Essays in the Political Economy and Industrial Organisation of Media: Infrastructure, Identity, and Influence

Abstract

This thesis investigates how economic and institutional forces shape political discourse by analysing the behaviour of media intermediaries: social media platforms, legacy newspapers, and religious radio networks, under varying structural constraints. Across three empirical studies, I apply causal inference methods to evaluate how media respond to incentives and how such responses affect symbolic representation, speech norms, and political behaviour.

The first paper examines infrastructure-led content moderation by exploiting a previously undocumented policy change by Google in July 2022, which required apps to restrict threatening and misinformation-related content to remain on the Play Store. The policy asymmetrically constrained Gettr and, to a lesser extent, Truth Social, while Gab remained unaffected. Using a novel panel of over 28 million posts from 62,000 users and a triple-differences (DDD) design, I find sharp, persistent declines in threatening content among constrained platforms, especially among high-risk users. No such decline occurred on Gab. I also find no evidence of harmful content being displaced to Gab or other forms of reactive user migration, suggesting that the policy reduced threatening discourse without triggering the spillovers commonly associated with public deplatforming. The findings suggest that upstream platform governance can significantly reshape online discourse, even in the absence of public enforcement or deplatforming.

The second paper studies how economic incentives shaped Southern newspapers' adoption of the term "Black" instead of "Negro" during the civil rights era. I leverage a coordinated campaign launched by civil rights leaders in June 1966 and track label diffusion using a corpus of 58 million digitised articles from 450 Southern newspapers. Adoption coincided with a positive shift in portrayals of Black actors and was predicted by market competition and white racial conservatism. To isolate the economic channel, I compare adoption rates across co-owned newspapers with shared editorial teams but different local readerships. Finally, I show that early adoption is associated with subsequent increases in Black political representation, suggesting that symbolic norm change facilitated substantive incorporation.

The third paper evaluates the political impact of Christian-conservative broadcasting by studying the expansion of Salem Communications following deregulatory shifts in the 1990s. Salem grew rapidly by acquiring stations and monetising through religious programming insulated from commercial pressures. Using plausibly exogenous variation in signal exposure from topography, I estimate the causal impact of Salem's entry on Republican vote share in presidential elections. I find persistent increases in Republican support, especially in evangelical counties, and evidence of long-run partisan realignment.

Together, these papers demonstrate how structural features of media markets, i.e. ownership constraints, platform governance, audience composition, and business models, mediate the relationship between identity, discourse, and democratic participation. The thesis contributes to political economy, media economics, economic history and the study of symbolic politics in institutional contexts.