CH 11 Risk

Lecture notes
 Degenerate lottery assigns probability
1
Questions 11.2, 11.3 (except for part (a) — notice
by the way that in part (b) the expected iso-value
lines are just the parallel lines that have the same
expected value), 11.4, 11.5 (the certainty
equivalent of a lottery is defined in page 195 of
the textbook), and 11.6.  Horizontal shows the probability p1
 associated with the worst prize x1, while
 How much are you willing to pay to the vertical axis shows the probability p3
insure against 100 000 loss? associated with the best prize x3
 Objective probabilities: probabilities  The probability of winning prize x2 is 1-
for which there is some statistical, 0.2-0.3=0.5 and can be inferred from the
experimental or analytical basis that horizontal length from c to the.
different people can agree upon Hypotenuse as shown by the dashed
 Expected utility hypothesis magenta arrow.

11.1 Expected Utility 11.1.2 Preferences

 Decision making under risk, the
commodity space has to be redefined
Reflexivity: for any lottery p in delta

Totality: for any two lotteries p and q in delta,

either or both

Transitivity: for any three lotteries p,q and r in

delta ir
Independence: for any three lotteries p,q and r in

11.1.1 Commodity space delta if then for any t in the range

 Consumer can win one of three prizes

Continuity: lotteries p,q and r in delta

 A lottery assigned
probability pj of winning prize xj then there is some t in the
 The probabilities have to sum to one range
 The consumer’s decision problem under
risk is to choose between different lotteries
 Commodity space is the set of all possible  Idea of continuity: if p>~q>~r so that p is
lotteries which we denote by delta the best of the three and the r the worst,
 The Masrchak triangle represents the then there is some average of the best and
commodity space delta in two dimensions worst lotteries that is indifferent to the one
in the middle
 Idea of independence: if p is weakly
better than q to begin with , then taking
any weighted average (or mixture) of p
with r should remain weakly better than
 the corresponding weighted average of q
with r
 Suppose: p=(0.2,0.5,0.3) is at least as good
as q=(0.5,0.3,0.2) and r= (0.4,0.4,0.3)
 Let t=1/2
 Then independence requires that the
mixture

Be at least as good as the mixture

The expected utility theorem shows that if a

consumer preference over lotteries satisfy these
axioms, then there is
a) Von neumann-morgenstern (vNM
utility function u(x) over the set of prizes  Vertical intercept:
X
b) That the consumer preferences over
lotteries can be represented by an  Positive slope:
expected utility (EU) function V, where
her utility from lottery p is given by
 Slope of any indifference curve is a
constant that depends on the vNM utilities,
 Theorem holds for any positive monotonic not on the probabilities of the prizes
transformation v of the vNM utility u so  Intercept is also independent of the
long as v=au+b where a>0 probabilities and increases with the value
 The preferences over lotteries are of V^.
unchanged if we replace each of the utility  Therefore for larger values of V^, the
from the prizes in v(p) by the consumer’s EU indifference curves are
transformation increasing to the northwest

11.2 Attitudes towards Risk

 vNM utility cannot be subject to any  We assume that any consumer is a VNM
positive monotonic transformation , it is
expected utility maximiser
not an ordinal utility function
 Set X:
 because u(xi) terms are multiplicative
constants, the EU function is linear in the o X1 = $6400
probabilities o X2= $9100
 this implies consumer’s preferences over o X3= $10 000
lotteries generates linear indifference  Consumer begins with wealth 10 000
curves by EU function V(p)  25% chance reduction to 6400
 P = (0.25,0,0.75)
  Expected value
 First replace p2 in V(p) with
 Second lottery q=(0,1,0) which yields
To obtain $9100 for sure – the same as the expected
under lottery p.
 When two lotteries yield same value of
To draw an indifference curve, fix the utility level wealth, we say they are actuarially fair.
V(p) at V^  Risk Adverse: if given a choice between
 Rearrange 11.2 and make p3 the subject lottery p or receiving the expected value
of p for sure, she prefers the latter. (a sure
 thing over a actuarially fair lottery) i.e.  horizontal – how much x in good state H
V(q) > V(p)  vertical: amount in state L
 Concave vNM utility function  diagonal line Ca is the line of certainty

 along which , the consumer

 Where u bar is the weighted average of receives the same amount in either state
the utilities u(x1) and u(x3)  consumer a’s endowment in the two

states is given by
which lies below the diagonal line Ca ,
meaning she is better off in the high state
as opposed to the low state
 assume a is risk averse and has vNM

