Game On : The Interplay between Celebrity Endorsements and Peer Networks in the Video Game Industry

This paper examines how celebrity endorsement and peer influence affect video game usage and sales using data from a large social network. To investigate these phenomena, we constructed a four-year longitudinal panel dataset of 1.2 million video game players from the popular game Dota 2. We leveraged variations in both endorsements and product characteristics to address key identification challenges. We analyzed how endorsements interact with existing peer influence and how product features such as observability and complexity moderate the effects of endorsements and peer effects. Our results show that endorsements and peer effects are positive but substitutive -- consumers rely less on their peers after endorsements. Both effects are stronger when the product is more observable or complex; observability plays a bigger role for peer effects than complexity. Endorsements enhance game usage, but only for the products that achieve a higher reach in tournaments. Endorser performance affects sales but not usage, while win rate has no effect on either outcome. Techniques such as regression discontinuity designs, Difference in Difference models, and instrumental variable models show that the peer and endorsement effects we find are robust