Taming Tech Giants’ Algorithms: The Amazon Buy-Box Case

Taming Tech Giants’ Algorithms: The Amazon Buy-Box Case

Overview
Many digital platforms operate as two-sided markets, facilitating the matching between sellers and buyers. Due to ICT developments, online marketplaces such as Amazon are increasingly relevant in our societies. The functioning of the Amazon platform is of great scientific interest because of the dual role the tech giant plays, which poses several policy-relevant questions. On the one hand, Amazon controls buyers and sellers’ participation and interactions through its platform design choices, governed by proprietary algorithms that are opaque and not transparent. On the other hand, Amazon operates as a seller itself, leveraging its gatekeeper power and the information collected on both sides of the market to gain a competitive advantage over other market participants and, ultimately, shape competition. Amazon controls the rules of the game by exploiting information and algorithmic systems, but also participates in it as a player. Regulators fear that algorithms can reduce competition and harm consumers: there is a raising awareness about the risks algorithms pose and a call for their effective regulation. The European Commission (10.11.2020) fears that “the use of seller data allows Amazon to avoid the normal risks of retail competition and to leverage its dominance in the market”. We will contribute to this debate providing new insights on proprietary algorithms. Our research will also answer regulatory questions about how these algorithms are used, how sellers and consumers’ information is exploited on two-sided platforms, and how consumer behavior is shaped by the platform’s strategic choices. Ultimately, we plan to provide solutions for antitrust in digital markets both from a competition policy and a consumer protection perspective.  We focus on a very specific concern raised by the design of Amazon’s online platform: a customer can directly proceed to checkout by clicking on the “add-to-cart” button, known as the Buy-Box. However, it is Amazon’s ranking algorithm that selects the seller to be placed in this prominent position, whereas all the other sellers listing the same product are relegated to subsequent pages. The Buy-Box algorithm represents an extreme case of ranking algorithms, where the selection process operated by the algorithm leads to a binary winning-losing outcome. Being in the Buy Box is crucial for sellers, as available evidence (Chen et al. 2016) suggests that roughly 80% of customer purchases go through this button. The main issue related to the Buy-Box is that the algorithm selecting the Buy-Box seller is neither observable nor decodable.  The criticality of this design choice is recognized also by the European Commission, which plans to investigate “whether the criteria that Amazon sets to select the winner of the Buy-Box lead to preferential treatment of Amazon's retail business”.

Research questions
We analyze Amazon’s dual role from the two sides of the market, investigating three interconnected questions. On the sellers’ side: (1) Which sellers’ features explain the Buy-Box winning dynamics and how does the prominent Buy-Box position granted to a specific seller impact competition?  On the consumers’ side: (2) How relevant is the Buy-Box channel for actual sales on the platform and how can information provision on the Buy-Box affect consumers’ search behavior? (3) Does Amazon, based on consumers’ characteristics, personalize search results or, even, displayed prices on own products?

Policy relevance
The first question responds to a fast-growing regulatory demand to investigate the functioning of algorithms and their impact on competition. The second and the third questions pertain to the design of the platform environment and the interface consumers interact with, aiming to quantify the salience of the Buy-Box for buyers and their literacy on its functioning, and to tackle the issue of consumers’ data exploitation and discrimination on digital platforms.The final objective of our project is to propose viable policy recommendations. The relevance of our findings could go beyond purely scientific interest and contribute to informing policy-makers on how to design a more effective regulation of algorithms. To this extent, we will also shed light on consumers’ behavior and decision making in online environments through a novel approach that combines observational and experimental data. Consumer protection is a first-order policy concern, and our findings could help understand how to implement measures to foster transparency and consumer awareness in online marketplaces, which would be less intrusive (and more viable for the regulator) than other remedies such as platforms’ design changes.

Where to meet us:

  • August 31 - September 3, 2022, European ESA Meeting, Bologna
  • August 25-27, 2022 - EARIE, Vienna
  • November 18, 2021 – EAYE: 3rd European Association of Young Economists Workshop on “Field Experiments and Experiments with Non-Standard Subjects”, University of Innsbruck
  • September 9, 2021 – The Sixth Meeting of the Behavioral and Experimental Economics Network (BEEN) Meeting, University of Bologna, Department of Economics
  • June 25, 2021 - ZEW Internal Seminar, Mannheim
  • June 9, 2021 - Lear Competition Festival, Rome
  • June 7, 2021 - Informal Discussions on Experimental Economics (IDEE) Meeting, University of Bologna, Department of Economics

Project members

Francesco Clavorà Braulin

Francesco Clavorà Braulin

Project Coordinator
Junior Research Associate

To the profile
Michela Boldrini

Michela Boldrini

Project Coordinator

To the profile

Contact

Francesco Clavorà Braulin
Junior Research Associate
Francesco Clavorà Braulin, PhD
To the profile