Data Governance with Vulnerable Individuals

Department of Economics and Decision Sciences

Speaker:  Rosella Argenziano (Essex)

Room T-019

Abstract :

Public service providers can improve service quality by collaborating with third parties and sharing sensitive data, but doing so exposes individuals to exploitation, with harm especially severe for highly vulnerable individuals. We model this trade-off in a game where a benevolent provider interacts with an exploitative third party, considering both individual-level and dataset-level data sharing. Even under full benevolence, sensitive information may be revealed explicitly or implicitly due to a commitment problem. Requiring individual consent for data sharing does not resolve this problem and instead creates new inference channels. If individuals have the option to withhold information from the provider, coordination failures arise: vulnerable individuals may strategically withhold—receiving lower-quality support as a result—or avoid public services altogether, while still facing exploitation.

Joint work with : Francesco Squintani (Warwick)