Академический Документы
Профессиональный Документы
Культура Документы
Introduction
It does not work when people are more likely to be wrong than right (conventional wisdom is incorrect)
Surveys are good in this regard.
The problem is that there is no penalty for a wrong answer, so there is no sorting in a survey. You can do potentially better.
Prediction Markets
Prices play the role of information in markets. You can get better results when people are sorted based on their own judgment of the value of their information
The Iowa experiment Google
These are modeled after market economies and the invisible hand.
Overview
2. Benefits of Centralization
Economies of scale
physical capital managerial talent brand name & reputation design
Better coordination
knowledge transfer across units consistency / standardization synchronization control common strategy
10
Benefits of Decentralization
11
Specific Knowledge
Cost of Transferring Knowledge Costly Specific Knowledge Cheap General Knowledge
costly to understand
requiring scientific or specialized technical skills subjective or experiential
12
The classical approach: centralized hierarchy Decentralization implies that coordination must happen at lower levels of the organization
roughly speaking, 2 kinds of coordination problems: simple & integration
13
Modularization
Putting people with the most interdependent jobs together amounts to modularizing overall structure
ex: break XP Consulting into smaller divisions
regional? (NA; Europe; Asia) type of customer? (Corporate; Government; Not-for-profit; etc.) practice area? (Strategy; IT; 6s)
You can structure authority w/ some decisions organized by one set of divisions, others a different set
ex: decisions about hiring & compensation determined by region; decisions about training & promotion determined by practice area
Note how complex it starts to get there are clear advantages to reducing overlapping lines of authority
hypothesis: the largest source of dis-economies of scale is bureaucracy (coordination costs)
14
Integration
Integration: for some decisions, pockets of specific knowledge throughout org. need to be combined
ex: Apple Computer laptop product design use lateral mechanisms
teams, matrix informal networks e.g., product design
CEO Engineering Sales Production Other
B A
A B B A
15
16
Hierarchy
Gladys Gladys Willie
Flat
Willie
18
Assume new ideas are binary (good or bad / profitable or unprofitable) At first stage, p = probability of correct decision; p > At second stage (hierarchy only), q = probability of correct decision; q > p
19
Hierarchy
Accepts p Good Rejects (1-p) Accepts pq
Rejects p(1-q)
Rejects (1-p)
New Idea
Accepts (1-p)
Bad Rejects p
Accepts
(1-p)(1-q)
Rejects (1-p)q
Rejects p
Willie Evaluates
20
Flat
Accepts p
Good Rejects (1-p)
New Idea
Accepts (1-p) Bad Rejects p
21
Results
Flat Rate For One New Idea Accept Good Idea False Negative False Positive Reject Bad Idea Overall Throughput Accept Good Ideas False Negatives False Positives Reject Bad Ideas p 1-p 1-p p Hierarchy p q 1-p q (1-p)(1-q) 1-(1-p)(1-q)
2N p 2N(1-p) 2N(1-p) 2N p
22
Flat structures
evaluate ideas more quickly evaluate more ideas for the same # of employees make more changes, good & bad have more successes & failures
Constraints on decisions
e.g., budgets
5. Implementation
So what should XP Consulting consider in its structure? First, Modularize overall structure, possibly in overlapping ways
ex: Cambridge Technology Partners makes the problem more manageable put most interdependent parts together, reducing coordination problems
Second, allocate decisions within each division: ask who / what / where / when / why? to identify key specific knowledge
who has valuable specific knowledge? what kind of knowledge? where in (& out) of the organization? when (is timing relevant)? why is it of economic value?
Third, think about what needs to be made consistent or coordinated across the division or whole organization
26
Implementation
The last two give strong guidance on what to decentralize & centralize Fourth, go back & refine the overall structure
try to streamline further to cut bureaucracy look for & address coordination problems integration problems require the most attention, & will create most of your day-to-day headaches
Sixth, design performance evaluation & incentives to match job design (after Midterm)
27
6. Economic Ideas
Decision making
decentralization v. centralization decision management v. control degrees of decision control / hierarchy, & their effects
28
Summary Points
The metaphor of a market highlights the role of economics in organizational design
design is largely about creating & making use of knowledge
by its nature, specific knowledge tends to have more economic value
but market approaches are limited when complex coordination (especially of the integration kind) is needed
The concepts apply to design of an individual job, to a workgroup, to structure of a global conglomerate
29