Global Average Treatment Effects for Individualized Randomization Experiments with Aggregate Data

2026-05-26

AI Summary

This study addresses the challenge of causal inference in two-sided markets, where strong temporal and cross-unit interference combined with access only to aggregated data renders conventional methods invalid. The work establishes, for the first time, the identifiability of the Global Average Treatment Effect (GATE) under purely aggregated observations. It introduces the IRE-VCDP model, which captures interference mechanisms in supply-demand dynamics through a varying-coefficient decision process, and develops a comprehensive framework for GATE estimation and statistical inference. By integrating individualized randomized experimental design, aggregated data analysis, and causal theory, the proposed approach demonstrates both theoretical guarantees and empirical validity, as confirmed through extensive simulations and real-world experiments on a leading ride-hailing platform.

Abstract

Individualized randomized experiments are central to online platforms for optimizing personalized decisions in complex environments. In two-sided markets, however, standard treatment effect estimation is often invalid due to strong temporal and cross-unit interference, a challenge compounded when only aggregated data are available because of privacy or system constraints. To address these issues, we identify the Global Average Treatment Effect (GATE) using only group-level data from treatment and control groups. We first establish identification conditions based on aggregated observations, and then propose the Individualized Randomized Experiment Varying Coefficient Decision Process (IRE-VCDP) model, which accounts for interference through supply-demand dynamics. Building on this framework, we develop a complete procedure for estimation and statistical inference of the GATE, along with theoretical guarantees for the proposed test. Extensive simulations and real-world experiments using data from a leading ridesharing platform demonstrate the effectiveness of our approach.

Problem

• interference
• aggregate data
• causal inference
• two-sided markets
• treatment effect

Innovation

• Global Average Treatment Effect
• aggregate data
• interference
• two-sided markets
• IRE-VCDP