EC537 Slides Lecture 1 PDF

EC537 Microeconomic Theory for Research Students,
Part II: Lecture 1
Leonardo Felli
CLM.G.4
8 November 2011
Course Outline
Topics in Contract Theory
Lecture 1: Contracts what are they? The Coase Theorem, Hidden

Information, Bilateral trading.
Lecture 2: Hidden Action, Multi-Tasking, Informed Principal,

Intertemporal Incentives.
Lecture 3: Implementation and the Foundations of Contracting with

Unverifiable Information.
Lecture 4: Transaction Costs.
Lecture 5: Hold-Up Problem, Specific Investments and Competition.
Leonardo Felli (LSE) EC537 Microeconomic Theory for Research Students, Part II:
8 November
Lecture 1 2011 2 / 79
Admin
My coordinates: S.478, x7525, lfelli@econ.lse.ac.uk
PA: Elizabeth Mirhady, S.686, e.mirhady@lse.ac.uk.
Office Hours:
Wednesday 3:30-4:30 p.m.
or by appointment (e-mail lfelli@econ.lse.ac.uk).
Course Material: available at:

http://econ.lse.ac.uk/staff/lfelli/teaching
8 November
Lecture 1 2011 3 / 79
References: Contract Theory
Oliver Hart, Firms Contracts and Financial Structure, Oxford: Oxford

University Press, 1995.
Jean-Jacques Laffont and David Martimort, The Theory of

Inncentives: The Principal-Agent Model, Princeton and Oxford:
Princeton University Press, 2002.
Patrick Bolton and Mathias Dewatripont, Contract Theory,

Cambridge: M.I.T. Press, 2004.
8 November
Lecture 1 2011 4 / 79
The Contract
The first natural question that needs to be answered is:
What is a contract?
Definition
A contract is the ruling of an economic transaction: the description of the
performance that the contracting parties agree to complete at a (possibly
future) date.
8 November
Lecture 1 2011 5 / 79
Example: a contract for the purchase of a specific item, say a meal. It
specifies:
the restaurant’s performance (number of courses, quality of food,

cooking details, etc. . . ),
the customer’s performance (payment in full upon completion).
Contracts involve not only the contracting parties, but also outsiders
(enforcing authority: Court or Enforcer).
8 November
Lecture 1 2011 6 / 79
We distinguish between implicit and explicit contracts.
A contract is implicit or self-enforcing whenever the environment in

which the contracting parties operate corresponds to the extensive
form of a game whose (unique) subgame perfect Nash equilibrium
(PBE) exactly corresponds to the outcome the parties would like to
implement.
If you believe in SPE or PBE then there is no need for explicit

communication. The two rational individuals will behave in the way
required.
8 November
Lecture 1 2011 7 / 79
If the outcome the parties would like to implement is not the
subgame perfect Nash equilibrium of the environment in which they
operate the parties might want to modify the environment.
This is accomplished through an explicit contract.
An explicit contract is a commitment device requiring:
an explicit agreement between the parties,
the intervention of a third party: the enforcer.
8 November
Lecture 1 2011 8 / 79
The role of the enforcer is to force the parties to behave in a way that
differs from the one that would arise in the absence of any agreement.
An explicit contract therefore specifies a new extensive form

corresponding to a new game for the parties.
The usual way for the enforcer to guarantee that the parties operate
in this new environment is by modifying the parties’ payoffs, when
necessary.
By agreeing to bring in an enforcer in the game the parties commit to

play a game that differs from the initial one they were in.
8 November
Lecture 1 2011 9 / 79
To see how the presence of an enforcer may work consider the
following example: (Kreps, 1984)
A buyer B and a seller S wish to trade an indivisible item at date 1.
The buyer’s valuation: v ,
The seller’s delivery cost: c.
Let
v >c
In other words, trade is socially efficient.
Leonardo Felli (LSE) EC537 Microeconomic Theory for Research Students, Part8II:
November
Lecture 12011 10 / 79
Let p be a reasonable price level (we abstract for the moment from
bargaining) such that:
v > p > c.
B’s and S’s situation may be described by the following normal form:
deliver not deliver

