Vertical Governance of Online Speech: Evidence from Google's Moderation Mandate

This paper provides the first causal evidence that upstream infrastructure providers can reshape social media discourse by enforcing content moderation through access-based leverage. I exploit a 2022 update to Google's Play Store policy requiring stricter removal of violent threats and misinformation, along with variation in platform exposure across three similar 'Alt-Tech' social media platforms, within a triple-differences design. Using a novel panel of over 28 million posts from 62,000 users, I find that threatening content declined sharply and persistently on the exposed platforms, particularly among high-risk users. These effects are not explained by user self-censorship, public awareness, contemporaneous events, or selective data loss. I also document significant reductions in lawful but politically sensitive narratives, including election denial and January 6 insurrection commentary. The findings show how infrastructure-level enforcement can durably alter the boundaries of permissible speech across platforms, contributing to literatures on platform governance, vertical restraints in digital markets, and the institutional foundations of online discourse.