Random Graph Asymptotics for Treatment Effect Estimation in Two-Sided Markets

Yichen Pan, Xiaowu Dai

Keywords: Two-sided markets, Random graph models, Network interference

Abstract: In two-sided markets, the accurate estimation of treatment effects is crucial yet challenging due to the inherent interference between market participants, which violates the Stable Unit Treatment Value Assumption (SUTVA). This paper introduces a novel framework that leverages random graph asymptotics to model and estimate treatment effects under network interference in two-sided markets. By incorporating a random graph model, we handle two-sided randomization by modeling customer interference within the potential outcome function as a function of graph topology and equilibrium dynamics, while capturing listing interference through the random graph structure. Our new estimation process provides asymptotically normal estimators with robust theoretical properties, suitable for large-scale market scenarios. Our theoretical findings are supported by extensive numerical simulations, demonstrating the effectiveness and practical applicability of our approach in estimating direct and indirect causal effects within these complex market structures.

Primary Area: causal reasoning

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 11085