pay p v − p, p − c −p, p
not pay p v , −c 0, 0
November
Lecture 12011 11 / 79
The unique Nash equilibrium (dominant solvable) is:
(B does not pay, S does not deliver).
This is clearly an inefficient outcome: no trade.
The situation does not change if any of the following two extensive
forms are played.
November
Lecture 12011 12 / 79
The unique SPE of the following game is:
{B does not pay, S does not deliver at both nodes}.
Bb
@
pay p @ not pay p
@
@
S @S
q @q
L L
L L
deliver L not deliver L not
L L
L deliver L deliver
L L
q Lq q Lq
(v − p, p − c) (−p, p) (v , −c) (0, 0)
November
Lecture 12011 13 / 79
The unique SPE of the following game is:
{S does not deliver, B does not pay at both nodes, }.
Sb
@
deliver @
@not deliver
@
B @B
q @q
L L
L L
pay p L not pay p L not
L pay p L pay p
L L
L L
q Lq q Lq
(p − c, v − p) (−c, v ) (p, −p) (0, 0)
November
Lecture 12011 14 / 79
Solution: to this inefficiency is an explicit contract enforced by a third
party (enforcer).
It specifies:
the payment p that B is supposed to make contingent on S delivering

the item,
the punishment FB > p (implicit in the legal system) imposed by the

enforcer on B in the event that S delivers and B does not pay,
the punishment FS > c (implicit in the legal system) imposed by the

enforcer on S in the event that B pays but S does not deliver.
November
Lecture 12011 15 / 79
In this case the normal form describing the contracting parties
problem once the contract is in place is:
deliver not deliver

pay p v − p, p − c FS − p, p − FS
not pay p v − FB , FB − c 0, 0
The unique Nash equilibrium is now:
(B pays p, S delivers).
This particular contract is budget balanced off-the-equilibrium-path

(renegotiation proof). The latter property does not always hold.
November
Lecture 12011 16 / 79
Consider now an environment in which when a party goes to the
enforcer (goes to court) detection is costly (κ) and is successful only
with probability π.
The payoffs associated with (not pay p, deliver) are:
v − π (FB + κ), π FB − (1 − π)κ − c
The payoffs associated with (pay p, not deliver) are:
π FS − (1 − π)κ − p, p − π (FS + κ)
November
Lecture 12011 17 / 79
Notice that as deterrence goes: the detection probability (policing,
monitoring) π and the size of the punishment, FB and FS , are
substitutes (Becker 1968).
The game, below, assumes that the enforcer’s costs κ are paid by the
loosing party (British system):
deliver not deliver

π FS − (1 − π)κ − p,
pay p v − p, p − c
p − π (FS + κ)
v − π (FB + κ),
not pay p 0, 0
π FB − (1 − π)κ − c
If court’s costs κ are too high the game has multiple Nash equilibria:
(pay p, deliver) and (not pay p, not deliver).
November
Lecture 12011 18 / 79
This example clearly shows the need for an enforcement mechanism.
This mechanism may be due to:
the parties being involved in a repeated relationship

relationship/implicit contracting, (multiplicity might be a problem).
the presence of a legal system that enforces the parties agreement

(explicit contracting).
Notice that according to this interpretation the enforcer is essentially

a commitment device available to the parties that can be used when
the parties agree to call it in.
November
Lecture 12011 19 / 79
An alternative interpretation is that the enforcer itself is one of the
players of the game.
It should therefore be endowed with a payoff function and an action

space and should be explicitly considered in the analysis of the
contractual situation (we will come back to this).
It should be mentioned that using this line of argument one could

obtain a rather extreme interpretation of a contract (a law) (Mailath,
Morris and Postlewaite 2000).
November
Lecture 12011 20 / 79
The view is that enforcement/punishment is the only relevant activity.
A contract (a law) can at best be interpreted as cheap talk that allows

the parties to coordinate on a particular equilibrium of the game.
No new equilibrium is introduced by the parties agreeing on a

contract or by the parliament passing a law.
November
Lecture 12011 21 / 79
From now on we will assume that the two (or more) parties involved
in the contractual relationship operate in a market economy with a
well functioning legal system.
Whatever contract the parties agree to, it will be enforced by the

court.
The penalties for breaching the contract will be assumed to be

sufficiently severe that no contracting party will ever consider the
possibility of not honoring the contract.
We will abstract from explicitly specifying these penalties.
November
Lecture 12011 22 / 79
Coase Theorem:
Once we have established what a contract is and how it works the

next natural question is:
What could parties achieve in an economic environment in which they

can costlessly negotiate a contractual agreement?
The answer to this question is the celebrated Coase Theorem.
November
Lecture 12011 23 / 79
Theorem (Coase Theorem: Coase (1960))
In an economy where ownership rights are well defined and transacting is
costless gains from trade will be exploited (a contract will be agreed upon)
and efficiency achieved whatever the distribution of entitlements.
That is rational agents write contracts that are individually rational and
Pareto efficient.
A contract is individually rational if each contracting party is not worse off

