Session 1: Review

Sciences Po

August 27, 2018

Sciences Po Session 1: Review August 27, 2018 1 / 30

Overview

1 Course overview

2 Useful concepts from CORE

Consumer Choice
Firm Production
Game Theory

Sciences Po Session 1: Review August 27, 2018 2 / 30

Outline

Social choice, eciency and welfare

Game theory and incentives
Institutions as mechanisms
Limits to eciency

Sciences Po Session 1: Review August 27, 2018 3 / 30

Organization

Two assigments. First assigment before the midterm and second

assignment before the nal.
One midterm (tentative date: 13th of October)
One nal exam
Each is (33% of nal grade). Additionally, there will be a multiple answers
quiz on Moodle between each lecture and the next conference. It will
consist in simple questions on what has been presented in the lecture.
Completing all quizzes will give you a bonus point on the assignements
grade.

Sciences Po Session 1: Review August 27, 2018 4 / 30

 Consumer Choice

Sciences Po Session 1: Review August 27, 2018 5 / 30

Consumer Choice

A consumer has a xed income and can choose between two goods: coee
and tea
R = consumer's income
C = Number of coees
T = Number of teas
The consumer can get as many coees and teas as his or her income
allows, that is if pC (resp. pT ) is the price of a cup of coee (resp. a cup
of tea) then we must have:
pc C + pT T ≤ R (1)

Sciences Po Session 1: Review August 27, 2018 6 / 30

Graphical representation
12

10 R
pT = 10
1 R = 10, pT = 1,pC = 2

6 Feasible allocations
T

0
0 2 4 R
pC = 10
2
6 8
C

Sciences Po Session 1: Review August 27, 2018 7 / 30

Relative price
Exercise : How many teas is the consumer ready to give up to obtain one
more cup of coee? What is the relative price of tea to coee? Look at the
graph below for indications.
12

10 R
pT = 10
1 R = 10, pT = 1,pC = 2

6 Feasible allocations
T

-2T
4 +1C
2

0
0 2 4 R
pC = 10
2
6 8
C
Sciences Po Session 1: Review August 27, 2018 8 / 30
Preferences

A bundle Q = (C , T ) is a combination of coee and tea consumed.

For instance (2,6) (two coees and 6 teas) and (3,4) (3 coees and 4 teas)
are both feasible bundles. Which will the consumer prefer?
Consumers' preferences are pairwise rankings of all available
consumption bundles.This ranking must satisfy the following properties:
Completeness
Transitivity
Non satiety
Exercise : what are the denitions of each of these three properties?

Sciences Po Session 1: Review August 27, 2018 9 / 30

Indierence curves
An indierence curve goes through all the bundles between which
the consumer is indierent.
12

10

6
T

0
0 2 4 6 8
C
Sciences Po Session 1: Review August 27, 2018 10 / 30
Indierence curves

Indierence curves:
are downward sloping
do not cross each other
are convex
the further from the origin, the higher the satisfaction they yield
Exercise : Explain how each of these properties is consistent with
preferences' denition.
The slope of an indierence curve is called the Marginal Rate of
Substitution (MRS). In our example, it reects how many teas (resp.
coees) the consumer is willing to give up for more coees (resp. teas).

Sciences Po Session 1: Review August 27, 2018 11 / 30

Optimal Choice
Given the consumer's budget constraint and indierence curves, which
bundle will he or she choose?
12

10

6
T

0
0 2 4 6 8
C
Sciences Po Session 1: Review August 27, 2018 12 / 30
Remember

Because indierence curves are convex, the optimal bundle exhausts

the budget constraint.
Hence the optimum is found at the tangency point between the
budget constraint and the indierence curve yielding the highest
poissble level of statisfaction.
At optimum: Marginal Rate of Substitution = Relative Price

Sciences Po Session 1: Review August 27, 2018 13 / 30

 Firm Production

Sciences Po Session 1: Review August 27, 2018 14 / 30

Firm production

When producing Q unit of good, rms face both a xed cost F and a
variable cost CV (Q). Hence their cost function is C (Q) = F + CV (Q).
The average cost is CA (Q) = C (Q)
Q .
Marginal cost CM (Q) is dened as the change in cost over the change in
quantity.
Q2
Example : Suppose a rm's cost function is C (Q) = |{z}
3 + .
10
F |{z}
CV (Q)
First the rm produces Q = 1 unit of good. Its cost is C (1) = 1031 .
If it wants to produce Q = 2 goods, then its cost will be C (2) = 1034 .

The marginal cost of producing 1 good to 2 goods is C (22)−C

−1
(1)
= 10 3

Sciences Po Session 1: Review August 27, 2018 15 / 30

Isoprot curves

Next, rms set a price P for their good. They make prot if P ≥ CM (Q).
Their prot function is
Π(Q) = PQ − C (Q)
Π(Q) + C (Q)
⇔P=
Q
Π(Q)
⇔P= + CA (Q)
Q
This last equation denes isoprot curves.

