Advanced Microeconomic

Advanced Microeconomics
Harald Wiese
University of Leipzig
Winter term 2012/2013
Harald Wiese (University of Leipzig) Advanced Microeconomics Winter term 2012/2013 1 / 70
What I like about economic theory and microeconomics
Not just talk, but theory with formal models (> sociology)
The core of neoclassical economics: Walras model of perfect
competition (chapter XIX)
interdependence of markets and
prices to bring markets into equilibrium
and also and beyond:
decisions under uncertainty (chapters III and IV)
detailed analysis of consumer behavior (chapters VI and VII)
interactive decisions = game theory (chapters X to XIII)
simple concepts like Pareto eciency with wide ramications (chapter
XIV)
auction theory (chapters XVII and XVIII)
adverse selection (chapter XXII) and hidden action (XXIII)
However: hard work ahead
Technical stu
3 class meetings
Monday 09:15 - 10:45 in SR 13 (I 274)
Wednesday 09:15 - 10:45 in SR 17 (I 374)
Friday 09:15 - 10:45 in SR 13 (I 274)
mixing lectures and exercises
written exam 120 min. at two dates
midterm on 07.12.2012 (60 min.)
nal in February (60 min.)
Literature
Harald Wiese: Advanced microeconomics (which is more detailed
than the lecture)
Andreu Mas-Colell, Michael D. Whinston, Jerry R. Green:
Microeconomic Theory
Hugh Gravell, Ray Rees: Microeconomics
Georey Jehle, Philip Reny: Advanced Microeconomic Theory
Samuel Bowles: Microeconomics Behavior, Institutions, and
Evolution
Thomas Hnscheid: Schwester Helga
Course overview
Part A. Basic decision and preference theory:
Decisions in strategic form
Decisions in extensive form
Ordinal preference theory
Decisions under risk
Chapter II
Decisions
in strategic form
Decision theory Game theory
Chapter III
Decisions
in extensive form
Chapter X
Games
in strategic form
Chapter XII
Games
in extensive form
Course overview
Part B. Household theory and theory of the rm:
The household optimum
Comparative statics and duality theory
Production theory
Cost minimization and prot maximization
Course overview
Part C. Games and industrial organization:
Games in strategic form
Price and quantity competition
Games in extensive form
Repeated games
Course overview
Part D. Bargaining theory and Pareto optimality:
Pareto optimality in microeconomics
Cooperative game theory
Course overview
Part E. Bayesian games and mechanism design:
Static Bayesian games
The revelation principle and mechanism design
Course overview
Part F. Perfect competition and competition policy:
General equilibrium theory I: the main results
General equilibrium theory II: criticism and applications
Introduction to competition policy and regulation
Course overview
Part G. Contracts and principal-agent theories:
Adverse selection
Hidden action
Part A. Basic decision and preference theory
1
Decisions in strategic (static) form
2
Decisions in extensive (dynamic) form
3
Ordinal preference theory
4
Decisions under risk
Overview
1
Introduction
2
Sets, functions, and real numbers
3
Dominance and best responses
4
Mixed strategies and beliefs
5
Rationalizability
Introduction
Decision in strategic form:
helps to ease into game theory;
concerns one-time (once-and-for-all) decisions where payos depend
on:
strategy;
state of the world.
Introduction
Example
state of the world
bad weather good weather
strategy
production
of umbrellas
100 81
production
of sunshades
64 121
Introduction
Example 1: production of umbrellas or sunshades
Denition
A decision situation in strategic form is a triple
= (S, W, u) ,
with
S (decision makers strategy set),
W (set of states of the world), and
u : S W R (payo function).
= (S, u : S R) is called a decision situation in strategic form without
uncertainty.
Introduction
In this example:
S = umbrella, sunshade;
W = bad weather, good weather;
payo function u given by
u (umbrella, bad weather) = 100,
u (umbrella, good weather) = 81,
u (sunshade, bad weather) = 64,
u (sunshade, good weather) = 121.
Note that:
decision maker can choose only one strategy from S;
only one state of the world from W can actually happen.
Introduction
Denition
Decision situation without uncertainty:
u (s, w
1
) = u (s, w
2
)
for all s S and all w
1
, w
2
W.
Trivial case: [W[ = 1.
Introduction
Example 2: Newcombs problem
Assumptions:
2 boxes:
1000 e in box 1;
0 e or 1 mio. e in box 2.
strategies: open box 2, only, or both boxes;
Higher Being put the money into the boxes depending on His
prediction of your choice;
Higher Beings predictions can be correct (She knows the books you
read) or wrong.
Problem
What would you do?
Introduction
Example 2: Newcombs problem, version 1
prediction:
box 2, only
prediction:
both boxes
you open
box 2, only
1 000 000 Euros 0 Euro
you open
both boxes
1 001 000 Euros 1 000 Euro
Introduction
Example 2: Newcombs problem, version 2
prediction
is correct
prediction
is wrong
you open
box 2, only
you open
both boxes
1 000 Euro 1 001 000 Euros
We come back to Newcombs problem later ...
Introduction
Example 3: Cournot monopoly
Denition
= (S, ), where:
S = [0, ) (set of output decisions),
: S R (payo function) dened by (s) = p (s) s C (s) with
p : S [0, ) (inverse demand function);
C : S [0, ) (cost function).
1
Introduction
2
3
4
5
Rationalizability
Sets
Set, element, and subset
Denition (set and elements)
Set any collection of elements that can be distinguished from each
other. Set can be empty: ?.
Denition (subset)
Let M be a nonempty set. A set N is called a subset of M (N _ M) if
every element from N is contained in M. are used to indicate sets.
M
1
= M
2
i M
1
_ M
2
and M
2
_ M
1
.
Problem
True?
1, 2 = 2, 1 = 1, 2, 2
1, 2, 3 _ 1, 2
Sets
Tuples
Denition
Let M be a nonempty set. A tuple on M is an ordered list of elements
from M. Elements can appear several times. A tuple consisting of n
entries is called n-tuple. () are used to denote tuples.
(a
1
, ..., a
n
) = (b
1
, ..., b
m
) if n = m and a
i
= b
i
for all i = 1, ..., n.
Problem
True?
(1, 2, 3) = (2, 1, 3)
(1, 2, 2) = (1, 2)
