Below are 10 exam-style questions on game theory, all drawn from Chapter 17 (“Oligopoly”) of your
Mankiw text. Questions 1–5 test core concepts; Questions 6–10 are more advanced and require
deeper analysis.
Basic Questions (1–5)
1. Prisoner’s Dilemma Payoffs
Two firms repeatedly face the choice to Cooperate (maintain high prices) or Cheat (cut price). A
single-period payoff matrix (profits in $M) is:
Firm B: Coop Firm B: Cheat
A: Cooperate (5, 5) (2, 8)
A: Cheat (8, 2) (4, 4)
Question: Identify the unique Nash equilibrium and explain why it is a prisoner’s dilemma.
Answer: Both firms Cheat (4,4). Each firm’s dominant strategy is Cheat (8>5 if the other Coops; 4>2 if
the other Cheats). Mutual cheating yields lower payoffs than mutual cooperation (4<5), capturing
the classic dilemma.
2. Cournot Best Response
Two identical firms choose quantities q1q_1 and q2q_2. Market price is P=100−(q1+q2)P = 100 - (q_1
+ q_2); constant MC = 20 for each.
Question: Derive firm 1’s best-response function q1(q2)q_1(q_2).
order: 100−2q1−q2−20=0100 - 2q_1 - q_2 - 20 = 0 ⇒ q1=80−q22q_1 = \tfrac{80 - q_2}{2}.
Answer: Firm 1 maximizes π1=[100−(q1+q2)]q1−20q1\pi_1 = [100 - (q_1+q_2)]q_1 - 20q_1. First-
3. Cournot Equilibrium
Using the best-response from Q2, find the symmetric Cournot equilibrium (q1∗,q2∗)(q_1^*, q_2^*)
Answer: Symmetry ⇒ q1=q2=qq_1 = q_2 = q. Then q=(80–q)/2q = (80 – q)/2 ⇒ 3q=803q = 80 ⇒
and equilibrium price.
q∗=26.67q^* = 26.67. Price: P=100–2(26.67)=46.66P = 100 – 2(26.67) = 46.66.
4. Bertrand Paradox
Two firms sell identical products at zero marginal cost. They simultaneously set prices p1p_1, p2p_2.
Consumers buy from the cheaper; if equal, they split demand.
Question: What is the unique Nash equilibrium price and profit?
Answer: Both set price = MC = 0. Any price above 0 invites undercutting. Profit = 0. This is the
Bertrand paradox.
5. Stackelberg Leader
In the duopoly of Q3, now firm 1 moves first (chooses q1q_1), firm 2 observes and then chooses
q2q_2.
Question: Solve for Stackelberg equilibrium quantities (q1S,q2S)(q_1^{S},q_2^{S}).
Answer: Firm 2’s best response: q2=(80−q1)/2q_2 = (80 - q_1)/2. Leader chooses q1q_1 to maximize
π1=(100−q1−q2)q1−20q1\pi_1 = (100 - q_1 - q_2)q_1 - 20q_1. Substitute q2q_2:
q_1/2]q_1. FOC: 60−q1=060 - q_1 = 0 ⇒ q1S=60q_1^{S}=60, then q2S=(80−60)/2=10q_2^S = (80 -
π1=[100−q1−(80−q1)/2−20]q1\pi_1 = [100 - q_1 - (80 - q_1)/2 - 20]q_1. Simplify: = [60−q1/2]q1[60 -
60)/2 = 10.
Advanced Questions (6–10)
6. Mixed-Strategy Equilibrium in a Coordination Game
Two drivers approaching an unmarked intersection must each choose Left or Right. Payoff is +1 if
they pick the same, –10 if they pick differently. There is no preference for Left vs Right.
Question: Find the mixed-strategy Nash equilibrium (probability pp of choosing Left).
Answer: Let driver 2 play Left with prob pp. Driver 1 is indifferent when
1⋅p+(−10)⋅(1−p)=(−10)⋅p+1⋅(1−p),1\cdot p + (-10)\cdot(1-p) = (-10)\cdot p + 1\cdot(1-p),
solving p−10+10p=−10p+1−pp -10 +10p = -10p +1 -p ⇒ 11p−10=−11p+1−p11p -10 = -11p +1 -p ⇒
combine: 11p−10=−12p+111p -10 = -12p +1 ⇒ 23p=1123p=11 ⇒ p=11/23≈0.478p=11/23\
approx0.478.