Sciences Po Session 1: Review August 27, 2018 16 / 30

Graphical representation
From our example above:
60
Price/marginal cost

40 Marginal cost

Isoprot curve: 9
20 Isoprot curve: 6
Isoprot curve: 3

0
0 5 10 15 20
Quantity of goods
Sciences Po Session 1: Review August 27, 2018 17 / 30
Prot maximization
Price/marginal cost 60

40

Feasible choice set

Isoprot curve: 9
20 Isoprot curve: 6
P* Isoprot curve: 3

Demand curve
0
0 Q* 5 10 15 20
Quantity of goods

Sciences Po Session 1: Review August 27, 2018 18 / 30

Remember

Isoprot curves describe price-quantity combinations that yield the

same prot.
Isoprot curves intersect with the marginal cost curve at their
minimum.
The rm's goal is to maximize its prot Π(Q). It can only do so
within consumers' feasible choice set.
Therefore a rm maximizes its prot by choosing the isoprot
curve that features the highest price and is tangent to
consumers' demand curve.

Sciences Po Session 1: Review August 27, 2018 19 / 30

Deadweight loss
Price/marginal cost 60

40 Marginal cost

Consumer surplus

20 Isoprot curve: 6
P*
Deadweight loss
Firm surplus Demand curve
0
0 Q* 5 10 15 20
Quantity of goods

Sciences Po Session 1: Review August 27, 2018 20 / 30

 Game Theory

Sciences Po Session 1: Review August 27, 2018 21 / 30

Game Theory

Concepts we are going to review:

Dominant strategies
Nash equilibrium
Repeated games
Mixed strategies

Sciences Po Session 1: Review August 27, 2018 22 / 30

Some notations

There are P players. Player i has a set of n actions qi = (qi1 , . . . , qin ).

Each player i reacts to other players' action through reaction function
Ri : (q1 , . . . , qi−1 , qi+1 , . . . , qP ) → qi . A best response is a reaction
function Ri that yields the best possible outcome for player i .
In the examples below, there will be only two players A and B and two
possible actions:
P=2
n=2
Player A has set of actions qA = (qA1 , qA2 ) and reaction function
RA : qB → qA (same for B)

Sciences Po Session 1: Review August 27, 2018 23 / 30

Dominant strategies in prisoner's dilemna

Two possible actions for each player A and B: qA = qB = (C,D). That is

players can either cooperate (C) or defect (D).

Robber 1
C D
Robber 2 C (−2, −2) (−6, 0)
D (0, −6) (−5, −5)

Exercise : What is each player's dominant strategy?

Sciences Po Session 1: Review August 27, 2018 24 / 30

Nash equilibrium in battle of sexes
Consider another game, in which a couple, Charlie and Alex, must choose
between two activities: going to the bar (B) or watching a movie at home
(M).

Charlie
B M
Alex B (2, 1) (0, 0)
M (0, 0) (1, 2)

Exercise
Show there are no dominant strategies in this game. How do we nd Nash
equilibria?

Sciences Po Session 1: Review August 27, 2018 25 / 30

Remember

Two actions constitute a Nash equilibrium if each action is a best

response to the other. Said dierently, no player has any incentive to
deviate and choose a dierent action.
In particular an equilibrium stemming from dominant strategies is a
Nash equilibrium.
John Nash proved in 1950 that there always exists at least one
Nash equilibrium in every nite game.

Sciences Po Session 1: Review August 27, 2018 26 / 30

Repeated games: back to the prisoner's dilemna
Imagine the two robbers are caught over and over again. They must then
decide a strategy that dictates their behavior each time they are caught. In
such a set up, the equilibrium is such that no one deviates from their
assigned strategy at any time.
Example :
Both players always defecting (i.e choosing D each time they are
caught) is an equilibrium: if player A and B always play D, none of
them wish to deviate and play C at any time, because it would result
in a smaller payo.
Both players always cooperating (i.e choosing C each time they are
caught) is not and equilibrium: if player A (resp. B) always plays C,
player B (resp. A) will want to deviate from C to D to get a larger
payo.

Sciences Po Session 1: Review August 27, 2018 27 / 30

Exercise : Is permanent retaliation (i.e. each player cooperates until one
defect. if player A (resp. B) defects, player B (resp. A) defects forever) an
equilibrium in this game?

Sciences Po Session 1: Review August 27, 2018 28 / 30

Mixed strategies: tax game

Until now, we've only discussed pure strategies, i.e strategies in which
players always choose the same action. However in the game below there is
no Nash equilibrium in pure strategies.

Tax authorities
Audit No audit
Tax payer Declare all income (3, 1) (3, 2)
Lie on income (0, 4) (5, 0)

But Nash said there always exists an equilibrium! Hence the concept of
mixed strategies: players play their actions with some probablity, so as to
make the other indierent to their actions.

Sciences Po Session 1: Review August 27, 2018 29 / 30

Let's nd the mixed strategies Nash equilibrium in the tax game

Tax authorities
Audit (A) No audit (N)
Tax payer Declare all income (T) (3, 1) (3, 2)
Lie on income (L) (0, 4) (5, 0)
Tax authorities are indierent to auditing or not if they get the same
payo for both actions. Call x the probablity that the tax payer tells
the truth. Then we are looking for x such that:
1 × x + 4 × (1 − x) = 2 × x + 0 × (1 − x)
| {z } | {z }
Tax authorities' expected Tax authorities' expected
payo if auditing payo if not auditing
−3 × x + 4 = 2 × x
4
x=
5
4 1
Therefore ( 5 T , 5 L) is the tax payer's equilibrium strategy.
Exercise : nd the equilibrium mixed strategy for tax authorities.
Sciences Po Session 1: Review August 27, 2018 30 / 30