We examine self-preferencing in Amazon's 'Similar item to consider' box. We document a stark difference across domains; while Amazon frequently recommends alternatives to Amazon Basics (AB) products on the US website, it systematically suppresses these recommendations on the Canadian website. Leveraging this discontinuity, we estimate the causal effect of algorithmic exclusion due to self-preferencing. We find that non-AB products that merit a recommendation in the US but are denied one in Canada suffer a 11% decrease in the volume of sales. This finding quantifies the significant market distortion caused when gatekeepers favor their own inventory over rival goods.
On platforms with peer-to-peer reviews, new products encounter the so-called “cold- start” problem. Little-known products are bought too rarely and remain little known. This paper studies how policies aimed at mitigating the cold-start problem, specifically by low- ering prices for new products, affect welfare and how much of the welfare effect is due to a change in the speed of social learning. Based on data from Airbnbs located in Manhattan, we estimate a structural model with short-lived consumers and long-lived hosts who make dynamic entry, exit, and pricing decisions. We then simulate a counterfactual reduction in Airbnb’s revenue fee for new listings. While this increases total welfare, we find that it does not accelerate social learning. By contrast, if the cost of the fee reduction is passed on to hosts, faster social learning in itself increases consumer welfare by up to 5% and is the main driver of the total welfare gain.
We want to find out whether the concern of seller and consumer harm as a result of Amazon entry finds emprical support beyond anecdotal evidence provided by the media. To this end, we measure the predictors and effects of Amazon first-party retail entry on consumer and third-party merchant outcomes in the Home & Kitchen department of Germany’s Marketplace between 2016 and 2021. While the empirical setting presented challenges for estimating causal effects, our results are broadly inconsistent with systematic adverse effects of Amazon entry on Amazon Marketplace.