7. Trigger Strategies in an Infinitely Repeated Game
Consider the Prisoner’s Dilemma payoffs from Q1 played infinitely. Discount factor is δ\delta. Firms
use the “grim-trigger” strategy: cooperate until any cheating, then defect forever.
Question: Derive the minimal δ\delta such that cooperation is sustained as a subgame-perfect
equilibrium.
Answer: Payoff to always Cooperate: ∑t=0∞5δt=5/(1−δ)\sum_{t=0}^\infty 5\delta^t = 5/(1-\delta). If
one cheats once, immediate gain = 8 (vs 5), then punishment yields 4 forever: total = 8+δ⋅4/(1−δ)8
by 1−δ1-\delta: 5≥8(1−δ)+4δ5 \ge 8(1-\delta) +4\delta ⇒ 5≥8−8δ+4δ5\ge8-8\delta+4\delta ⇒
+ \delta \cdot 4/(1-\delta). Need 5/(1−δ)≥8+4δ/(1−δ)5/(1-\delta) \ge 8 + 4\delta/(1-\delta). Multiply
5≥8−4δ5\ge8 -4\delta ⇒ 4δ≥34\delta\ge3 ⇒ δ≥0.75\delta\ge0.75.
8. Public Goods and Free Riding
Two individuals decide simultaneously how much to contribute gi≥0g_i\ge0 to a public good. Total
public good is G=g1+g2G=g_1+g_2. Each’s payoff is Ui=αln(G+1)−giU_i = \alpha\ln(G+1) - g_i, with
α>0\alpha>0.
Question: Find the Nash equilibrium contributions (g1∗,g2∗)(g_1^*,g_2^*).
alpha/(G+1) -1=0 ⇒ G∗=α−1G^* = \alpha -1. By symmetry g1∗=g2∗=(α−1)/2g_1^*=g_2^*=(\alpha -
Answer: Each solves maxgiαln(gi+gj+1)−gi\max_{g_i}\alpha\ln(g_i+g_j+1) - g_i. FOC: α/(G+1)−1=0\
1)/2.
9. Bertrand with Capacity Constraints
Two firms have identical MC=20 but limited capacity KK. Market demand: P=100−QP=100 - Q. They
set prices simultaneously. If one sets a lower price, it sells up to its capacity KK; the residual goes to
the other (if it priced as well).
Question: Describe conditions under which there is a pure-strategy equilibrium with both firms
setting the same price.
Answer: If K≥QC/2K\ge Q^C/2, where QCQ^C is competitive quantity (80), then capacity is large
enough; equilibrium price = MC. If K<QC/2K < Q^C/2, then with capacities binding, any price above
MC yields incentive to undercut until the residual demand equals capacity. Equilibrium entails prices
strictly above MC, found where residual demand = capacity.
10. Signaling in a Market for Lemons
Two types of sellers (High-quality H or Low-quality L) sell cars of value vHv_H or vLv_L to buyers who
cannot distinguish types. Buyers’ offer price is expected value E[v]E[v]. A seller can buy a $c
diagnostic report that reveals quality. Only H-type buys it.
Question: Show that for some cc, there exists a separating perfect-Bayesian equilibrium where only
H-types pay for the report and get price vHv_H, and L-types do not pay and get price E[v]E[v]. Derive
the bounds on cc.
Answer: In a separating equilibrium, H pays cc to get full information and sells at vHv_H, netting
vH−cv_H - c. It must prefer this over not paying and selling at E[v]E[v]:
vH−c ≥ E[v]⇒c≤vH−E[v].v_H - c \;\ge\; E[v] \quad\Rightarrow\quad c \le v_H - E[v].
Meanwhile, L-types must prefer not paying and selling at E[v]E[v] rather than paying and selling at
vLv_L:
E[v] ≥ vL−c⇒c≥vL−E[v].E[v] \;\ge\; v_L - c \quad\Rightarrow\quad c \ge v_L - E[v].
Thus a separating equilibrium exists if
vL−E[v] ≤ c ≤ vH−E[v].v_L - E[v] \;\le\; c \;\le\; v_H - E[v].
These 10 questions span core duopoly models, dynamic enforcement, public-goods contributions,
capacity constraints, and signaling—ranging from textbook applications to frontier analyses.