Centralization vs Decentralization Flashcards
Fundamental Tension
- The relevant information for decision-making isdispersed and costly to obtain (local taste)
- Ideally, the agents with the relevant information should take the decisions.
- However, given the diverging interests and incompleteness of contracts, there may be incentive problems.
Decentralization
Use the local information without the need to transfer it
(efficiency and time savings)
- The top of the hierarchy can focus more on strategic
decision making
- Motivation to subjects taking local decisions
- Attracts talent in the periphery.
- Increases workers’ engagement.
- Increase human capital of employees
Decentralization pt 2
-Coordination:
- Duplication of Costs (R&D, Marketing, etc)
- Price Competition across branches.
- Local vs Global optimum.
- Losses in the information from the center.
- Agency problems: Adverse selection, moral hazard,
incentives.
Information and Agency Costs
- Information Costs:
- Costs of producing and transmitting the necessary/relevant information for decision making, or Costs of taking decisions with insufficient information.
- Control Costs:
- Costs of Interest alignment
Fixed decision process
Milgrom and Roberts (1988): people have incentives to distort information collection and to influence decisions to their advantage through transmission but face costs of doing it.
A basic model with two players
see slides
Cheap talk: Crawford and Sobel (1982)
When messages are costless, misalignment of preferences prevents full revelation of information - Sender receiver game:
Even when the manager says the truth the CEO will think that he lies because of diverging interests. Because the CEO does not believe the manager - the manager has the incentive to lie.
- The receiver (CEO) wants to gear her decision to the local information that only a biased sender (manager) has.
- Parties have quadratic payoffs and the sender knows the state of nature s.
- The bias of the manager represents the imperfect alignment of the parties’ interests. And the information he gives: which range belongs the state of nature.
- The result shows that if the bias is sufficiently small, the information that gives the sender about the state of nature is more precise (narrower intervals) and so the receiver’s decision.
Signal Jamming: Gibbons (2005)
Resources are spent on lobbying even though, in equilibrium, lobbying does not affect the decision. Consider an employee-owner game:
- There is symmetric uncertainty about the state s
- The employee (player 1) can influence the signal that the owner (player 2)
observes about the state
- The owner will try to extract from the signal whatever information about the
state, prompting the employee to try to influence the signal
- In equilibrium, the owner anticipates the employee’s attempts to influence the signal and make her decision weighing this
- Even though, the owner is not fooled, the equilibrium level of lobbying is positive because the employee has incentives.
- Moreover, this level increases with the employee’s bias
Signalling: Austen-Smith and Banks (2000) and Kartik (2007)
Lesson: Some actions may convey information because only
certain types would be willing to take those actions
Funny ads - your so good you dont need proper ads
- Signalling model of money burning adapted:
-> Player1’s action is burning money. The cost
of money burning increases with the state
-> The sender’s action could not perfectly
reveal the state
-> The result is a separating equilibrium.
Endogenous decision allocation
Now, consider actions that the owner can take in the first stage to mitigate the inefficiencies of organizational communication and decision making in the second stage
Endogenous decision allocation:T hree suggestions aimed at controlling influence activities:
- Limiting discretion
- Cut communication
- Altering several aspects of organizational design to limit the cost of influence activities
1.Limiting discretion
Manager and employee
-> The employee can allocate a fixed amount of effort between
productive activities and influence
-> If the manager is allowed discretion over adopting the change,
then the probability that she does adopt it is a function
increasing on influence and the employee devote time to
production and to influence
Manager does not have discretion to promote employees
2.Cut communication
Two employees - does not matter if good or mad are promoted & we care a lot about effort
- > Each one allocate a fixed amount of time between influence activity and productive effort and the principal can not observe these choices
- > There is a promotion and the principal has to decide on wage payments for the current period, wages to pay to each employee once the promotion decision is made, and the criteria to use in promotion
Results: if the importance of assigning the better person to the
new job is sufficiently low and the productivity of effort is high,
promotion will be random
2.Cut communication pt 2
If the manager does not have discretion, the employee
has no incentive to exert influence and devote all his time
to production.
-> Result : when redistribution effects are greater or when
the change is less valuable is optimal to deny the
decisionmaker discretion over implementing the change.
- Altering several aspects of organizational design
o limit the cost of influence activities
- If the marginal impact of employee quality on performance in the
new job is sufficiently high, qualifications in the promotion
choice are more important. - This would incentivize workers to concentrate solely on building
credentials - Some adjustments :
-> Pay a bonus for individual output
-> Promote partially on current performance