player if both escalate. The standard chicken incentives require \(G>0\) and \(K>D>0\). These conditions mean that escalating is better than yielding against a player who yields, while yielding is better than mutual escalation against a player who escalates. The loss from mutual escalation is therefore worse than the loss from yielding, but unilateral escalation is attractive.

Definition: A chicken game is a strategic-form game in which each player prefers to escalate when the opponent yields, prefers to yield when the opponent escalates, and both players receive a particularly bad outcome if they both escalate. Under the payoff conditions \(G>0\) and \(K>D>0\), neither action is a dominant strategy.

The game is symmetric when the players have the same payoffs and incentives. The strategic pattern can also occur with different payoffs for the two players. In that case, each player may face a different loss from mutual escalation, gain from unilateral escalation, or cost of yielding.

Pure-Strategy Equilibria

A pure-strategy Nash equilibrium is an outcome at which neither player can improve their payoff by changing their own action alone. In the chicken-game matrix, if Player 2 yields, Player 1 prefers to escalate: \(G>0\). Given that Player 1 escalates, Player 2 prefers to yield: \(-D>-K\), because \(K>D\). Thus (Escalate, Yield) is an equilibrium. By the same reasoning, (Yield, Escalate) is also an equilibrium.

Neither (Escalate, Escalate) nor (Yield, Yield) is a pure-strategy equilibrium under these strict conditions. At (Escalate, Escalate), either player can improve by yielding, since \(-D>-K\). At (Yield, Yield), either player can improve by escalating, since \(G>0\). The two pure equilibria assign the roles differently; the matrix alone does not select which player will yield.

Worked Example: Finding Both Pure Equilibria

Two fictional firms are deciding whether to launch a costly promotional campaign or hold back. Their payoffs, in points, follow the symmetric chicken-game pattern: mutual campaigns give each firm \(-100\); if only one campaigns, that firm gets \(40\) and the other gets \(-20\); if both hold back, each gets \(0\). Identify all pure-strategy equilibria and explain why the other outcomes are not equilibria.

Method. Compare each firm’s payoffs from its two actions while holding the other firm’s action fixed. Here \(G=40\), \(D=20\), and \(K=100\), so \(G>0\) and \(K>D>0\).

If Firm 2 holds back, Firm 1 compares \(40\) from campaigning with \(0\) from holding back and chooses to campaign. Given that Firm 1 campaigns, Firm 2 compares \(-20\) from holding back with \(-100\) from campaigning and chooses to hold back. Therefore (Campaign, Hold back) is an equilibrium.

Reversing the firms gives the other equilibrium, (Hold back, Campaign): the campaigning firm earns \(40\), and the firm that holds back earns \(-20\).

If both campaign, either firm can switch to holding back and improve its payoff from \(-100\) to \(-20\), so that outcome is not an equilibrium. If both hold back, either can campaign and improve its payoff from \(0\) to \(40\), so that outcome is not an equilibrium. Check: exactly the two asymmetric outcomes are mutual best responses.

Deriving the Mixed-Strategy Equilibrium

A mixed strategy assigns probabilities to a player’s actions. Assume the players choose their randomizations independently and that each maximizes expected payoff. In a fully mixed equilibrium, each player gives the other player a probability that makes the other indifferent between escalating and yielding. If the opponent were not indifferent, they would choose one action for sure rather than assign positive probability to both.

For the symmetric game, let \(p\) be the probability that the opponent escalates. The expected payoff from escalating is \(p(-K)+(1-p)G\). The expected payoff from yielding is \(p(-D)+(1-p)0\). Equating the two gives the probability at which the player is indifferent:

$$ p(-K)+(1-p)G=p(-D) $$

Rearrange the equation by expanding the left side and collecting the terms involving \(p\):

$$ -pK+G-pG=-pD \quad\Longrightarrow\quad G=p(G+K-D) \quad\Longrightarrow\quad p=\frac{G}{G+K-D}. $$

The denominator is greater than \(G\), because \(K-D>0\); therefore \(0<p<1\). In a symmetric fully mixed equilibrium, each player escalates with this probability. The formula identifies the opponent’s escalation probability that makes a player indifferent; in a symmetric game, both players use the same probability.

Result: In a symmetric chicken game with \(G>0\) and \(K>D>0\), each player escalates with probability \(\frac{G}{G+K-D}\) in the fully mixed equilibrium. The game also has the two asymmetric pure-strategy equilibria described above.

Worked Example: Checking the Symmetric Mixed Equilibrium

For the fictional firms in the previous example, find the mixed-strategy equilibrium probability and verify that each firm is indifferent between campaigning and holding back.

