Network Heterogeneity and Value of Information
Kota Murayama
TL;DR
This paper shows that the value of public information in networked settings is inherently non-monotonic and highly shaped by network structure. Using a linear-quadratic-Gaussian beauty-contest framework, it demonstrates that public disclosure can improve welfare when interactions are uniform but hurt welfare in core-periphery networks where dense cores overreact to noisy public signals, imposing volatility on peripheral agents. It further identifies an endogenous-information-sharing channel that can lead to under-provision of public information in such networks. The work highlights practical policy implications: transparency policies may backfire in tiered markets, and obtaining granular network data is crucial for welfare-enhancing information design.
Abstract
Does greater connectivity enhance the value of public information? I study a networked beauty contest game where agents balance adaptation to the fundamental with local coordination. The analysis reveals a stark non-monotonicity: while public disclosure improves welfare when interactions are uniform, regardless of their intensity, it can be detrimental in core-periphery structures. This welfare loss stems from a distortion driven by the core, where core agents over-respond to a noisy public signal, forcing peripheral neighbors to absorb this volatility to maintain alignment. These findings suggest that standard transparency policies can backfire in tiered markets where dominant hubs propagate excess volatility.