Scribd
Upload
152 views

Expected Utility Theory: XX X X P P P

The document discusses expected utility theory as it relates to principal-agent problems. It introduces the concepts of risk neutrality, risk aversion, and moral hazard. Under symmetric information, the principal designs a contract to maximize their payoff while satisfying the agent's participation constraint. This leads to an efficient allocation of risk depending on the risk preferences of the principal and agent. Under asymmetric information, like moral hazard, the principal must consider that the agent's actions may not be observable.

Uploaded by

gopshal
Copyright
© Attribution Non-Commercial (BY-NC)
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
152 views

Expected Utility Theory: XX X X P P P

The document discusses expected utility theory as it relates to principal-agent problems. It introduces the concepts of risk neutrality, risk aversion, and moral hazard. Under symmetric information, the principal designs a contract to maximize their payoff while satisfying the agent's participation constraint. This leads to an efficient allocation of risk depending on the risk preferences of the principal and agent. Under asymmetric information, like moral hazard, the principal must consider that the agent's actions may not be observable.

Uploaded by

gopshal
Copyright
© Attribution Non-Commercial (BY-NC)
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/ 6

Introduction

Expected Utility Theory


We will consider models where players face uncertainty. Their final profit is a random variable. Profit of the principal is W = ( ~ x w~ x ) (benefit wage) Profit of the agent is w~ x. They take actions that maximize their expected utility. In the case of a discrete random variable. x probability x1 p1 x2 p2 x3 p3

The expected utility of the principal is given by

B = p1B ( x1 wx1 ) + p2 B( x2 wx 2 ) + p3 B( x3 wx3 ) U = p1u ( wx1 ) + p2u ( wx 2 ) + p3u( wx3 )

The expected utility of the agent is given by (exempted from cost of effort)

We say that an individual is risk neutral if he is indifferent between accepting a gamble and receiving for sure the expected prize of the gamble:

~ )) = E ( w ~) (E(w

We say that an individual is risk averse if he is prefers receiving for sure the expected prize of the gamble to gambling:

~ )) > E ( w ~) (E(w

If the function If the function If the function If the function

B(.) is concave: principal is risk averse B(.) is linear: principal is risk neutral u(.) is concave: agent is risk averse u(.) is linear: principal is risk neutral

General approach
I.1 Elements of the Problem Principal: contractor Agent: contractee

Timing of relationship considered 1-Principal offers the contract Contact composed of elements that observable and verifiable. 2-Agent accepts or rejects Reservation Utility: minimum utility guaranteed to agent 3-Agent carries out an effort on behalf of principal. 4-Outcome and payoffs Payoff to principal: B(x(e)-w) Payoff to agent: w: wage to agent e: effort exerted v(e): cost of exerting effort e x(e): result obtained when exerting effort e Crucial: Conflict of interest Equilibrium concept: Sub-game Perfect equilibrium I.2 Types of Asymmetric Information Problems 1-Moral Hazard Agent has private information after the contract has been accepted. u(w) v(e)

P designs Contact

A accepts (or rejects)

A supplies non verifiable result effort

N determines Outcome & Payoff

The effort is not verifiable, the outcome partially informs the principal.

Alternative: after agent accepts the contract, Nature moves and its move is observable only to the agent.

2-Adverse selection Agent has private information before the contract is offered. E.g. insurance company facing new client.

N chooses A's type Observed by A only

P designs contract

A accepts A supplies N chooses Outcome (or rejects) effort result & payoffs

Effort is verifiable but not informative regarding type. 3-Signalling Similar to adverse selection except that Agent can send a signal that is observed by Principal to potentially reveal his type.

N chooses A's type observed By A only

A sends signal

P designs contract

A accepts A supplies (or rejects) effort

N determines result

Outcome & payoff

Alternative: P could have a private type that she signals via the contract she offers.

Procedure: Characterize first best contracts (contracts under symmetric information) and look at what happens when information becomes asymmetric.

The base model


Bilateral relationship between P and A. Let xi = result i, i = 1,,n

Pr ob[ x = xi / e] = pi (e), for i = 1,..., n.

p (e) = 1
pi (e) > 0 for all i, e.
2.1 Symmetric Information Contracts All relevant information is verifiable. Contract includes effort demanded (e) and wage w(xi)i=1,,n. Principal must only make sure Agent will accept the contract. For all e, P finds acceptable contract and picks the one maximizing his payoff:
i =1 i

i =n

e , w ( xi ) i =1 ,..., n

Max i pi (e) B ( xi w( xi ))

s.t. i pi (e)u ( w( xi )) v(e) U .


Participation constraint is necessarily binding. Lagrangian implies

B ' ( xi wi ) = for all i. u ' ( wi )

Marginal rates of substitutions are equal => efficiency! If P is risk neutral, B(.) is constant, A is fully insured. If A is risk averse, u(.) is constant, P is fully insured. Optimal payment (w) for a given effort level is such that (1) Participation constraint binds (2) Risk is allocated optimally Whoever is risk neutral takes all the risk. If both risk averse, they share the risk.

From F.O.C. w.r.t. wi :

B ' ( x i w i ( x i )) + u ' ( w i ( )) = 0
Differentiating w.r.t. xi :

dw dw B ' ' ( xi wi ) 1 i + u ' ' ( wi ) i = 0 dxi dxi


At the solution Thus:

B ' ( xi wi ) . u ' ( wi ) dwi r = P , dxi rP + rA

where

rP =

B' ' B'

and

rA =

u' ' u'

are the Arrow-Pratt measures of risk aversion.

In a discrete context, (1) +(2) lead to a system of n equations, n unknowns (n = number of possible results). Optimal effort level Discrete case: for each optimal payment, calculate the profit of the principal. The optimal e is the one maximizing these profits.

A Simple Example
Consider the case of discrete effort. el em eh x=10000 x=5000

Game: t=0: principal offers contract (e,w(e,x)) t=1: agent accepts or rejects offer U = 0 t=2: if accepted, contract is executed, result obtained and payments effectuated. Assume that payoffs are given by B ( x w) for the principal and U = u ( w) v ( e) for the agent with

v ( el ) = 20

v ( em ) = 30 v ( eh ) = 40
This is a sequential game with complete information. The principal can choose between 3 effort levels and for each a wage if x=5000 is realized and a wage if x=10000 is realized. For each offer the agent can accept or reject. The way to solve is to find, for each possible effort, the least expensive accepted contract. Case 1: B ( x w) = ( x w), U = Case 2: B ( x w) = Case 3: B ( x w) =

w v ( e)

( x w) , U = w v ( e ) ( x w) , U = w v ( e )

You might also like