The values are \(G=40\), \(K=100\), and \(D=20\). The probability of campaigning is:

$$ p=\frac{40}{40+100-20} =\frac{40}{120} =\frac{1}{3}. $$

Suppose the other firm campaigns with probability \(1/3\) and holds back with probability \(2/3\). The expected payoff from campaigning is:

$$ \frac{1}{3}(-100)+\frac{2}{3}(40) =-\frac{100}{3}+\frac{80}{3} =-\frac{20}{3}. $$

The expected payoff from holding back is:

$$ \frac{1}{3}(-20)+\frac{2}{3}(0) =-\frac{20}{3}. $$

Both actions therefore have the same expected payoff, \(-20/3\) points. By symmetry, both firms use the same probability. Check: substituting \(p=1/3\) into either expected-payoff calculation gives \(-20/3\), as required for indifference.

Different Players, Different Mixing Probabilities

When payoffs are asymmetric, the probability that makes Player 1 indifferent need not equal the probability that makes Player 2 indifferent. Use the general matrix below, with \(K_i\), \(G_i\), and \(D_i\) denoting the mutual-escalation loss, unilateral-escalation gain, and yielding loss for Player \(i\):

$$ \begin{array}{c|cc} & \text{Player 2: Escalate} & \text{Player 2: Yield} \\ \hline \text{Player 1: Escalate} & (-K_1,-K_2) & (G_1,-D_2) \\ \text{Player 1: Yield} & (-D_1,G_2) & (0,0) \end{array} $$

Assume \(G_i>0\) and \(K_i>D_i>0\) for each player. Let \(q\) be the probability that Player 2 escalates and \(p\) the probability that Player 1 escalates. Making Player 1 indifferent gives \(q=G_1/(G_1+K_1-D_1)\). Making Player 2 indifferent gives \(p=G_2/(G_2+K_2-D_2)\). The key is to solve each player’s indifference equation using the opponent’s probability.

Worked Example: Asymmetric Stakes and Opponent Probabilities

Two fictional organizations are considering whether to pursue an aggressive negotiating position or yield. Their payoff matrix is:

$$ \begin{array}{c|cc} & \text{Organization 2: Escalate} & \text{Organization 2: Yield} \\ \hline \text{Organization 1: Escalate} & (-90,-80) & (30,-20) \\ \text{Organization 1: Yield} & (-30,40) & (0,0) \end{array} $$

Find the two pure equilibria and the fully mixed equilibrium. For Organization 1, the payoff parameters are \(G_1=30\), \(K_1=90\), and \(D_1=30\). For Organization 2, they are \(G_2=40\), \(K_2=80\), and \(D_2=20\). Both satisfy the chicken-game conditions.

If Organization 2 yields, Organization 1 prefers escalation (\(30>0\)); if Organization 1 escalates, Organization 2 prefers yielding (\(-20>-80\)). Thus (Escalate, Yield) is a pure equilibrium. Reversing the roles gives the other pure equilibrium, (Yield, Escalate). The two remaining outcomes are not equilibria: mutual escalation invites either player to yield, while mutual yielding invites either player to escalate.

Let \(q\) be the probability that Organization 2 escalates. Organization 1 is indifferent when:

$$ q=\frac{30}{30+90-30} =\frac{30}{90} =\frac{1}{3}. $$

Let \(p\) be the probability that Organization 1 escalates. Organization 2 is indifferent when:

$$ p=\frac{40}{40+80-20} =\frac{40}{100} =0.4. $$

Check Organization 1’s payoffs when \(q=1/3\). Escalating gives \((1/3)(-90)+(2/3)(30)=-30+20=-10\). Yielding gives \((1/3)(-30)+(2/3)(0)=-10\).

Check Organization 2’s payoffs when \(p=0.4\). Escalating gives \((0.4)(-80)+(0.6)(40)=-32+24=-8\). Yielding gives \((0.4)(-20)+(0.6)(0)=-8\). Thus Organization 1 escalates with probability \(0.4\), while Organization 2 escalates with probability \(1/3\). Check: each player’s two expected payoffs agree at the opponent’s calculated probability.

How Escalation Incentives Shape Choices

The mixed-strategy formula shows how the payoff structure affects the rate of escalation in equilibrium. In a symmetric game, \(p=G/(G+K-D)\). Holding the other payoffs fixed, a larger unilateral gain \(G\) raises \(p\), while a larger mutual-escalation loss \(K\) lowers \(p\). A larger yielding loss \(D\) raises \(p\): if yielding when the opponent escalates becomes more costly, a player must face a higher escalation probability to be indifferent.