 then the expected utility from the bundle

(xH, xL) is
 Concave – risk loving
 Risk neutral consumer is one whose vNM
utility function is linear – indifferent  expected utility from endowment wA is
between risky lotter and the expected
value of that lottery for certain
 if each state of the world is equally likely ,
11.4 Pareto Efficient Risk Sharing
then p=1/2 and a’s utility from wA is 20.
 Simplest possible situation to share risk:
 All consumption bundles that yeildutility
state contingent claims
20 is given by the EU indifferent curve
 Single good c: which shows a’s preferences over the
 Two states of the world: bundle is convex
o high H which occurs with  MRS between two states
probability p
o low state L which pccurs with
probability 1-p
 state contingent claims bundle
 MRS along the certainty line is always
is the lottery over prizes xH
and xL with probabilities p and 1-p equal to no matter what
the utility function
 A(400,400) is the certainty equivalent of
the lottery wA. It gives the same utility as
wA except that is it risk free because it lies
on the certainty line.
 Say consumer b with endowment wb=
(1324,604), the aggregate endowment
becomes wA+wB = (2000, 800)
 Look at the possibility of risk sharing
between consumers a and b
11.4.1 Risk averse and risk neutral consumer
  Pay offs in each cell add up to zero
meaning that it’s a zero sum game where
one player’s gain is always
 Intial endowment w lies below each counterbalances by the loss of some other
consumers line of certainity, both are player/s
better off in the high stte than in low state
 Consumer A is risk averse 12.2 solving static games

 Consumer b is risk neutral 12.2.1 Dominant strategy equilibrium

Dominant strategy: one strategy that is the best
in terms of the player’s payoffs, regardless of
 Then b;s expected utility is what others are doing

Which is linear with MRSb = p/(1-p) shown by the

green EU indifference curve
 Interior contact curve is found by setting
MRSA given

equated to MRSb = p/(1-p)  Aruna charges 10, then mahala will charge
 Solving we get xLa = xHa i.e. the contract 10 as well  payoff of 50
curve coincides with the line of certainty  Is aruna charges 15, then mahala will
Ca. charge 10 - payoff of 80
 Therefore at any pareto efficient  Regardless of what aruna charges,
allocation, consumer a is fully insured charging 10 is always the best for mahala
aginst he state contingent risk i.e. dominant strategy
 Continue….  Dominated strategy: the payoff is always
11.4.1 Two risk averse consumers worse off regardless of what is chosen
Continue..  Dominant strategy equilibrium (DSE):
when players play their dominant
strategies. Each person has a strategy that
is the besno matter hwat others are
CH 12 Game Theory doing; no person has an incentive to
switch to a different strategy
 Strategic interaction: outcomes depend
on all agents  (10,10) was the dominant strategy
 Normal form game: three basic elements equilibrium
– players, strategies and payoffs  DSE cannot be applied when neither
e.g. player has a dominant strategy i.e. if one
 player does not have a dominant strategy,
that game cannot have a DSE.
 Prisoner’s dilemma:


 Possible NE

 Talk is a dominant strategy for each
prisoner (talk, talk) is a DSE
 The prisoner’s dilemma because when the
prisoners follow their own self interst and
play their dominant strategies they end up
in prison for 10 years each as the DSE,
when instead they bouth would have 
been better off being in prison for one  (call, call) cannot be a NE
year if the two could find a way to remain  Ii is a NE
silent  Iii is also a NE
 Being quiet is never in one’s self interest  Iv is not a NE
 Conditions: Method of mutual best response:
o Each player must have a dominant
strategy
o The DSE outcome must yield
payoffs that are worse for each
player than some other outcome
that is potentially possible in the
pay off matrix

12.2.2 Nash equilibrium

* At NE, unilateral deviation does not pay – no  Two players with three strategies each
player gains by changing her strategy on her own  Up, middle and down
 Unilateral deviation which eliminates the  Left, centre and right
possible strategy combination where at  Find alito’s best response to breyer’s
least one player wants to deviate strategies
 Any strategy combination that does no  Suppose breyer plays left. The highest
get eliminated is then a NE payoff for A is to play Middle
 Second is the method of mutual best  NE is a mutual best response i.e. the
response which determines each players choice of the row players strategy is the
best response to different strategies best given the column players choice and
played by other players viceversa
 NE arises when each player’s strategy is a  U,C and M,L are both Nash equilibria
mutual best response: every player
Relationship between NE and DSE
maximises her payoff given what others
 Nash equilibria and Dominant strategy
are doing.
equilibria
Method of unilateral deviation
 A NE is a configuration of strategies where
no single player wants to change their
strategy given what the other players are
doing
  At DSE, no player wants to deviate  Mirror image BR functions – this is a
regardless of what the others are doing ; symmetric game
in particular, no player will want to  Unique NE is where the BR functions
deviate when everyone else plays their intersect E= (0.5,0.5)
dominant strategies  We write the mixed strategy NE as the
 Every DSE must be a NE list of probabilities associated with each
 DSE is a special case of NE strategy for each player
 A game can have NE but no DSE
Solving the battles of the sexes game
12.2.3 Mixed strategies
 No NE which implies no DSE either
 No NE in pure strategies – meaning each
player chooses a strategy for sure
 Play the game many times
Solving the matching pennies game
 P probability that ann chooses heads and
(1—p) probability that ann chooses tails
 Similarly for bob but q
 Verify the two Nash equilibria in pure
strategies is (F,F) and (O,O)
 Expected payoffs

