Information Design for Adaptive Organizations

TL;DR

This paper explores how organizations can optimally share information to enhance coordination and performance, using graph theory and Bayesian persuasion to determine effective public signals and transparency levels.

Contribution

It introduces a novel framework combining Laplacian matrices and Bayesian persuasion to design optimal information sharing strategies in organizations.

Findings

Optimal public signals include local states and their averages.

Laplacian eigenvalues determine transparency and revelation conditions.

Algebraic connectivity influences full information disclosure.

Abstract

This paper examines the optimal design of information sharing in organizations. Organizational performance depends on agents adapting to uncertain external environments while coordinating their actions, where coordination incentives and synergies are modeled as graphs (networks). The equilibrium strategies and the principal's objective function are summarized using Laplacian matrices of these graphs. I formulate a Bayesian persuasion problem to determine the optimal public signal and show that it comprises a set of statistics on local states, necessarily including their average, which serves as the organizational goal. When the principal benefits equally from the coordination of any two agents, the choice of disclosed statistics is based on the Laplacian eigenvectors and eigenvalues of the incentive graph. The algebraic connectivity (the second smallest Laplacian eigenvalue) determines the condition for full revelation, while the Laplacian spectral radius (the largest Laplacian eigenvalue) establishes the condition for minimum transparency, where only the average state is disclosed.