Congestion v Content Provision in Net Neutrality: The Case of Amazon's Twitch.tv

Net neutrality encourages content provision but also creates congestion externalities from the increase in data traffic. I study the consequences of net neutrality in Twitch.tv, a popular internet platform. Twitch is non-neutral because it gives priority to the most popular content providers by compressing their data, which makes them accessible to more consumers. I estimate a two-sided-market model that considers the interactions between content provision, its consumption, and congestion. Using an exogenous technological upgrade that increased data traffic, I identify the costs of congestion for content providers and for their consumers and, using exogenous time-series variations within panels, I identify the benefits of prioritization. I use the estimated preferences and technological parameters to study the counterfactual in which net neutrality is imposed in the platform, which requires priority to be allocated anonymously. Consumer welfare drops 5%, whereas content provision does not increase, but its average quality drops. I then consider a counterfactual rent-extractive platform that charges for prioritization under the non-neutral regime. In this case, net neutrality, which prohibits priority charges, increases content provision, but consumer welfare still drops due to lower content quality and congestion externalities.