 NE in mixed strategies is one where no

one wants to change their won mixed
Equate the expected pay offs for the husband to
strategy given the mixed strategy of the
get q=0.25 since his expected payoff is the same
other players regardless of the value of p he chooses
 Ann’s choice of p has to be the best - Represented by the horizontal line BRH
response to bob’s choice of q - Set q=0 and 1
 Best response functions for each player - When q=0, he is better off going opera i.e.
 How to plot the best response for each p=0
player - For q=1, pay off is maximised when he
 For anne: sub in random values of q into goes football i.e. p=1
her two equations to reval which one she - Do the same for wife
is better off playing
 E.g when p =1, she is better off playing
p=1
Generally:

 dashed line means the nodes assigned

means bob cannot distinguish between
those nodes i.e. he cannot surmise
Cross three times whether anna has played H or T
NE is ((0,1),(0,1))  the collection of nodes joined together
EM NE ((0.25,0.75),(0.5,0.5)) are known as a information set
NE EF ((1,0),(1,0))  if a players information set consists of one
All best response functions in two player games node it is called a singleton information
with two strategies each are eicewise linear set
- one or three nash equilibria  if all information sets ina game are
- the idea of mixed strategy can be extended to singleton sets, we say that it is a game of
more than two players and more than two perfect information
strategies but it is no longer possible to draw best 12.3.2 Subgame perfect Nash equilibrium
response functions and derive them graphically  assume common knowledge of rationality
- all players know that at any decision
12.3 Dynamic Games node, if the player at that node has a
 Players move sequentially choice that leads to a strictly higher pay
 Two features: off compared to any other strategy
o Timing of play  subgame: subset of the extensive form
o What information the players have game that containsa singleton initial node
when they move and all the nodes and all the node sbelow
12.3.1 Extensive form games that can be reached from it.
I: incumbent firm Solving the potential entry game
E: potential entrant who is threatening to turn
the market from a monopoly to a duopoly
Initial node: open
D: don’t enter
E: enter
Terminal node (0,4)
First number denotes the pay off of entrant and
second the incumbent

 it is common knowledge that the game

would proceed along the red branch
 from common knowledge of rationality, it
follows that E chooses e,
 subgame perfect nash equilibrium (SPNE)
os reached when player E plays e and
player I plays a, the equilibrium path of
the game is shown by the red branches
Solving the centipede game

A: accommodate entry
F: fight
 Simultaneous games can be written in
extensive form
 Profit maximised
‘

OR MR=MC

 The price charged for the output Q*, is

 two piles of cash on a table one
from from the inverse demand so
containing $4 and the other containing $1
 t: terminate
 p: push the pile to the other player in replacing
which the cash pile doubles  Sub this into the eqn above
 ends after 4 rounds
 four subgames
 the only SPNE is that player 1 chooses to Factor out p*
terminate the game at the initial node

Price elasticity of demand on the inverse

demand

Sub into eqn above

CH 10 Monopoly
 monopoly can engage in uniform pricing,
charging the same price to all its
customers or it can engage in differential  Therefore LHS which is = MR is >0 when
pricing (price discrimation) and charge |elasticity of demand| >1
different prices from different customers  Therefore a monopoly can earn a positive
MR only if its demand is price elastic.
10.1 Uniform Pricing
 can set both price and quantity
 find the quantity price combination that
lies on the market demand curve so as to
 numerator is the absolute mark up
maximise its profit
(difference between what the monopolist
 setting price determines how much the
charges and what is costs to produce the
monopoly can sell and therefore how
last unit
much it should produce
 RHS is always >0 since e is always
 setting the quantity determines the max
negative
price that the monopoly can sell the
 P* >0 so the absolute mark up is positive
quantity for
– monopolist charges a price that is
greater than the cost of producing the last
10.1.1 Profit maximisation unit
 let p=D(Q) be inverse market demand  Dividing the absolute mark up by p*, we
curve obtain the relative mark up – a unit free
 then profit function is measure of market power known as the
lerner index

  The more price elastic the demand, the Total revenue is
lower the relative mark up: goods are
relatively elastic the demand the lower the
relative mark up
10.1.2 Calculating monopoly output and price So MR is

Two types of demand curves

10.2 Differential Pricing, 10.3 Personalized
 Linear inverse demand
Pricing (except subsection 10.3.2), and 10.4
 Constant price elasticity through out
Group Pricing.
Linear demand:
 P=abQ
 TR=

 So
 The MR for a linear inverse demand has
the same vertical intercept as the inverse
demand but is twice as steep
Example:
P=120-Q
C(Q)=Q^2
MR=120-2Q
MR=MC
MC=2Q
Q*=30
Sub Q* into inverse demand function to find
p*=$90.
 Price is more than MC of producing the
last unit of output $60 so the absolute
mark up is $30.

