Professional Documents
Culture Documents
Laffont, J.-J. and D. Martimort [2002] The Theory of Incentives, Princeton Univ. Press Salani B. [1997], The Economics of Contracts, A Primer, MIT, Cambridge MA: ch. 2 (2.1, 22.), pp. 11-26
Why principal-agent?
One party (the principal) contracts another party (the agent) to perform some action or to take some decision. The agent has an information advantage he knows something the principal does not know, will know something the principal does not know, he can take secret actions. The principal knows that the agent has this advantage (making it a potential handicap!).
Examples
owner manager insurance company insured creditor debtor firm salesmen voters government investor portfolio manager
Hidden Information
t=1 P offers a contract t=2
Adverse Selection
t=3 time The contract is executed
A accepts or refuses
Story 2
t=0 P offers a contract
Hidden Action
t=1 A accepts or refuses t=2
Moral Hazard
t=3 time A exerts an effort or not The outcome is realized and the contract is executed
Adverse selection
You consider buying a second hand vehicle. You know that there are different qualities the cars' owners know their car's quality you cannot distinguish cars Owners do not sell for less than the true value You do not bid more than the expected value What will happen?
5000
10000
Price offered
V = 7500 5000
E[V | V p]
5000
10000
Price offered
5000
10000
Price offered
45
10000 E[aVs | Vs p]
5000
5000
7500 10000
Price offered
45
5000
7500 10000
Price offered
45
10000 1.2 6'250 = 7500 5000 Quantity=50% E[aVs | Vs p] Partial market failure Efficiency loss best cars not optimally used (worst cars subsidized)
5000
7500 10000
Price offered
45
10000 1.2 6'250 = 7500 5000 Quantity=50% E[aVs | Vs p] Partial market failure Efficiency loss best cars not optimally used (worst cars subsidized) Higher a smaller range of qualities unsold higher efficiency loss for best cars
5000
7500 10000
Price offered
Preliminary conclusion
Informational asymmetries create inefficiencies. Quantities traded are smaller than under first best; market may fail completely. Less informed party knows that they know less and anticipate opportunistic behavior by better informed party. The information advantage is a handicap for the owners of the best qualities. is an advantage for the owners of the worst qualities (they can hide behind the intermediate qualities). The presence of bad cars creates a negative externality to the owners of good cars.
Optimal contracts
an attempt to overcome informational asymmetries one party proposes contract (mostly principal) the counterparty accepts or rejects contract creates incentives: good types accept, bad types reject (hidden info) exert effort (hidden action) mechanics: offering party maximizes utility subject to constraints: participation incentives wealth, feasibility
Rain in Bilbao
For several months not a drop of rain has fallen in Bilbao. The desperate mayor is contacted by a sourcerer who says he is able to make rain. If the sourcerer is a bluffer the chance of rain for the next week stays at 2/100. If he really is a sourcerer, the chance of rain goes up to 20/100. The pretending sourcerer has a utility function of u(w) = w0.5. If he is a sourcerer, he accepts a contract if he gets at least u=10. If he is a bluffer, he accepts if he gets u>1. The mayor does not want to be caught with a bluffer. a) What contract should the mayor offer? b) What is the cost of the information asymmetry to the city of Bilbao?
Source: I. Macho-Stadler und J.D. Perez-Castrillo, An Introduction to the Economics of Information, Oxford University Press, 1977, p. 161
good type
bad type
c EU(c)
reject EU(rej)
c EU(c)
reject EU(rej)
good type
bad type
c EU(c)
reject EU(rej)
c EU(c)
reject EU(rej)
Participation Constraints
contract space
Contracts can define payments that are contingent on observable (and verifiable) outcomes. Type of agent (S, B): not observable (not even ex post) Outcome (R, 0): observable Contract can specify payments for R and 0. This is the contract space. The principal should use her contract space. The general form of the contract thus is c = {wR, w0} The optimal contract c* = {w*R, w*0} maximizes the principal's utility subject to constraints (S accepts, B rejects)
Participation constraints
Sourcerer: Bluffer: 0.20(wR 0.5) + 0.80(w0 0.5) 10= uS 0.02(wR 0.5) + 0.98(w0 0.5) 1 = uB
both constraints bind, because if not - for sourcerer: we could have him cheaper
- for bluffer: sourcerer bears unnecessary risk (we could make outcomes more similar, thus paying sourcerer less in expected terms)
Graphical solution
w0
45
uB uS 0 uB 2500 uS
wR
Numerical solution
Optimal contract when types are unobservable: w*R = 2500, w*0 = 0 Optimal contract when types are observable (first best): wf = wR = w0 = 100 (only offered to sourcerer) The cost of the information asymmetry? Cost = Ew* Ewf = 0.2(2500) -100 = 400 is here paid by the principal.
Which of the four optimal contracts maximizes the principal's utility? => the overall optimal contract
Graphical solution
w0 OF
45
100 uB 0 3600 uS
wR
Numerical solution
Optimal contract when types are unobservable: w*R = 3600, w*0 = 100 Interpretation:
performance wage 100 = fixed wage; 3500 = bonus for rain 3600 = fixed wage; 3500 = penalty if no rain
Examples
Guarantee Health Insurance: Franchise Mortgage Credit Baby 81 King Salomo's Judgement
Hidden Information
t=1 P offers a contract t=2
Adverse Selection
t=3 time The contract is executed
A accepts or refuses
Story 2
t=0 P offers a contract
Hidden Action
t=1 A accepts or refuses t=2
Moral Hazard
t=3 time A exerts an effort or not The outcome is realized and the contract is executed
accept
reject
no effort
<
EU(reject) EU(reject)
You cannot observe the broker's effort, but you observe the outcome (the price at which you can sell). You are risk neutral (i.e. you maximize expected profit) The broker maximizes U = w1/2 - e, where w is financial income and e effort cost. He only becomes active if he gets U=120 What contract do you offer the broker? What are the effects of unobservability of effort?
Hints:
What are the best contracts for each possible effort level when effort is not observable? (Draw the constraints in a graph with w1/2-axes!!!) Which among these best contracts yields the highest profit overall? What effort level would P want to get if effort was observable? What contract would he offer? What difference does unobservability make on: the optimal effort level? A' welfare? P's welfare?
Hints:
An important idea in the economics of contracts (or: mechanisms) is the Revelation Principle The Revelation Principle says (roughly): whatever outcome you can achieve, you can achieve by giving the counterparty an incentive to tell the truth. you do not loose anything by making the counterparty tell the truth. Example: the second price auction (also Vickrey auction): the highest bidder gets the good for the amount of the second highest bid