Scribd
Upload
84 views2 pages

EC 246: Decisions & Games

This document contains the answers to problems from an EC 246 course. It provides the correct mixed strategy Nash equilibria for an extended matching pennies game, and analyzes the Nash equilibria for a related game where players' utilities depend on whether they have a Bernoulli or square root utility function. It then solves two problems: 1) determining all mixed strategy Nash equilibria for a game where players choose heads, tails, or covered, and 2) solving a system of equations to find a mixed strategy Nash equilibrium for an insurance claim game.

Uploaded by

Parmesh Gupta
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
84 views2 pages

EC 246: Decisions & Games

This document contains the answers to problems from an EC 246 course. It provides the correct mixed strategy Nash equilibria for an extended matching pennies game, and analyzes the Nash equilibria for a related game where players' utilities depend on whether they have a Bernoulli or square root utility function. It then solves two problems: 1) determining all mixed strategy Nash equilibria for a game where players choose heads, tails, or covered, and 2) solving a system of equations to find a mixed strategy Nash equilibrium for an insurance claim game.

Uploaded by

Parmesh Gupta
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

EC 246: Decisions & Games

Answer Key III

Let be begin by presenting the correct answer to the first problem in Problem Set II. I am
grateful to two EC 246 students Divyanshu Gupta and Milind Mandlik for pointing out the
correct answer to me.
Specifically, in addition to the following Nash equilibria of the “Extended Matching Pennies
Game – 1”:
[Row plays cover, Column plays cover], [Row plays cover, Column plays {heads, tails; p,
1–p (for some p  [0.4, 0.6])}], [Row plays {heads, tails; p, 1–p (for some p  [0.4, 0.6])},
Column plays cover].
there are two other mixed-strategy Nash equilibria of the game that are mirror-images of each
other (and also give each player expected utility of 6):
[Row plays cover, Column plays {heads, tails, cover; 2/9, 3/9, 4/9}] and [Column plays
cover, Row plays {heads, tails, cover; 2/9, 3/9, 4/9}].
First, verify that the stated strategies are indeed best-responses to each other. Then verify
that if any player randomizes over {heads, tails, cover} with any set of non-negative
probabilities {p, q, (1– p – q)} that are different from {2/9, 3/9, 4/9}, then the other player
will strictly prefer not to play ‘cover’. And there does not exist a Nash equilibrium in the
game in which no player plays ‘cover’ with positive probability [think about a proof (by
contradiction) of this last statement!].

We now turn to Problem Set III.


________________________________________________
Problem 1: I have given you the outcome matrix which tells you how much money each
player gets depending on alternative outcomes.
When the players’ common Bernoulli utility function is: u(x) = x, then the following matrix
is also the players’ payoff-matrix.
Column Player
Heads Tails Covered
Row Player Heads 100, 00 00, 100 036, 036
Tails 00, 100 100, 00 36, 036
. Covered 036, 036 036, 036 036, 036

Of course, [covered, covered] is the only pure-strategy Nash equilibrium. Let me try to
explicitly determine all the mixed-strategy Nash equilibrium, to give students an idea of the
logic behind my successive arguments.
Can there exist a Nash equilibrium where Column randomizes over Heads and Tails with
probabilities x and (1–x)? Note that for x > 0.5, Row’s unique best response will be Heads;
for x < 0.5, Row’s unique best response will be Tails; and for x = 0.5, Row’s pure best
response set will be {Heads, Tails}. Use these facts to establish the following result: [each
1
EC 246: Decisions & Games

player plays {Heads, tails; 0.5, 0.5}] is the only mixed strategy equilibrium in which any one
player randomizes over Heads and Tails.
Can there exist a Nash equilibrium where Column randomizes over Heads and Cover with
probabilities y and (1–y)? Note that irrespective of the magnitude of y, Row’s unique best
response will be Heads. Use this fact to establish the following result: there is no mixed-
strategy equilibrium in which any player randomizes over Heads and Cover. Then extend the
argument to prove that: there is no mixed-strategy equilibrium in which any player
randomizes over Tails and Cover.
Finally, can there exist a Nash equilibrium where Column randomizes over Heads, Tails and
Cover with probabilities x, y, and (1–x–y)? Note that for Row, cover will never be a best-
response; the best response will be either only Heads, or only Tails, or {Heads, Tails} if and
only if x = y. Use these facts to establish the following result: there is no mixed-strategy
equilibrium in which any player randomizes over Heads, Tails, and Cover.
So, conclusion: there exists a Pareto-superior equilibrium in which each player randomizes
over Heads and Tails with equal probability (getting expected utility 50), and a Pareto-
inferior equilibrium in which each player plays Covered (getting expected utility 36).

In contrast, when the players’ common Bernoulli utility function is: v(x) = + x, then the
game’s payoff-matrix is exactly the payoff-matrix specified in Problem I of Problem Set II,
and therefore, the equilibrium set will be as described at the beginning of this Answer Key.

So, for the same game-outcome matrix, players’ risk attitudes do determine the set of
equilibrium Nash strategies, establishing which was the whole point of this exercise.

Problem 2: Solving for two unknowns {x, y} using any combination of the following three
equations:
(i) 11x + 12.075y + 12.6(1–x–y) = 10.925x + 12y + 13.125(1–x–y),
(ii) 10.925x + 12y + 13.125(1–x–y) = 11.4x + 11.875y + 13(1–x–y),
(iii) 11x + 12.075y + 12.6(1–x–y) = 11.4x + 11.875y + 13(1–x–y),

will give you the solution: x = 5/24 and y = 2/3.


Then, from the payoff-matrix note that if Column randomizes over the pure strategies: claim
11K, and claim 12K, the Row will strictly prefer to claim 11K; if Column randomizes over the
pure strategies: claim 11K, and claim 13K, the Row will strictly prefer to claim 13K; and if
Column randomizes over the pure strategies: claim 12K, and claim 13K, the Row will strictly
prefer to claim 12K. The symmetry of the game then establishes the uniqueness of the mixed
strategy specified in the Session 4 slides.

You might also like