by deciding to sign the contract rather then choosing not to sign it.
November
Lecture 12011 24 / 79
This is the reflection of an other basic principle of a well functioning
legal system known as: freedom of contract.
This is equivalent to assume that the action space of the contracting

parties always contains the option not to sign the contract.
A contract is Pareto efficient if there does not exists an other feasible

contract that makes at least one of the contracting party strictly
better off without making any other contracting party worse off.
November
Lecture 12011 25 / 79
Consider the following simple model of a production externality.
Consider two parties, labelled A and B.
Party A generates revenue RA (eA ) (strictly concave) by choosing the

input eA at a linear cost c eA (c > 0).
A’s payoff function is then:
ΠA (eA ) = RA (eA ) − c eA
November
Lecture 12011 26 / 79
Party B generates revenue RB (eB ) (strictly concave) by choosing the
input eB at the linear cost c eB (c > 0).
Party B also suffers from an externality γ eA (γ > 0) imposed by A on

B.
B’s payoff function is then:
ΠB (eB ) − γ eA = RB (eB ) − c eB − γ eA
November
Lecture 12011 27 / 79
Assume first that the parties choose the amounts of input eA and eB
simultaneously and independently without any prior agreement.
Party A’s problem:

max ΠA (eA )
eA
Party B’s problem:

max ΠB (eB ) − γ eA
eB
In equilibrium the inputs chosen (êA , êB ) are:
RA0 (êA ) = c, RB0 (êB ) = c
November
Lecture 12011 28 / 79
Consider now the social efficient amounts of input eA∗ and eB∗ .
These solve the problem:
max ΠA (eA ) + ΠB (eB ) − γ eA

eA ,eB
In other words (eA∗ , eB∗ ) are such that:
RA0 (eA∗ ) = c + γ
RB0 (eB∗ ) = c
November
Lecture 12011 29 / 79
Comparing (êA , êB ) and (eA∗ , eB∗ ) we obtain using concavity of RA (·):
eB∗ = êB , eA∗ < êA
In other words:
ΠA (eA∗ ) + ΠB (eB∗ ) − γ eA∗ − [ΠA (êA ) + ΠB (êB ) − γ êA ] =
= [ΠA (eA∗ ) − ΠA (êA )] + γ (êA − eA∗ ) > 0
The joint surplus is reduced by the inefficiency generated by the

externality.
November
Lecture 12011 30 / 79
Assume now that the two contracting parties have the opportunity to
get together and agree on a contract before the amounts of input are
chosen.
There exists strictly positive gains from trade.
A reduction in the amount of input eA from êA to eA∗ will generate:
a decrease in the net revenues from A’s technology:
ΠA (eA∗ ) < ΠA (êA )
reduction in the negative externality
γ eA∗ < γ êA
November
Lecture 12011 31 / 79
The former effect is more than compensated by the latter one. This
may create room for negotiation.
Normalize for simplicity the total size of the surplus that is available
to share between the two contracting parties to have size 1 (simple
normalization).
To establish a well defined negotiation ownership rights need to be

specified.
Entitlements/ownership rights define the outside option of each party

to the contract.
In other words they define the payoff each party is entitled to without
need for the other party to agree.
November
Lecture 12011 32 / 79
Denote wA and wB the entitlements of party A, respectively B where:
wA + wB < 1.
We assume the following extensive form for the costless negotiation

between the two parties:
Infinite horizon, alternating offers bargaining game with discounting

and outside options.
November
Lecture 12011 33 / 79
Denote:
δ the parties’ common discount factor,
x the share of the pie to party A,
(1 − x) the share of the pie to party B.
November
Lecture 12011 34 / 79
Extensive Form:
Odd periods:
Stage I : A makes an offer xA to B,
Stage II : B observes the offer and has three alternatives:
he can accept the offer, then x = xA and the game

terminates;
he can reject the offer and take his outside option wB

and the game terminates;
he can reject the offer and do not take his outside

option, then the game moves to Stage I of the following
period.
November
Lecture 12011 35 / 79
Even periods:
Stage I : B makes an offer xB to A,
Stage II : A observes the offer and ha three alternative choices:
he can accept the offer, then x = xB and the game

