Merger Analysis in the App Economy: An Empirical Model of Ad-Sponsored Media

This paper proposes a new model of imperfect competition of ad-sponsored media, which can sell "free" products, for a merger analysis applicable to the mobile app industry. To analyze developers' monetizing with both price and advertising in an app, we consider a consumer who faces both budget and time constraints. Moreover, to catch up with newly created and quickly redefined markets, we automate the conversion from in-text product descriptions to numerical product attributes by combining word embedding and dimension reduction techniques. The model defines an equilibrium over consumers' downloads, usage, and in-app purchase decisions and app developers' price and non-price competition. We estimate the model using mobile app data from Japan from 2015 to 2017. The estimate revealed that the marginal disutility of watching advertisements is 12.4% of the ad price for games and 3.1% for other apps. The relevant markets defined by a Small, Non-transitory but Significant Increase in Cost (SSNIC) test are larger than the product categories. Merger simulations show that the app market is, at the static level, competitive and even a merger among the top 10 apps has negligible effects on surplus. The proportional transaction fee imposed by the platform, whose welfare implication is ambiguous because it increases the advertisements and decreases the download prices, is more influential. For game apps, the total surplus is maximized at 12%-15% rather than the actual 30%, increasing welfare by 2.4% and app developers' profits by 44%. For other apps, the total surplus curve is almost flat around 30%.