Double Coordination in Small Groups

1
Double Coordination in Small Groups

• Luigi Mittone, Matteo Ploner, Ivan Soraperra
• Computable and Experimental Economics Laboratory
  University of Trento, Italy
• IAREP/SABE - World Meeting 2008
• Roma 4 September 2008

2
The Experimental Approach to Innovation a New
Research Field for the CEEL

• Two work in progress projects
• A nation wide field study on coordination in
  non-profit firms knowledge sharing and
  dissemination within organizations (U-Know
  Project financed by the EU)
• Coordination and innovation within the network
  externalities frame

3
Motivations and Related Literature (1)

• Coordination has been studied exclusively in
  single groups (BoS, WLG, Minimum Game, etc.).
• Many interesting situations in which two groups
  of people must coordinate their actions on two
  levels
• two groups of stakeholders in a firm,
• two departments of the same firm,
• consumers and producers of goods with network
  externalities, etc.

4
Motivations and Related Literature (2)

• A crossroad between two streams of literature
  coordination failures and network externalities
• Network externalities considered within the
  problem of introducing a new product
• Coordination failures in large groups the weak
  link coordination game

5
Technology Adoption and Network Externalities

• Katz and Shapiro (AmEcRew 1985, JourPolEc 1986)
• Liebowitz and Margolis (JourEcPersp 1994)

6
Katz and Shapiro (1986) p.822-823
7
Katz and Shapiro (1986)
8
Network Externalities

• Network consumption externalities require that at
  least one specific attribute, using the
  Lancasters Theory of Consumption terminology, is
  almost perfectly homogeneous
• Supply side competition is therefore restricted
  to the other attributes (first of all price)
• The common attribute works as an entry barrier
  for the newcomers.

9
Network Externalities

• Network externalities as a public good
• Need for coordination to produce a Pareto
  efficient solution.
• The specific case when the consumers must
  coordinate themselves to switch from a
  traditional product (already characterized by
  network externalities) to an innovative one
  (which we assume can produce even stronger
  positive externalities due to the use of a more
  innovative technology)

10
Coordination in Experimental Games

• WEBER (AmEcRew 2006)
• BORNSTEIN et al (GamesEcBehav 2002)
• COOPER et al (AmEcRew 1990)

11
COOPER et al (1990) p. 218

• We study a class of symmetric, simultaneous
  move, complete information games called
  coordination games. This term refers to games
  which exhibit multiple Nash equilibria which are
  Pareto-rankable.' That is, all players are better
  off in one equilibrium relative to another
• yet may be unable to explicitly coordi-

• nate their strategies to achieve the preferred
  outcome. When this occurs, a coordination failure
  arises.

12
WEBER (2006)BORNSTEIN et al (2002)

• Weak link coordination game
• Coordination failure in large groups
  (experimental)
• Competition between groups
• Progressive increase in the size of the group

13
Weak Link Coordination Games
Players choice Minimum choice of all players Minimum choice of all players Minimum choice of all players Minimum choice of all players Minimum choice of all players Minimum choice of all players Minimum choice of all players
Players choice 7 6 5 4 3 2 1
7 .90 .70 .50 .30 .10 -.10 -.30
6 .80 .60 .40 .20 .00 -.20
5 .70 .50 .30 .10 -.10
4 .60 .40 .20 .00
3 .50 .30 .10
2 .40 .20
1 .30
Source Weber, 2006
14
Two Groups, Two Goods, Double Coordination

• One group are the consumers
• One group are the producers
• Positive network externalities for both groups
• Multiple Nash equilibria
• Not a weak link game
• One innovative good
• One traditional good

15
Interaction Structure

• Coordination Game
• 2 Pareto-ranked equilibria
• Two actions
• T(raditional good)
• I(nnovative good)

16
Interaction Structure

• Groups of 10
• 5 players role A
• 5 players role B
• 30 repetitions with feedback
• 1 repetition rewarded (random pick)
• 2 experimental treatments
• Baseline (two different payoffs structures)
• Treatment (two different payoffs structures)

17
Baseline(1)

• Symmetric game
• 2 Nash equilibria
• All players choose I All players choose T
• Dominant strategy is to choose what the majority
  of the members of the other group chooses
  (independently from ones own group)

18
Treatment (1)

• Asymmetric game
• 2 Nash equilibria
• All players choose I All players choose T
• Dominant strategy for role A is to choose what
  the majority of role B chooses
• Dominant strategy for role B is to choose what
  the majority of role B chooses

19
Baseline(2)

• Same properties of Baseline 1 but
• Focal point in the NW corner
• Weaker risk perception for the I move

20
Treatment (2)
21
Summary

• Baseline
• Players A (e.g., consumers) and Players B (e.g.,
  producers) must build a belief on the preferred
  choice of the other group
• Treatment
• Players B (e.g., producers) have a greater power
  in determining the equilibrium
• In setting (2) option I is less risky than in (1)
  setting

22
Predictions

• Baseline
• Due to the simmetry of the incentives across
  groups and to the balanced payoff structure of
  the two matrices we expect a fast convergence
  towards one of the two equilibria.
• Treatment
• Players B can pull towards one of the two goods
• If no coordination at the beginning then the
  choice of majority of Bs will attract the other
  players in the game
• In (2) more global coordination on I,I than in
  (1)

23
Procedures and Participants

• 60 Participants
• Students of the University of Trento, Italy
• Computerized experiment
• Web-based
• Average earnings
• Time required
• About 1h 30min

24
Results Payoffs1

• In the baseline a fast coordination on I is
  observed
• In the treatment coordination on I is slower (or
  it does not even occur !)

25
Results Payoffs 2

• When the risk of choosing I is low all the
  people immediately coordinates on the Pareto
  Dominant equilibrium

26
Preliminary Conclusions

• A very high rate of coordination compared to the
  coordination levels reported in the literature
• Slower coordination in Treatment(1) when compared
  to Baseline(1)
• Need for a more detailed individual level
  analysis