terminates;
he can reject the offer and take his outside option wA

and the game terminates;
he can reject the offer and do not take his outside

option, then the game moves to Stage I of the following
period.
November
Lecture 12011 36 / 79
Payoffs:
If parties agree on x in period n + 1:
πA (σA , σB ) = δ n x, πB (σA , σB ) = δ n (1 − x),
If they do not agree and either party takes his outside option in period
n + 1:
πA (σA , σB ) = δ n wA , πB (σA , σB ) = δ n wB .
November
Lecture 12011 37 / 79
Theorem (Deal Me Out)
For any discount factor δ < 1, and any pair (wA , wB ), wA + wB < 1, the
bargaining game has a unique subgame perfect equilibrium.
Agreement between the parties is immediate and the outside options are
never exercised.
November
Lecture 12011 38 / 79
Proof: (sketch)
Denote xiH , respectively xiL , i ∈ {A, B}, the highest, respectively the
lowest, possible share that A can receive in a subgame that starts
with i making the offer.
We then have that:

xBH ≤ max{wA , δ xAH }, 1 − xAL ≤ max{wB , δ 1 − xBL }
Moreover:

xBL ≥ max{wA , δ xAL }, 1 − xAH ≤ max{wB , δ 1 − xBH }
November
Lecture 12011 39 / 79
Solving these inequalities we obtain:
xAH = xAL = xA , xBH = xBL = xB
We also obtain that:
If
δ δ
wA ≤ , wB ≤
1+δ 1+δ
then
1 δ
xA = , xB =
1+δ 1+δ
November
Lecture 12011 40 / 79
If
δ
wA ≥ , wB ≤ δ(1 − wA )
1+δ
then
xA = 1 − δ(1 − wA ), xB = wA
If
δ
wA ≤ δ(1 − wB ), wB ≥
1+δ
then
xA = 1 − wB , xB = δ(1 − wB )
If
wA ≥ δ(1 − wB ), wB ≥ δ(1 − wA )
then
xA = 1 − wB , xB = wA
November
Lecture 12011 41 / 79
These offers characterize a pair of strategies (σA , σB ).
It is easy to show that these strategies constitute the unique subgame

perfect equilibrium of the bargaining game.
Notice that an efficient agreement is reached independently of the

size of the entitlements.
November
Lecture 12011 42 / 79
In particular if each party is entitle to the choice of his input, then:
ΠA (êA )
wA =
ΠA (eA∗ ) + ΠB (eB∗ ) − γ eA∗
ΠB (êB ) − γ êA
wB =
ΠA (eA∗ ) + ΠB (eB∗ ) − γ eA∗
If instead party B is entitled to preclude party A from operating his

technology, then:
ΠB (êB )
wA = 0, wB =
ΠA (eA∗ ) + ΠB (eB∗ ) − γ eA∗
November
Lecture 12011 43 / 79
In either case the result above implies that we would get the efficient
outcome: (eA∗ , eB∗ ).
However, the share that accrues to each party depends on the

entitlements wA and wB .
The equilibrium contract specifies a transfer between the two parties

and A’s choice of input eA∗ .
Also the transfer depend on the entitlements wA and wB .
November
Lecture 12011 44 / 79
From now on we are going to focus on models in which the Coase
Theorem fails.
The classic cause for the failure of the Coase Theorem is the presence
of asymmetric information between the parties.
This is a situation in which each party has private information on his

own preferences (hidden information model).
Recall that a game of incomplete information (a player does not know

the preferences of one opponent) can always be recast as a game of
imperfect information (a player does not know the history of the
game) (Harsanyi 1967).
November
Lecture 12011 45 / 79
Asymmetric Information:
We are going to consider first a very simple model of bargaining

under bilateral asymmetric information (a specific extensive form)
with no externalities.
We will show that in this situation efficiency cannot be achieved.
Recall that the Coase Theorem implies efficiency even in the presence
of externalities therefore if inefficiency arises in the absence of
externalities we can conclude that the Theorem fails.
November
Lecture 12011 46 / 79
Notice however that this does not imply that we cannot find an
extensive form that will achieve efficiency.
Fortunately an other fundamental principle of contract theory will

help in this case: Revelation Principle.
Using the revelation principle we will be able to conclude that