Sets
Cartesian product
Denition
Let M
1
and M
2
be nonempty sets. The Cartesian product of M
1
and M
2
(M
1
M
2
) is dened by
M
1
M
2
:= (m
1
, m
2
) : m
1
M
1
, m
2
M
2
.
Example
In a decision situation in strategic form, S W is the set of tuples (s, w).
Problem
Let M := 1, 2, 3 and N := 2, 3. Find M N and depict this set in a
two-dimensional gure where M is associated with the abscissa (x-axis)
and N with the ordinate (y-axis).
Functions
Injective and surjective functions
Denition
Let M and N be nonempty sets.
A function f : M N associates with every m M an element from
N, denoted by f (m) and called the value of f at m.
Sets
M domain (of f )
N range (of f )
f (M) := f (m) : m M =
_
mM
f (m) image (of f ).
Injective function if f (m) = f (m
/
) implies m = m
/
for all
m, m
/
M.
Surjective function if f (M) = N holds.
Bijective function both injective and surjective.
Functions
value versus image
m
x
( ) m f
M
( ) M f
f (M) := f (m) : m M =
_
mM
f (m)
Functions
Exercise
Problem
Let M := 1, 2, 3 and N := a, b, c.
Dene f : M N by
f (1) = a,
f (2) = a and
f (3) = c.
Is f surjective or injective?
Functions
Two dierent sorts of arrows
1
f : M N (domain is left; range is right).
2
m f (m) on the level of individual elements of M and N.
Example
A quadratic function may be written as
f : R R,
x x
2
.
or in a shorter form: f : x x
2
or very short (and in an incorrect manner): f (x) = x
2
Functions
Inverse function
Denition
Let f : M N be an injective function. The function f
1
: f (M) M
dened by
f
1
(n) = m = f (m) = n
is called f s inverse function.
Problem
Let M := 1, 2 and N := 6, 7, 8, 9. Dene f : M N by
f (m) = 10 2m. Dene f s inverse function.
Cardinality
Denition
Let M and N be nonempty sets and let f : M N be a bijective function.
== M and N have the same cardinality (denoted by [M[ = [N[).
If a bijective function f : M 1, 2, ..., n exists, M is nite and contains
n elements. Otherwise M is innite.
Problem
Let M := 1, 2, 3 and N := a, b, c. Show [M[ = [N[.
Real numbers
Sets of real numbers
Real numbers:
sets symbol elements
natural numbers N 1, 2, 3, ...
integers Z ..., 2, 1, 0, 1, 2, ...
rational numbers Q
_
p
q
: p Z, q N
_
irrational real numbers RQ
_
2 = 1.414 2..., e = 2.7183..., etc.
Note N Z Q.
Problem
1
8
and
4
7
are rational numbers. Write these numbers as 0.1... and 0.5... and
show that a repeating pattern emerges.
Real numbers
Theorem
Theorem (cardinality)
The sets N, Z, and Q are innite and we have
[N[ = [Z[ = [Q[ .
However:
[Q[ < [R[
and even
[Q[ < [x R : a _ x _ b[
for any numbers a, b with a < b.
Real numbers
Intervals
[a, b] : = x R : a _ x _ b ,
[a, b) : = x R : a _ x < b ,
(a, b] : = x R : a < x _ b ,
(a, b) : = x R : a < x < b ,
[a, ) : = x R : a _ x and
(, b] : = x R : x _ b .
Problem
Given the above denition for intervals, can you nd an alternative
expression for R?
Convex combination
Denition
Let x and y be elements of R. =
kx + (1 k) y, k [0, 1]
the convex combination (the linear combination) of x and y.
Problem
0 x + (1 0) y =?
Problem
Where is
1
4
1 +
3
4
2, closer to 1 or closer to 2?
[a, b] = x R : a _ x _ b
= ka + (1 k) b : k [0, 1]
Decisions in strategic form:
1
Introduction
2
3
4
5
Rationalizability
Weak and strict dominance
Denition
= (S, W, u) decision situation in strategic form.
Strategy s S (weakly) dominates strategy s
/
S i
u (s, w) _ u (s
/
, w) holds for all w W and
u (s, w) > u (s
/
, w) is true for at least one w W.
Strategy s S strictly dominates strategy s
/
S i
u (s, w) > u (s
/
, w) holds for all w W.
Dominant strategy a strategy that dominates every other strategy
(weakly or strictly).
s
/
is called (weakly) dominated or strictly dominated.
Newcombs problem, version 1
prediction:
box 2, only
prediction:
both boxes
you open
box 2, only
you open
both boxes
1 001 000 Euros R 1 000 Euro R
Which strategy is best
for the rst state of the world,
for the second state of the world?
Newcombs problem, version 2
prediction
is correct
prediction
is wrong
you open
box 2, only
1 000 000 Euros R 0 Euro
you open
both boxes
1 000 Euro 1 001 000 Euros R
Which strategy is best
for the rst state of the world,
for the second state of the world?
Power set
Denition
Let M be any set. The set of all subsets of M is called the power set of M
and is denoted by 2
M
.
Examples
M := 1, 2, 3 has the power set
2
M
= , 1 , 2 , 3 , 1, 2 , 1, 3 , 2, 3 , 1, 2, 3 .
Note that the empty set also belongs to the power set of M !
Problem
Can you derive a general rule to calculate the number of elements of any
power set?
arg max
Example
Consider a rm that tries to maximize its prot by choosing the output
x optimally. The output x is taken from a set X (e.g. the interval [0, ))
and the prot is a real (Euro) number.
(x) R : prot resulting from the output x,
max
x
(x) R : maximal prot by choosing x optimally,
argmax
x
(x) _ X : set of outputs that lead to the maximal prot
Of course,
max
x
(x) = (x
+
) for all x
+
fromargmax
x
(x) .
arg max
The following cases are possible:
argmax
x
(x) contains several elements.
argmax
x
(x) contains just one element.
argmax
x
(x) = .
Best response
Denition
= (S, W, u) .
The function s
R
: W 2
S
a best-response function (a best response, a
best answer) if s
R
is given by
s
R
(w) := arg max
sS
u (s, w) .
Problem
Use best-response functions to characterize s as a dominant strategy.
Hint: characterization means that you are to nd a statement that is
equivalent to the denition.
1
Introduction
2
3
4
5
Rationalizability
Probability distribution
Denition
Probability distribution on M : prob : 2
M
[0, 1] with
prob () = 0,
prob (A'B) = prob (A) + prob (B) for all A, B 2
M
obeying
A B = and
prob (M) = 1 (summing condition).
Subsets of M are also called events.
Problem
Probability for the events
A : "the number of pips (spots) is 2",
B : "the number of pips is odd", and
A'B
Mixed strategy
Denition
Let S be a nite strategy set. A mixed strategy is a probability
distribution on S :
(s) _ 0 for all s S and

