Internet advertising with information overload.

Authors
  • CHOI Dong ook
  • RENAULT Regis
  • BIANCINI Sara
  • DONNI Olivier
  • CELIK Gorkem
  • TREGOUET Thomas
  • CHONE Philippe
  • SOVINSKY Michelle
Publication date
2015
Publication type
Thesis
Summary Attention is a scarce resource. This study examines how users' limited attention affects the Internet advertising market. I analyze the Korean search engine market and present a structural estimate of the demand for banner ads that takes into account the limited attention of users. The theoretical model underlying this estimation assumes that users' attention is heterogeneously distributed. In this framework, I show that a monopoly platform tends to serve too many ads and that the probability of ads reaching their target depends on users' and websites' characteristics. A discrete choice model with random coefficients is used to estimate users' demand function. The two demand functions and the profit function of the platforms are used to derive the margin rate. The latter is composed of three components: a network effect on the user side, a congestion effect and the Lerner index.The reduced form estimation models are discussed as an introduction.They seem to reveal the existence of an information congestion phenomenon in the Korean search engine market.The structural estimation that I implement consists first of simulating the probability that a user pays attention to an advertisement posted on a Web site. The estimated model predicts the probability that a given advertisement will be perceived by the users of a website. The estimated model predicts the probability that a given advertisement will be perceived by the users of a website. This probability varies from 0.19 to 0.58 depending on the website and also varies according to the gender, age and education level of the users. A 1% increase in the number of ads displayed on a Web page results in a 0.06% decrease in the probability that an ad message will be perceived by users.The estimated user demand reveals that advertising is a nuisance for 92.17% of users. By estimating the demand for advertising, I conclude that the externality related to the limited attention of users is an important determinant of the margin rate of platforms.The study finally shows that the level of advertising on websites is systematically below the socially optimal level, which implies that the market power enjoyed by the sites leads to a restriction of the number of banner ads that goes beyond what would be socially optimal, given the negative externality of information congestion. This results in an increase in the advertisers' surplus but a decrease in the social surplus (deadweight loss).I finally perform a counterfactual experiment on the impact of ad blocking software. This experiment reveals that the generalization of such software reduces ad levels, because the valuation of users by a platform corresponds to the average number of advertising messages they receive. These results highlight that taking into account the phenomenon of information congestion is essential in estimating the demand for audiovisual content and advertising space.
Topics of the publication
Themes detected by scanR from retrieved publications. For more information, see https://scanr.enseignementsup-recherche.gouv.fr