efficiency cannot be achieved whatever extensive form governs the
bargaining between the two parties under bilateral asymmetric
information.
November
Lecture 12011 47 / 79
Consider the following simple model of bilateral trade (double
auction) (Chatterjee and Samuelson, 1983).
Two players, a buyer and a seller: N = {b, s}.
The seller names an asking price: ps .
The buyer names an offer price: pb .
November
Lecture 12011 48 / 79
The action spaces:
As = {ps ≥ 0}, Ab = {pb ≥ 0}.
The seller owns and attaches value vs to an indivisible unit of a good.
The buyer attaches value vb to the unit of the good and is willing to
pay up to vb for it.
The valuations for the unit of the good of the seller and the buyer are
their private information of each player.
Player i ∈ {b, s} believes that the valuation of the opponent v−i

takes values in the unit interval.
November
Lecture 12011 49 / 79
The type spaces:
Ts = {0 ≤ vs ≤ 1}, Tb = {0 ≤ vb ≤ 1}
Player i ∈ {b, s} also believes that the valuation of the opponent is

uniformly distributed on [0, 1]:
µs = 1, µb = 1.
The extensive form of the game is such that:

If pb ≥ ps then they trade at the average price:
(ps + pb )
p= .
2
If pb < ps then no trade occurs.
November
Lecture 12011 50 / 79
The payoffs to both the seller and the buyer are then:

 (ps + pb )
if pb ≥ ps
us (ps , pb ; vs , vb ) =
 v 2 if pb < ps
s
and 
(ps + pb )
vb − if pb ≥ ps

ub (ps , pb ; vs , vb ) = 2
 0 if pb < ps
Players’ strategies: ps (vs ) and pb (vb ). We consider strictly monotonic

and differentiable strategies.
November
Lecture 12011 51 / 79
Consider now the seller’s best reply.
This is defined by the following maximization problem:
max Evb {us (ps , pb ; vs , vb ) | vs , pb (vb )}

ps
Consider now the seller’s payoff, substituting pb (vb ) we have:


 (ps + pb (vb ))
if pb (vb ) ≥ ps
us = 2
 v if pb (vb ) < ps
s
November
Lecture 12011 52 / 79
or 
 (ps + pb (vb ))
if vb ≥ pb−1 (ps )
us = 2
 v
s if vb < pb−1 (ps )
The seller’s maximization problem is then:

Z pb−1 (ps ) Z 1
(ps + pb (vb )
max vs dvb + dvb
ps vb =0 vb =pb−1 (ps ) 2
November
Lecture 12011 53 / 79
Recall that by Leibniz’s rule:
!
Z β(y )
∂
G (x, y )dx =
∂y α(y )
= G (β(y ), y ) β 0 (y ) − G (α(y ), y )α0 (y ) +
Z β(y )
∂G (x, y )
+ dx
α(y ) ∂y
November
Lecture 12011 54 / 79
Therefore the first order conditions are:
dpb−1 (ps ) 1 dpb−1 (ps )
vs − ps + pb (pb−1 (ps )) +
dps 2 dps
Z 1
1
+ dvb = 0
pb−1 (ps ) 2
or from ps = pb (pb−1 (ps )):
dpb−1 (ps ) 1 1
(vs − ps ) + vb p−1 (ps ) = 0
dps 2 b
which gives:
dpb−1 (ps ) 1
1 − pb−1 (ps ) = 0

(vs − ps ) +
dps 2
November
Lecture 12011 55 / 79
The buyer’s best reply is instead defined by:
max Evs {ub (ps , pb ; vs , vb ) | vb , ps (vs )}

pb
Consider now the buyer’s payoff obtained substituting ps (vs ):


(ps (vs ) + pb )
vb − if vs ≤ ps−1 (pb )

ub = 2
 0 if vs > ps−1 (pb )
we then get
Z ps−1 (pb )
(ps (vs ) + pb )
max vb − dvs
pb vs =0 2
November
Lecture 12011 56 / 79
Therefore the first order conditions are:
(ps (ps−1 (pb )) + pb ) dps−1 (pb )

vb − +
2 dpb
Z ps−1 (pb )
1
− dvs = 0
2 vs =0
or
dps−1 (pb ) 1 ps−1 (pb )
[vb − pb ] − vs 0 =0
dpb 2
which gives:
dps−1 (pb ) 1 −1
(vb − pb ) − ps (pb ) = 0
dpb 2
November
Lecture 12011 57 / 79
To simplify notation we re-write pb−1 (·) = qb (·) and ps−1 (·) = qs (·).
The two differential equations that define the best reply of the seller
and the buyer are then:
1
[qs (ps ) − ps ] qb0 (ps ) − [1 − qb (ps )] = 0
2
1
[qb (pb ) − pb ] qs0 (pb ) − qs (pb ) = 0
2
November
Lecture 12011 58 / 79
Solving the second equation for qb (pb ) and differentiating yields:
qs (pb )qs00 (pb )

