On the Amazon.com marketplace, both Amazon and small businesses compete in offering retail products. However, Amazon chooses what products consumers see when they search. Products sold by Amazon may have a better position compared to small business products, but the effects on consumers and sellers are unclear. Policymakers have expressed antitrust concerns about this, suspecting "self-preferencing" and "gatekeeper" market power. To study this, I develop a model where heterogeneous consumers search for differentiated products arranged on an acyclic graph (i.e., tree). Firms price in response to consumer search and how their products are arranged—highlighting how search design determines market structure. The model endogenizes consideration set formation and recovers the correlated distribution of consumer preferences and search costs. Estimated on Amazon data, I show that not accounting for product arrangement (e.g., search results and BuyBox) leads to incorrect price elasticity estimates. I provide three results on market power and antitrust policies using counterfactual product arrangements. (i) To isolate the effect of Amazon's position advantage, I remove it through a "neutral" product arrangement. Profits shift from Amazon to small businesses, confirming Amazon's sizable market power. However, consumer welfare falls when consumers reduce their search intensity in response to reduced value from searching. This suggests Amazon's incentives and consumers' preferences are aligned, weakening the claim of self-preferencing. (ii) Banning the platform owner from also being a seller reduces consumer welfare through price rather than product variety. (iii) I propose an alternate policy, splitting the platform into an Amazon side and a small-business side. Giving consumers the ability to search for and "support small businesses" would alleviate the market power imbalance without harming consumers.
Artists and policy makers are concerned that Generative Artificial Intelligence (GenAI) may lead to disappearance of non-GenAI content. In this paper we study the implications of GenAI for the production and consumption of creative goods; such as images, music, and writing. We first introduce a simple model of technology adoption and production that highlights how GenAI may influence market equilibrium. Then, using a difference-indifferences design, we causally estimate the impact of GenAI on production, firm entry, a measure of product quality, variety, and sales. We find that GenAI is a substitute for non-GenAI content, increases competition in markets, and crowds-out the production of non-GenAI content. Overall this leads to an increase in the quality and variety of produced and sold goods, and increased sales. Thus, our results imply that unregulated GenAI poses a substantial threat to non-GenAI production but is likely beneficial for most consumers. We leverage heterogeneity across markets to examine the role of market structures and legal differentiation and labeling in mitigating or enhancing GenAI adoption and influencing market equilibria. Evidence suggests policy that regulates content labeling and enforces clear disclosure can mitigate concerns that poor quality GenAI products may lead to market collapse.
Many valuable digital products (e.g., Youtube, Google Maps, LLMs) are provided at no charge. But they are not "free" nor costless to supply, rather the "price" consumers pay is with attention to advertising, which funds their production. The importance of ads for enabling attention product markets are not well understood and elasticities important for welfare lack study. Despite this, the Department of Justice is suing Google to improve competition in the $224b upstream advertising market, with unclear effects on downstream consumers. Lower ad prices may reduce production and increase ad load, but ad quality improvements may benefit consumers. Using data on Youtube's $29b market, I apply RDDs and IVs to derive causal estimates of ad attention (price) elasticities of demand and supply to inform the welfare consequences of proposed antitrust action. Content creators can choose between platform ads (ad rolls), and disintermediated ads (sponsorships), with an additional ad unit reducing viewership by 22% and 7%, respectively. I use a simple structural model of video consumption and content creation to analyze the effects of potential reduction in ad revenue and improved ad quality to understand consumer welfare under different antitrust outcomes.
The share of US students enrolled in entirely-online college degrees has doubled in recent decades (from 5% in 2008 to 10% in 2015). The importance of online colleges as a differentiated product in the higher education market is likely to increase, particularly after the online learning experiment of the 2020 pandemic. Policymakers concerned about tuition and student debt growth are interested in whether online degrees could put downwards pressure on tuition and increase access. However, the effect of online degrees on student enrollment, tuition competition and post-graduation outcomes remains little studied. In principle, the online degree market should be highly competitive, as it lacks the geographic market power that in-person colleges wield and is highly scalable. However, limited awareness of online colleges may limit competition and advertising may influence the student online college search process. We provide evidence that online colleges: (i) cater to specific segments of students (e.g. mature age students with family); (ii) deliver a wide range of post-graduation outcomes that are no worse on average than in-person colleges; (iii) face stronger competition, but with little measurable spillover on in-person colleges; and (iv) use advertising to distort the information gathering process for potential students and increase enrollment.
Competing retail platforms (such as Amazon and eBay) choose different ways of arranging the products shown to consumers. Why does this occur, and is there an optimal arrangement of products? I propose a model where consumers search through an acyclic graph (i.e., tree) to model a retail platform’s incentives to arrange products. I show that a monopoly platform may obfuscate search in equilibrium, using product arrangement to extract surplus from participants. However, under competing duopoly platforms, I show that pooling and separating equilibria are possible. In particular, there is a separating equilibrium where one platform groups similar products under intense price competition and the other platform does not, mirroring the observed search designs of Amazon and eBay. Search design is used to discriminate between consumers with different search costs. High search cost consumers prefer the platform that groups products as this minimizes searching, while low search cost consumers prefer the other platform since they can benefit from considering a wider range of products. The resulting pricing on each platform exhibits different dispersion that supports the separating equilibrium.