Targeting Viewers' Heterogeneous Ad Aversion: Evidence from a Two-Sided Market

Abstract

This paper studies the role of viewers’ heterogeneous ad aversion for media content demand and advertisers’ willingness to pay. Using high-frequency individual level data from free-to-air TV permits us to observe viewers’ minute level choices from the consideration set of alternatives. We first illustrate the potential selection challenges of using market level data to estimate viewers’ demand for content and average ad elasticity. We find that ad elasticity varies considerably by content; however this could be due to heterogeneous individuals self-selecting into particular content. To address this problem we exploit minute-level data on individual choices to estimate viewers’ heterogeneous ad aversion adapting the demand model in (Dubois et al 2020) to media content. With this approach we prioritize estimating heterogeneous preferences without placing distributional assumptions on individuals’ ad aversion or on how it correlates with observable demographic characteristics. We find that ad aversion is highly heterogeneous and not strongly correlated with observable socioeconomic characteristics such as socio-economic status or gender. We can also disentangle ad aversion from cohort preferences for content and idiosyncratic inertia/state dependence, which are relevant also for the advertisers’ side of the market. Our findings indicate that advertisers’ willingness to pay per impression is positively associated with the content’s ability to reach audiences with high ad aversion. We find robust evidence of a per-impression price premium for ad slots that target individuals with higher levels of ad aversion.