0 1
qb (pb ) = 3−
2 [qs0 (pb )]2
Substituting this expression into the first differential equation we get:
qs (ps )qs00 (ps )

qs (ps )
[qs (ps ) − ps ] 3 − − 1 − ps − 0 =0
[qs0 (ps )]2 qs (ps )
November
Lecture 12011 59 / 79
This is a second-order differential equation in qs (·) that has a
two-parameter family of solutions.
The simplest family of solution takes the form:
qs (ps ) = α ps + β
Then the values α = 3/2 and β = −3/8 solve the second-order

differential equation.
The definition of qs (·) and qb (·) imply that:

2 1 2 1
ps = vs + , pb = vb +
3 4 3 12
November
Lecture 12011 60 / 79
This is the (unique) Bayesian Nash equilibrium of this game.
Notice now that it is efficient to trade whenever:
vb ≥ vs
However in this double auction game trade occurs whenever:
pb ≥ ps
or
2 1 2 1
vb + ≥ vs +
3 12 3 4
November
Lecture 12011 61 / 79
In other words, in equilibrium trade occurs whenever:
1
vb ≥ vs +
4
vb 6
vs = vb
1 .................................................................
..
..
Trade ..
..
..
..
..
..
.
vb = vs + 4 .... 1
..
..
p 6 .. -
(0, 0) vs
1
November
Lecture 12011 62 / 79
Revelation Principle:
The obvious question is now: how can we make sure that there does
not exists an alternative way for the parties to achieve efficiency?
The tool that allows us to give an answer to this question is:

Revelation Principle
The Revelation Principle (Green and Laffont 1977, Myerson 1979,

Harris and Townsend 1981, Dasgupta, Hammond and Maskin 1979)
greatly simplify the set of feasible mechanisms for the parties.
November
Lecture 12011 63 / 79
This Revelation principle says that there is no loss in generality in
restricting attention to direct revelation mechanisms that satisfy
truth-telling constraints.
Recall:
the indirect mechanism is the one in which parties agree to a trade, set
prices etc. . .
the direct mechanism is the one in which parties report their private
information to a mechanism designer who according to the reports
enforces the mechanism.
Looking for the truth-telling equilibrium of the direct mechanism that

maximizes the principal’s utility is the way to identify the best
possible indirect mechanism from the principal’s view point.
November
Lecture 12011 64 / 79
Since every BNE of every indirect mechanism has an associated
truth-telling BNE of a direct mechanism if we find the truth-telling
BNE of the direct mechanism that maximizes the principal’s utility
there cannot exist any BNE of the indirect mechanism that is better
for the principal.
Notice that this way to proceed does not require us to specify the
space of all possible indirect mechanisms.
It is critical that the principal can commit to the mechanism in

advance: renegotiation may lead to a failure of the revelation
mechanisms (Dewatripont, 1989).
November
Lecture 12011 65 / 79
Bilateral Trade (Myerson and Satterthwaite 1983)
In our setting there is no principal, but the two parties at an ex-ante

stage — before they learn their private information — will commit to
a mechanism via the contract.
They will choose their contract in a way that maximizes their ex-ante
welfare.
Assume further that this is a pure bilateral contract transfers cannot

involve a third party.
In jargon the contract has to be budget balancing.
November
Lecture 12011 66 / 79
A seller and a buyer trade a single unit of a good.
The seller’s cost of delivering is c and it is the seller’s private

information:
c ∼ PS (c), c ∈ [c, c]
The buyer’s valuation is v and it is the buyer’s private information:
v ∼ PB (v ), v ∈ [v , v ]
A contract in this environment is a pair (φ, t) where

φ is the probability of trade,
t is the transfer from the buyer to the seller.
November
Lecture 12011 67 / 79
By revelation principle restrict attention to truth-telling direct
mechanisms.
The seller’s indirect utility is then:
US (ĉ, v |c) = t(ĉ, v ) − φ(ĉ, v ) c
The buyer’s indirect utility is instead:
UB (c, v̂ |v ) = φ(c, v̂ ) v − t(c, v̂ )
Denote:
US (ĉ) = Ev [t(ĉ, v ) − φ(ĉ, v ) c] = t(ĉ) − φ(ĉ) c
UB (v̂ ) = Ec [φ(c, v̂ ) v − t(c, v̂ )] = φ(v̂ ) v − t(v̂ )
November
Lecture 12011 68 / 79
Therefore every truth-telling BNE direct mechanism has to satisfy the
following set of incentive compatibility constraints (IC):
US (c) ≥ t(ĉ) − φ(ĉ) c ∀c, ĉ ∈ [c, c]
UB (v ) ≥ φ(v̂ ) v − t(v̂ ) ∀v , v̂ ∈ [v , v ]
Since once again we insist on freedom of contract we also require the

