Transcript Delegation in politics

Delegation in politics
•The principal-agent literature is concerned with how one
individual, the principal, can design a compensation system (a
contract) which motivates another individual, his agent, to act
in the principal’s interests (Stiglitz).
•Def. Delegation (Lupia): an act where one person or group,
called a principal, relies on another person or group, called an
agent, to act on principal’s behalf
•Def. Agency relationship (Arrow): There are two individuals.
One (the agent) must choose an action from a number of
alternative possibilities. The action affects the welfare of both
the agent and another person, the principal. The principal has
the additional function of prescribing payoffs rules: before the
agent chooses the action, the principal determines a rule that
specify the fee to be paid to the agent as function of the
principal’s observation of the results of the action.
Delegation in politics
•Def. accountability (Lupia): an agent is accountable to a
principal if a principal can exercise control over the agent:
more control , greater accountability.
•Def. Agency loss: difference between the actual consequence
of delegation and what the consequence would have been had
the agent been “perfect. Agency loss is zero when the agent
takes actions that the principal would have taken given
unlimited information and resources.(agent with the same
preferences of the principal but with all necessary resources
and skills)
•Def. Agency (delegation) success: if the outcome of
delegation improves the principal’s welfare relative to what
would have happened if the principal had chosen not to
delegate. (status quo as reverse point)
Delegation in politics
•Every delegation involves at least a principal
and an agent
•Many delegations entail the possibility of
conflicting interests. (principal and agent do not
share the same preferences)
•Many delegations contain the possibility of
limited information. (principal and (more rarely)
agent can be ignorant)
Romer and Rosenthal Delegation model
1
A
Propose
X[0,1]
0
Accept X
X
Reject X
SQ
P
• The agent moves first by choosing whether or
not to act. In the game the principal has
already chosen to delegate.
•The principal reacts by accepting the action or
rejecting it in favour of the status quo.
= outcome
1
SQ
P=A
2
SQ
SQ
A
P
P
A
3
SQ
P
A
SQ
A
4
P
Equilibrium
Outcome
Agency loss
2 P-SQ
P
SQ
P-SQ
Agency loss=0
SQ
P
2 P-SQ
Agent position
Equilibrium
Outcome
Agency loss
2 P-SQ
P
SQ
Agency loss=0
P-SQ
SQ
P 2 P-SQ
Agent position
Comments to R.R. model
•
Agency loss can grow when the agent’s ideal
point policy or the status quo moves away from
the principal’s ideal policy.
• Agency loss can grow with the distance
between P and SQ because a “bad” status quo
for the Principal makes a wider range of
proposals attractive to the principal.
• Predicitions depend on two assumptions
1. Delegation is particularly large: principal can
only accept or reject the agent’s action
2. Principal and agent have complete information
Incomplete information
•
•
1.
2.
•
1.
2.
Information is incomplete when an actor is unable to
predict the consequence of actions with certainty.
Two kind of informations
Information about actions
Information about people and their context.
When principals know less than agents two kind of
problems
Moral Hazard with hidden actions (“Opportunity makes
the thief” )
Adverse selection (“Put the fox to guard the poultry”)
How Moral Hazard Affects
Delegation (RR model)
Ex.
Complete information
Incomplete
information best
case
Outcome
Agency
loss
Outcome

Loss
Outco  Loss
me
1
P
None
P
0
SQ
|P-SQ|
2
A
-|A-P|
A
0
SQ
|A-SQ|
3
ε Closer to P
than SQ
-|SQ-P|- ε SQ
ε
A
|A-(|SQ-P|+ ε)|
4
SQ
-|SQ-P|
0
A
|A-SQ|
SQ
Incomplete information
worst case
= outcome witf complete information
= outcome with incomplete
information, worst scenario
SQ
SQ
P=A
A
SQ
P
P
SQ
P
A
SQ
A
= additional agency loss in case of i.c.
= agency loss in case of c. c.
A
P
Adverse selection
•
•
The agent is unwilling: the agent has
preferences different from the
preferences the principal thought he/she
had
The agent is unable.
How Institutions affect agency loss
•
Institutional designs affect the provision
of information to political principals
1. Ex ante mechanisms to deal with
adverse selection problems
2. Ex post mechanisms to deal with moral
hazard problems
Ex ante mechanisms: Screening
•
1.
2.
3.
4.
•
Competition: when a principal forces potential agents
to compete for the right to be his/her agent, he/she can
induce the potential agents to reveal information about
themselves. However the conditions that make such a
competition reducing agency loss are:
It is easy to determine whether or not the agent can
perform the task
Potential agents know their preferences and abilities
Potential agents know if the other potential agents
know their preferences and abilities
It is relatively easy and costless for the principal to
punish the agent for failing to accomplish the task ex
post.
Sometimes the skills required to win a competition are
different from the skills required to be an effective
agent
Ex ante mechanisms: Selection
•
•
When an agent chooses an action that
reveals to the principal a previously
hidden personal characteristic.
Ex. Diploma:
Ex ante mechanisms: Contract Design
•
•
Principal can offer potential agents a menu of
contracts to induce agents to reveal things
about themselves.
Problem of risk sharing: when structuring an
agent’s incentives ex ante , a principal faces a
trade-off between risk-sharing – which
prevents the agent’s rewards from depending
too strongly on observable outcomes- and
incentives which induce the principal to
condition the agent’s rewards on observable
outcomes.
Ex post mechanisms
•
•
•
•
Direct monitoring (“patrolling”) from the
Principal
Self reporting
Third party testimony (fire alarms
oversight)
Institutional checks : instead of striving
for an unbiased source of information a
principal may do better obtaining biased
reports from different agents who have
conflicting interests.