sS
(s) = 1 (summing condition)
(set of mixed strategies) > also stands for sum, as above
(s) = 1 for one s S > identify s with
(s) > 0 and (s
/
) > 0 for s ,= s
/
> properly mixed strategy
Notation:
_
(s
1
) , (s
2
) , ...,
_
s
[S[
__
.
Mixed strategy
Decision situation in strategic form with mixed strategies
Denition
= (S, W, u)
Example
Choosing umbrella with probability
1
3
and sunshade with probability
2
3
is a
mixed strategy.
Belief
Denition
Let W be a set of states of the world.
(set of probability distributions on W);
(belief);
Notation:
_
(w
1
) , ...,
_
w
[W[
__
.
Example
Bad weather with probability
1
4
and good weather with probability
3
4
Extending the payo function
So far,
u : S W R
Two extensions:
probability distribution on W
rather than a specic state of the world w W
> lottery
mixed strategy on S
rather than a specic strategy s S
beliefs and lotteries
state of the world
bad weather,
1
4
good weather,
3
4
strategy
production
of umbrellas
100 81
production
of sunshades
64 121
For example,
L
umbrella
=
_
100, 81;
1
4
,
3
4
_
Lottery and expected value
Denition
A tuple
L = [x; p] := [x
1
, ..., x
`
; p
1
, ..., p
`
]
is called a lottery where x
j
R is the payo accruing with probability p
j
/ (set of simple lotteries).
` = 1
L is called a trivial lottery
identify L = [x; 1] with x.
Denition
Assume a simple L = [x
1
, ..., x
`
; p
1
, ..., p
`
] . Its expected value is denoted
by E (L) and given by
E (L) =
`
j =1
p
j
x
j
.
mixed strategies
Denition
The payo under and w is dened by
u (, w) :=