following individual rationality constraints (IR) to be satisfied:
US (c) ≥ 0, ∀c, ĉ ∈ [c, c]
UB (v ) ≥ 0, ∀v , v̂ ∈ [v , v ]
November
Lecture 12011 69 / 79
Notice now that both parties’ preferences satisfy the Spence-Mirrlees
single crossing conditions:

∂ ∂UB /∂t
− >0
∂v ∂UB /∂φ
The following result helps us to write (IC) and (IR) constraints in a

manageable form.
November
Lecture 12011 70 / 79
Theorem (Myerson and Satterthwaite 1983)
For any probability φ(c, v ) there exists a transfer function t(c, v ) that
satisfies (IR) and (IC) if and only if:
Ec,v [φ(c, v ) (JB (v ) − JS (c))] ≥ 0
where

1 − PB (v ) PS (c)
JB (v ) = v − , JS (c) = c +
pB (v ) pS (c)
and
dφ(c) dφ(v )
≤ 0, ≥0
dc dv
November
Lecture 12011 71 / 79
Proof:
From the (IC) constraints since for every v̂ > v we must have:
UB (v ) ≥ φ(v̂ ) v − t(v̂ ) = UB (v̂ ) − (v̂ − v ) φ(v̂ ),
UB (v̂ ) ≥ φ(v ) v̂ − t(v ) = UB (v ) − (v − v̂ ) φ(v ),
Summing the two inequalities we get:
(v̂ − v ) φ(v̂ ) ≥ (v̂ − v ) φ(v )
Dividing by (v̂ − v ) and letting v̂ tend to v we obtain:
UB0 (v ) = φ(v )
Since φ(v , c) ∈ [0, 1] we obtain for every v ∈ [v , v ]:
UB (v ) ≥ UB (v )
November
Lecture 12011 72 / 79
We therefore conclude that the only relevant (IR) constraint is:
UB (v ) ≥ 0
Notice that since for every v̂ > v we have:
(v̂ − v ) φ(v̂ ) ≥ (v̂ − v ) φ(v )
we also obtain that:

dφ(v )
≥0
dv
Symmetrically for the seller we can prove that (IC) constraint implies:
dφ(c)
≤0
dc
Consider now the differential equation obtained above:
UB0 (v ) = φ(v )
November
Lecture 12011 73 / 79
Integrating it we obtain:
Z v
UB (v ) = UB (v ) + φ(ν)dν
v
and symmetrically for the seller we obtain:

Z c
US (c) = US (c) + φ(γ)dγ
c
By budget balancing we now get:
0 = Ec [t(c)] − Ev [t(v )] =
Z c Z c
= φ(c) c + φ(γ)dγ pS (c) dc + US (c) +
c c
Z v Z v
+ φ(ν)dν − v φ(v ) pB (v )dv + UB (v )
v v
November
Lecture 12011 74 / 79
Integrating by parts we get:
US (c) + UB (v ) =
Z c
PS (c)
= − c+ φ(c) pS (c) dc +
c pS (c)
v
1 − PB (v )
Z
+ v− φ(v ) pB (v )dv
v pB (v )
or
US (c) + UB (v ) = Ec,v [φ(c, v ) (JB (v ) − JS (c))]
Since (IR) is such that US (c) ≥ 0 and UB (v ) ≥ 0 then:
Ec,v [φ(c, v ) (JB (v ) − JS (c))] ≥ 0
November
Lecture 12011 75 / 79
Sufficiency is a bit more complex to prove it requires us to solve the
partial differential equation that is represented by the FOC of the (IC)
constraints.
The parties’ ex-ante problem is now:
max Ec,v [φ(c, v ) (v − c)]

φi
s.t. Ec,v [φ(c, v ) (JB (v ) − JS (c))] ≥ 0
dφ(c) dφ(v )
≤ 0, ≥0
dc dv
November
Lecture 12011 76 / 79
Ignoring monotonicity conditions and denoting µ the lagrange
multiplier of the remaining constraint we get a lagrangian function
that is linear in φi :

