Causal ML Estimates Marketplace Supply Incrementality

Understanding the causal link between increased supply and marketplace performance is crucial for two-sided platforms with diverse products. This research addresses the challenge of quantifying how additional supply translates into outcomes like total transactions or value. The study uses the Airbnb marketplace as a case study, specifically examining the effect of new listings on overall bookings, though the approach is broadly applicable. The proposed methodology integrates double/debiased machine learning with a hierarchical Bayesian framework. This combination allows for robust causal inference while incorporating prior knowledge. A key innovation is the use of tractable and informative features derived from geospatial literature to measure product segment similarity, enhancing the model's ability to differentiate impacts across various product categories. The model yields plausible estimates for the returns on additional supply within the marketplace and demonstrates strong out-of-sample performance. This provides marketplace operators with a more accurate tool to predict the impact of supply-side interventions and optimize growth strategies.