Self-Preferencing and Consumer Welfare: Evidence from a Field Experiment

How to cite

Fradkin, Andrey; Farronato, Chiara; MacKay, Alexander. 2023. "Self-Preferencing and Consumer Welfare: Evidence from a Field Experiment." AEA RCT Registry. AEARCTR-0011370. https://doi.org/10.1257/rct.11370-3.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Andrey Fradkin* , Chiara Farronato , Alexander MacKay

* Primary investigator

Registry ID
AEARCTR-0011370
DOI
https://doi.org/10.1257/rct.11370-3.0
First registered
Last updated
Status
Completed
Trial dates
2023-06-13 – 2023-10-31
Countries
United States of America ()
Registry record
http://www.socialscienceregistry.org/trials/11370

Abstract

We seek to study how self-preferencing by digital platforms affects consumer welfare and competition.

Intervention

Group 1 is the control group, for which the extension simply monitors participants' behavior. The extension tracks browsing behaviors on Amazon.com, ensuring that confidential information is removed before storing the data. Additionally, the extension tracks URLs visited on other major e-commerce websites, but it does not collect more detailed tracking data or intervene with what the participants see on these websites.
Group 2 is the Amazon treatment group. For this group, the extension hides major Amazon-owned brands from the participants' browsing experience on Amazon.com. This mainly includes removing Amazon-branded products from search and product pages.
Group 3 is the random treatment group. This group is similar to group 2, except that instead of hiding the Amazon-branded products from a web page, the extension hides a random set of products, where the number of hidden products is equal to the number of Amazon-branded products on that web page.

See the included document for details.

Experimental design

See the pre-registration document and analysis plan.

Design details

Randomization method
Simple randomization by computer
Randomization unit
The unit of randomization is the study participant.
Clusters
1200 participants, with 1000 minimum if costs of recruitment increase substantially
Observations
1200 participants, 6 wishlist task choices per participant
Arms
400 control, 400 amazon brands hide, 400 random hide
Primary outcomes
See the pre-registration document and analysis plan.

Papers from this trial

  1. Farronato, Chiara, Andrey Fradkin, and Alexander MacKay. Vertical Integration and Consumer Choice: Evidence from a Field Experiment. No. w34135. National Bureau of Economic Research, 2025. https://andreyfradkin.com/assets/amazon_exp_welfare.pdf

Cited by

No citation data available for this registration yet.

Keywords