µ 1 − PB (v ) PS (c)
Ec,v φ(c, v ) (v − c) − −
1−µ pB (v ) pS (c)
The solution is to set φ = 1 if and only if the term in brackets is

strictly positive.
In other words trade occurs if and only if, for µ ≥ 0:

µ 1 − PB (v ) µ PS (c)
v− ≥c+
1−µ pB (v ) 1 − µ pS (c)
November
Lecture 12011 77 / 79
This φ(c, v ) is weakly monotonic in:

µ 1 − PB (v )
v−
1−µ pB (v )
and
µ PS (c)
c+
1−µ pS (c)
Then MHRP implies that both monotonicity conditions are satisfied

and hence local and global (IC) holds.
Clearly if µ > 0 there will be inefficiencies in trade.
November
Lecture 12011 78 / 79
Theorem (Myerson and Satterthwaite 1983)
If c > v and v > c then necessarily µ > 0.
Proof: Immediate by substituting the efficient probabilities φi into the

constraint of the parties problem and showing that is violated.
Clearly with bilateral asymmetric information the Coase Theorem fails in a

very relevant sense: efficiency is no longer guaranteed by the use of
contracts.
November
Lecture 12011 79 / 79

EC537 Slides Lecture 1 PDF

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

EC537 Slides Lecture 1 PDF

Загружено:

Авторское право:

Доступные форматы

EC537 Microeconomic Theory for Research Students,

Part II: Lecture 1

Topics in Contract Theory

Lecture 1: Contracts what are they? The Coase Theorem, Hidden

Lecture 2: Hidden Action, Multi-Tasking, Informed Principal,

Lecture 3: Implementation and the Foundations of Contracting with

Lecture 4: Transaction Costs.

Lecture 5: Hold-Up Problem, Specific Investments and Competition.

My coordinates: S.478, x7525, lfelli@econ.lse.ac.uk

PA: Elizabeth Mirhady, S.686, e.mirhady@lse.ac.uk.

Course Material: available at:

Oliver Hart, Firms Contracts and Financial Structure, Oxford: Oxford

Jean-Jacques Laffont and David Martimort, The Theory of

Patrick Bolton and Mathias Dewatripont, Contract Theory,

The first natural question that needs to be answered is:

the restaurant’s performance (number of courses, quality of food,

the customer’s performance (payment in full upon completion).

A contract is implicit or self-enforcing whenever the environment in

If you believe in SPE or PBE then there is no need for explicit

This is accomplished through an explicit contract.

An explicit contract is a commitment device requiring:

an explicit agreement between the parties,

the intervention of a third party: the enforcer.

An explicit contract therefore specifies a new extensive form

By agreeing to bring in an enforcer in the game the parties commit to

A buyer B and a seller S wish to trade an indivisible item at date 1.

The buyer’s valuation: v ,

The seller’s delivery cost: c.

deliver not deliver

(B does not pay, S does not deliver).

This is clearly an inefficient outcome: no trade.

{B does not pay, S does not deliver at both nodes}.

{S does not deliver, B does not pay at both nodes, }.

the payment p that B is supposed to make contingent on S delivering

the punishment FB > p (implicit in the legal system) imposed by the

the punishment FS > c (implicit in the legal system) imposed by the

deliver not deliver

The unique Nash equilibrium is now:

This particular contract is budget balanced off-the-equilibrium-path

The payoffs associated with (not pay p, deliver) are:

v − π (FB + κ), π FB − (1 − π)κ − c

The payoffs associated with (pay p, not deliver) are:

deliver not deliver

This mechanism may be due to:

the parties being involved in a repeated relationship

the presence of a legal system that enforces the parties agreement

Notice that according to this interpretation the enforcer is essentially

It should therefore be endowed with a payoff function and an action

It should be mentioned that using this line of argument one could

A contract (a law) can at best be interpreted as cheap talk that allows

No new equilibrium is introduced by the parties agreeing on a

Whatever contract the parties agree to, it will be enforced by the

The penalties for breaching the contract will be assumed to be

We will abstract from explicitly specifying these penalties.

Once we have established what a contract is and how it works the

What could parties achieve in an economic environment in which they

The answer to this question is the celebrated Coase Theorem.

A contract is individually rational if each contracting party is not worse off

This is equivalent to assume that the action space of the contracting

A contract is Pareto efficient if there does not exists an other feasible

Consider two parties, labelled A and B.

Party A generates revenue RA (eA ) (strictly concave) by choosing the

A’s payoff function is then: