Direct and Indirect Benefits of Introducing Purchase Verification in E-commerce Platforms: Evidence from a Natural Experiment
39 Pages Posted: 11 Nov 2019 Last revised: 25 May 2021
Date Written: May 26, 2020
Abstract
E-commerce platforms struggle to create and maintain high-quality reputation systems. One promising option is the ``Verified Purchase'' (VP) badge, which confirms that the user reviewing a product purchased the product from the platform. Previous works comparing reputation platforms that require purchase verification with platforms that do not offer purchase verification found that review manipulation is easier in the latter. But what happens in platforms where purchase verification is optional (VP-optional)? How does the introduction of purchase verification alter review and reviewer characteristics?
In such VP-optional platforms, there is no monetary cost for posting fake reviews. Therefore, it is not clear why introducing purchase verification would have any effect on posting fake reviews. Yet, we argue that indirect costs related to response bias and the resulting disconfirmation between expected and realized product quality should discourage fake reviews. To investigate, we leverage a natural experiment to analyze 336,043 book reviews from two platforms (Amazon U.S. and Amazon U.K.). Indeed, we find empirical evidence that introducing purchase verification in a VP-optional market reduces fake reviews, the majority of which are positive. Directly, this reduction of fake reviews results in lower product ratings and longer reviews posted by more experienced reviewers. Indirectly, through increased attention, purchase verification increases the overall review helpfulness. These novel findings extend our understanding of how purchase verification can improve a platform's reputation system, and suggest managerial interventions for platforms that have yet to develop a verification mechanism.
Keywords: Online reputation systems, Purchase verification, E-commerce, Natural experiment, Difference in Differences, Document Embedding, Fake reviews
Suggested Citation: Suggested Citation