Constant -elasticity demand

Q=Ap^epsilon
Where episilon is a negative number

Inverse demand is
CH 13 Oligopoly
 Oligopolistic equilibrium, the behaviour of
firms corresponds to that of a Nash
Equilibrium (NE) i.e. each firm is
maximising its profit given the actions of
others.
 Two ways to model:
o Quantity competition
o Price competition

13.1 Static Quantity Competition

13.1.1 Cournot duopoly
 Two firms produce homogeneous good
 Total quantity Q=q1+q2 -
BR show show much q1 it should produce
 Inverse market demand p=200-Q in response to different levels of firm 2’s
 Cost each firm $20 to produce each unit output q2
of output - Note that firm 1’s BR line connects the
 Each firm chooses its own output taking tops of the isoprofits and firm one’s profits
as given the other firm’s production level increase to the sound as shown by the
Graphical representation to find NE arrows
 Plot each firms best response to the other Maximise firm 2’s profit
firms output choice and find a point of of
mutual best response
BR

Maximise

Plot both BR
The value of q1 maximises firm 1’s profit for the
given q2
Express q1 as a function of q2

 NE is mutual best response of each firm

to the others choice of output level
 Point E intersection is the NE (60,60)
 Such quantity into inverse market demand
cruve to find optimal price of 80.
Breaking it down
 Suppose firm 2 decides to produce q2=60
  Market price even before firm 1 produces
anything is at 140 as a result of firm 2’s
output
 Any production of firm 1 is going to lead
to a total output level in excess of 60 units
and reduce the market price further
e.g. p=200-q1-q2
MC1= $30
MC2 = $10

Solve simultaneously to get q1*=50 and q2*= 70

 Demand curve for firm one starts from

(60,140) to (200,0)  firm one’s 13.1.2 Cournot ogliopoly
resideual inverse demand  Extend with n identifical firms
 Set MR to MC of 20  P=a-Q
 When market inverse demand is linear  Q=q1+q2+q3+q4…
and firms have identical, constant MC,  MC= c
then each duopolist output is one third the  Then profit function is
output where the MC crosses the demand
point A
Algebraic derivation in the case of a symmetric Maximise
duopoly (when MC is the same of the two firms)

Since all firms are identical, each will choose the

same level of output at the NE, so set

 Imposing symmetry, we get

Where
Impose symmetry Then total quantity produced is

and
So q* = 60
Assymetric duopoly (different MC)

13.2 Static Price Competition

 Two models
o When products are identical
 o When products are differentiated
BR function
13.2.1 Bertrand duopoly
 Sell at different prices
 Products are identical
 Inverse market demand p=200-Q
 Prices are easily comparable This is how firm 2 will choose its output in stage 2
 Assume each firm MC= 10 per DVD Firm one will incorporate this into its profit
 NE prices? (p1*,p2*)? function
 Successive rounds of undercutting imply
that only (p1*,p2*) = (10,10) each ifrm
chooses to set its price equal to the MC in
equilibrium
 Bertrand paradox : how is it that MC
pricing requires the presence of many
firms under perfect competition while
price competition requires two firms?
 Two underlying assumptions:
 Each firms sells an identical homogenous
products
 each firm can handle the entire market
demand when it undercuts its rival
 if a firm sells differentiated products, -
then because of branding and consumer
loyalty one firm could charge above its
MC and not be fareful that its entire
clientele could be captured by the other
firm undercutting its price
 price undercuts by less known firms are
not likely to woo an amazon customer
 when a firm finds it impossible to meet
market demand when it undercuts a rival’s Step 4. Maximise to optain q1*= 90
price, we say it faces capacity constraints Then q2*= 45 so SPNE is (90,45) at point S
 then CM is no longer a NE  Since firm 2 is a follower, it will choose
on the BR2
13.3 Dynamic competition  Firm 1 will choose an output level so as to
maximise its profit i.e. choose highest
 firm one chooses q1 then firm 2 chooses isoprofit line that it can reach given BR2
q2
 Stackelberg price is p*=65
 solved backwards for the subgame perfect
 Pi= 4050 and
nash equilibrium (SPNE)
 Pi2 = 2025
13.3.1 Stackelberg Cournot duopoly

 Inverse market demand p=200-q1-q2 and

each firm has MC =20
 Assume firm 1 has already produced an
output q1 in the first stage
 Step 1. Maximise firm 2’s profit in the
second sate