Abstract: We study the optimal allocation of a good whose consumption imposes negative externalities on non-consumers. A surplus-maximizing designer can impose access restrictions (e.g., bans or time limits), screen agents by charging for access, and condition the allocation on agents' observable characteristics (e.g., age). The main application is regulating social media platforms when non-users face social exclusion. We show that partial access restrictions are not optimal when non-consumers face worse externalities than consumers. We then provide conditions under which, for each observable group, the optimal allocation either grants access to all agents or bans the good entirely. Our results rationalize minimum age requirements for social media that several countries have adopted or proposed. We next study how non-consumer externalities affect agents' welfare when the designer maximizes revenue. We show that, if consumer externalities are positive, reducing the negative externalities on non-consumers need not increase aggregate welfare. This provides new insights into policies aimed at mitigating externalities on non-consumers.

Abstract: We study the stability of social networks when agents can monitor others' linking decisions. We introduce a novel notion of network stability: a network is Peer-Monitored Stable if any unilateral deletion or mutual addition of a link harms at least one neighbor. We show that, for two widely studied types of externalities — distance and degree externalities — every Pareto-efficient network is Peer-Monitored Stable. This result does not rely on any specific structure for the efficient network and allows for arbitrary heterogeneity across agents. We then generalize our definition of monitoring by allowing agents to monitor indirect connections within a given distance, and introduce a measure of network stability defined as the minimum monitoring level required to stabilize the network. We apply this measure to models that mix both types of externalities and derive an upper bound on the monitoring level required to stabilize Pareto-efficient networks.

Abstract: We provide a general sufficient theorem to solve the delegation problem with and without money burning. For interval allocations, our main theorem relaxes the conditions obtained in previous literature. Significantly, our theorem extends beyond this, successfully addressing cases with discontinuous allocations. We demonstrate the theorem's power through examples, connecting with existing results and uncovering novel optimal allocation structures.

Abstract: This paper develops a theory of network regulation in environments where connections provide benefits but expose one side to risk. Agents form bilateral links in a setting with two types of agents: risky agents, who impose externalities on their partners, and vulnerable agents, who bear exposure. We first characterize efficient networks and show that they take one of four forms: two cliques, only cross-links, largest-group clique with cross-links, or the complete network. We then compare efficiency with pairwise-stable outcomes, showing that decentralized formation can lead to over- and under-connected networks. Finally, we examine policy interventions — fines on linking and degree caps — that improve the efficiency of stable networks. We also analyze the key-player allocation problem in organizations and design choices between deeper hierarchies and flatter structures. Overall, our analysis shows how organizations can design regulatory measures that preserve network value while mitigating exposure risk.