These are comparisons of equilibrium probabilities, not guarantees about a particular play. A mixed strategy describes randomization across plays under the model; it does not mean that the players are confused or that the outcome must be mutual escalation. Nor does reducing the equilibrium probability remove the two asymmetric pure equilibria as long as the strict chicken-game payoff conditions continue to hold.

Worked Example: Reducing the Gain From Unilateral Escalation

In a hypothetical negotiation, two identical parties receive payoffs measured in thousands of dollars. Mutual escalation gives each \(-96\), unilateral escalation gives the escalator \(24\) and the party that yields \(-16\), and mutual yielding gives each \(0\). A revised settlement arrangement reduces the unilateral gain from escalation to \(12\), leaving the other payoffs unchanged. Compare the mixed-equilibrium escalation probabilities before and after the revision.

Initially, \(G=24\), \(K=96\), and \(D=16\), satisfying \(G>0\) and \(K>D>0\). Each party’s equilibrium probability of escalation is:

$$ p=\frac{24}{24+96-16} =\frac{24}{104} =\frac{3}{13} \approx 0.2308. $$

After the revision, \(G=12\), while \(K=96\) and \(D=16\). The new probability is:

$$ p=\frac{12}{12+96-16} =\frac{12}{92} =\frac{3}{23} \approx 0.1304. $$

The equilibrium escalation probability falls from \(0.2308\) to \(0.1304\), or from about 23.08% to 13.04%. Both values are rounded to four decimal places. Check: the revised payoff still has \(G>0\) and \(K>D>0\), so the chicken-game incentives and the two asymmetric pure equilibria remain.

Common Mistakes and Exam Traps

  • Using the wrong player’s probability. To make Player 1 indifferent, solve for the probability that Player 2 escalates. In the asymmetric example, this probability is \(1/3\), not \(0.4\). The \(0.4\) value is Player 1’s probability, found from Player 2’s indifference condition.
  • Treating mutual escalation as an equilibrium. A disastrous outcome is not automatically an equilibrium. Check whether either player can improve by changing action alone; here, yielding improves a payoff from \(-K\) to \(-D\).
  • Calling mutual yielding an equilibrium because both avoid the worst outcome. If \(G>0\), either player can gain by escalating when the other yields. Mutual yielding is therefore not stable under the stated conditions.
  • Assuming the two pure equilibria are identical in their distribution of payoffs. They assign the escalator and the yielding player to different parties. With asymmetric payoffs, the consequences of those roles may differ.

A reliable analysis starts by checking the payoff conditions, then tests best responses in each pure outcome. For a mixed equilibrium, write each player’s expected payoff from both actions and solve the indifference equation. This sequence helps prevent both matrix-reading errors and the common mistake of assigning a player their own mixing probability in the wrong indifference calculation.

Check Your Understanding

For each problem, use the stated payoffs and interpret probabilities as the chance that a player chooses Escalate.

  1. A symmetric game has \(G=20\), \(K=70\), and \(D=10\). State its pure-strategy equilibria and calculate each player’s mixed-equilibrium probability of escalation. (Answer: (Escalate, Yield) and (Yield, Escalate); \(p=1/4=0.2500\).)
  2. A symmetric game has \(G=30\), \(K=60\), and \(D=15\). Calculate the mixed-equilibrium escalation probability and each player’s expected payoff there. (Answer: \(p=0.4000\); expected payoff \(-6\) payoff units per player.)
  3. In an asymmetric game, Player 1 has \(G_1=30\), \(K_1=75\), \(D_1=15\), and Player 2 has \(G_2=25\), \(K_2=100\), \(D_2=25\). Find the probability each player escalates in the fully mixed equilibrium. (Answer: Player 1 escalates with probability \(1/4=0.2500\); Player 2 escalates with probability \(1/3=0.3333\).)
  4. A symmetric game has mutual-escalation payoffs of \(-50\) each, unilateral-escalation payoffs of \(18\) for the escalator and \(-12\) for the player who yields, and mutual-yielding payoffs of \(0\). Identify all pure equilibria and state why mutual escalation is not one. (Answer: (Escalate, Yield) and (Yield, Escalate); at mutual escalation either player improves by yielding, from \(-50\) to \(-12\).)
  5. In a symmetric game, \(G=30\) and \(D=20\). If \(K\) increases from \(80\) to \(140\), calculate the mixed-equilibrium escalation probability in each case and describe the change. (Answer: it falls from \(30/(30+80-20)=1/3=0.3333\) to \(30/(30+140-20)=1/5=0.2000\).)
Clinical Biostats