Abstract:
This paper suggests that even if it is costless to inform all team members about the quality of a project, there are reasons to concentrate information in the hands of one person (a leader) and prevent full revelation to the rest. This deprives others of the information necessary for profitable defections; they (the followers) therefore will have no reasonable strategy other than following the informed leader because he has more information than they themselves have. Such leaders then can lead the ignorant group into cooperation in cases where information gives them an incentive not to do so. Unlike the common belief, this paper shows that lack of information transparency in a group or an organization may increase cooperation and thus efficiency compare to a regime of information dispersal.
Machine summary:
"In a theoretical setting, Komai, Stegeman, and Hermalin (2007, henceforth KSH), study a team production problem in which the leader leads by example but the leader's information is only partially revealed to his ignorant followers.
In the second stage of the game, having observed the leader's endorsement decision, followers update their beliefs about and simultaneously decide how much to contribute to the public project.
One is a trivial equilibrium in which the leader endorses the project with zero probability, and followers never contribute (we will argue- in footnote 14- that this equilibrium is not reasonable).
Remark 2: Because t*<, Theorem 2 implies that the unique endorsement equilibrium of the leader-follower game produces positive contributions for strictly more values of than does the complete information equilibrium.
The endorsement equilibrium of the leader-follower game yields positive contributions when t*<<, while the complete information equilibrium does not.
According to Theorem 3, if the cost of a bad endorsement is large enough (if θ is sufficiently large), then expected contributions will be higher ex-ante in the leader-follower setting than under complete information.
Theorem 3 simply shows that, if preventing full revelation of information does not improve the followers' cooperation ex-post in some states, it improves them ex-ante (on average) if the leader's cost of a bad endorsement is sufficiently large.
2 We show that if preventing full revelation of information does not improve the followers' cooperation ex post in some states, it improves them ex ante (on average) if the leader's cost of a bad endorsement is sufficiently large."