sS
(s) u (s, w)
Problem
Calculate the expected payo in the umbrella-sunshade decision situation
if the rm chooses umbrella with probability
1
3
and sunshade with
2
3
.
Dierentiate between w =bad weather and w ="good weather".
Thus, the payo for a mixed strategy is the mean of the payos for
the pure strategies.
How are best pure and best mixed strategies related?
Best pure and best mixed strategies
Lemma
Best pure and best mixed strategies are related by the following two claims:
Any mixed strategy that puts positive probabilities on best pure
strategies, only, is a best strategy.
If a mixed strategy is a best strategy, every pure strategy with
positive probability is a best strategy.
Mixing strategies and states of the world
Exercise
Problem
Consider again the umbrella-sunshade decision situation in which
the rm chooses umbrella with
1
3
and sunshade with probability
2
3
and
the weather is bad with probability
1
4
and good with probability
3
4
.
Calculate u
__
1
3
,
2
3
_
,
_
1
4
,
3
4
__
!
summary
The payo function u : S W R has been extended:
beliefs:
u : S R
(s, ) u (s, ) =

wW
(w) u (s, w) = E (L
s
)
mixed strategies
u : W R
(, w) u (, w) =

sS
(s) u (s, w)
both beliefs and mixed strategies
u : R
(, ) u (, ) =

sS

wW
(s) (w) u (s, w)
Best-response functions
Denition
Given = (S, W, u) , there are four best-response functions:
s
R,W
: W 2
S
, given by s
R,W
(w) := arg max
sS
u (s, w) ,
R,W
: W 2
, given by
R,W
(w) := arg max
u (, w) ,
s
R,
: 2
S
, given by s
R,
() := arg max
sS
u (s, ) , and
R,
: 2
, given by
R,
() := arg max
u (, )
s
R
or
R
instead of s
R,W
, etc. (if there is no danger of confusion)
Problem
Complete the sentence:
R,W
(w) implies (s) = 0 for all ... .
Theorem
Theorem
= (S, W, u) . We have
and
ss
R,
()
(s) = 1 imply
R,
() and

