Decentralized Signaling Mechanisms
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866910915069214720 |
|---|---|
| author | Boroujeni, Niloufar Mirzavand Iyer, Krishnamurthy Cooper, William L. |
| author_facet | Boroujeni, Niloufar Mirzavand Iyer, Krishnamurthy Cooper, William L. |
| contents | We study a system composed of multiple distinct service locations that aims to convince customers to join the system by providing information to customers. We cast the system's information design problem in the framework of Bayesian persuasion and describe centralized and decentralized signaling. We provide efficient methods for computing the system's optimal centralized and decentralized signaling mechanisms and derive a performance guarantee for decentralized signaling when the locations' states are independent. The guarantee states that the probability that a customer joins under optimal decentralized signaling is bounded below by the product of a strictly positive constant and the probability that a customer joins under optimal centralized signaling. The constant depends only on the number of service locations. We provide an example that shows that the constant cannot be improved. We consider an extension to more-general objectives for the system and establish that the same guarantee continues to hold. We also extend our analysis to systems where the locations' states are correlated, and again derive a performance guarantee for decentralized signaling in that setting. For the correlated setting, we prove that the guarantee's asymptotic dependence upon the number of locations cannot be substantially improved. A comparison of our guarantees for independent locations and for correlated locations reveals the influence of dependence on the performance of decentralized signaling. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_14163 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Decentralized Signaling Mechanisms Boroujeni, Niloufar Mirzavand Iyer, Krishnamurthy Cooper, William L. Computer Science and Game Theory Multiagent Systems Theoretical Economics 91A28, 90B99, 68W25 We study a system composed of multiple distinct service locations that aims to convince customers to join the system by providing information to customers. We cast the system's information design problem in the framework of Bayesian persuasion and describe centralized and decentralized signaling. We provide efficient methods for computing the system's optimal centralized and decentralized signaling mechanisms and derive a performance guarantee for decentralized signaling when the locations' states are independent. The guarantee states that the probability that a customer joins under optimal decentralized signaling is bounded below by the product of a strictly positive constant and the probability that a customer joins under optimal centralized signaling. The constant depends only on the number of service locations. We provide an example that shows that the constant cannot be improved. We consider an extension to more-general objectives for the system and establish that the same guarantee continues to hold. We also extend our analysis to systems where the locations' states are correlated, and again derive a performance guarantee for decentralized signaling in that setting. For the correlated setting, we prove that the guarantee's asymptotic dependence upon the number of locations cannot be substantially improved. A comparison of our guarantees for independent locations and for correlated locations reveals the influence of dependence on the performance of decentralized signaling. |
| title | Decentralized Signaling Mechanisms |
| topic | Computer Science and Game Theory Multiagent Systems Theoretical Economics 91A28, 90B99, 68W25 |
| url | https://arxiv.org/abs/2504.14163 |