R,
() implies s s
R,
() for all s S with (s) > 0.
These implications continue to hold for W and w rather than and .
Example
w
1
w
2
s
1
4 1
s
2
1 2
Let := (w
1
) be the probability of w
1
. We have s
1
s
R,
() in the
case of
4 + (1 ) 1 _ 1 + (1 ) 2,
i.e., if _
1
4
holds.
Best-reply function is given by
R,
: 2
.
For ,=
1
4
, there is exactly one best strategy,
= 0 ( = (0, 1) = s
2
) or
= 1 ( = (1, 0) = s
1
).
=
1
4
implies that every pure strategy (> every mixed strategy) is
best.
R,
() =
_
_
_
1, >
1
4
[0, 1] , =
1
4
0, <
1
4
4
1
1
1

Exercise
Problem
Sketch the best-reply function
R,
for
w
1
w
2
s
1
1 3
s
2
2 1
Rationalizability
1
Introduction
2
3
4
5
Rationalizability
Rationalizable strategies
Example
w
1
w
2
s
1
4 4
s
2
1 5
s
3
5 1
Can a rational decider choose s
1
although it is not a best response to
w
1
and not a best response to w
2
?
Yes: the rational decider may entertain the belief on W with
(w
1
) = (w
2
) =
1
2
.
Given this belief, s
1
is a perfectly reasonable strategy.
Problem
Show s
1
s
R,
__
1
2
,
1
2
__
!
Rationalizable strategies
Denition
= (S, W, u) .
A mixed strategy is called rationalizable with respect to W if a
w W exists such that
R,W
(w).
Strategy is called rationalizable with respect to if a belief
exists such that
R,
().
Nobel price 2010
In 2010, the Sveriges Riksbank Prize in Economic Sciences in Memory of
Alfred Nobel was awarded to the economists
1/3 Peter A. Diamond (Massachusetts Institute of Technology,
Cambridge, MA),
1/3 Dale T. Mortensen (Northwestern University, Evanston, IL), and
1/3 Christopher A. Pissarides (London School of Economics and Political
Science)
for their analysis of markets with search frictions.
Further exercises: Problem 1
(a) If strategy s S strictly dominates strategy s
/
S and strategy s
/
strictly dominates strategy s
//
S, is it always true that strategy s
strictly dominates strategy s
//
?
(b) If strategy s S weakly dominates strategy s
/
S and strategy s
/
weakly dominates strategy s
//
S, is it always true that strategy s
weakly dominates strategy s
//
?
Further exercises: Problem 2
Prove the following assertions or give a counter-example!
(a) If is rationalizable with respect to W, then is rationalizable
with respect to .
(b) If s S is a weakly dominant strategy, then it is rationalizable with
respect to W.
(c) If s S is rationalizable with respect to W, then s is a weakly
dominant strategy.
Further exercises: Problem 3 (without gure)
Consider the problem of a monopolist faced with the inverse demand
function p(q) = a b q, in which a can either be high, a
h
, or low, a
l
.
The monopolist produces with constant marginal and average cost c.
Assume that a
h
> a
l
> c and b > 0. Think of the monopolist as setting
the quantity, q, and not the price, p.
(a) Formulate this monopolists problem as a decision problem in
strategic form. Determine s
R,W
!
(b) Assume a
h
= 6, a
l
= 4, b = 2, c = 1 so that you obtain the plot given
in the gure. Show that any strategy q /
_
a
l
c
2b
,
a
h
c
2b
_
is dominated
by either s
R,W
_
a
h
_
or s
R,W
_
a
l
_
. Show also that no strategy
q
_
a
l
c
2b
,
a
h
c
2b
_
dominates any other strategy q
/

_
a
l
c
2b
,
a
h
c
2b
_
.
(c) Determine all rationalizable strategies with respect to W.
(d) Dicult: Determine all rationalizable strategies with respect to .
Hint: Show that the optimal output is a convex combination of
s
R,W
_
a
h
_
and s
R,W
_
a
l
_
.
Further exercises: Problem 3 (gure)
0.0 0.5 1.0 1.5 2.0 2.5 3.0
0
1
2
3
4
quantity
profit

Advanced Microeconomic

Enviado por

Dados do documento

Descrição original:

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Advanced Microeconomic

Enviado por

Direitos autorais:

Formatos disponíveis

Advanced Microeconomics

Harald Wiese (University of Leipzig) Advanced Microeconomics Winter term 2012/2013 61 / 70

Você também pode gostar