e-commerce marketplaces

This study investigates how online platforms jointly leverage three strategic instruments—pricing (commissions and transaction prices), matching (recommendation and search mechanisms), and bundling (product assortment)—to simultaneously enhance platform revenue and improve overall market welfare. By developing a game-theoretic model of multi-sided interactions and integrating equilibrium analysis with mechanism design theory, the paper systematically uncovers the mechanisms through which the interplay of these levers shapes participant behavior, transaction structures, and value distribution. The findings elucidate the intrinsic coupling among key platform design dimensions and offer theoretical foundations for governance strategies that balance efficiency and fairness in digital markets.

This paper investigates the design of optimal search-and-matching mechanisms for a monopolistic online platform operating under private user information and market frictions. Employing tools from game theory, search-and-matching theory, and information economics—specifically screening models—the study establishes, for the first time, that under complementary matching and transferable utility assumptions, the platform can achieve Pareto efficiency via an optimal screening mechanism: eliminating equilibrium mismatch entirely, inducing perfect sequential matching, and maximizing total social surplus. This result challenges conventional wisdom by reconciling informational asymmetry, platform market power, and allocative efficiency. It identifies precise structural conditions under which profit maximization is compatible with social efficiency—thereby resolving a longstanding theoretical tension between incentive compatibility, market design, and welfare optimization in two-sided platform markets.

This paper addresses the two-sided attention market formed by platforms, users, and content creators in online content ecosystems, aiming to unify the modeling of their dynamic interactions and heterogeneous evolutionary outcomes. Method: It innovatively formalizes bilateral attention allocation as mirror descent over a non-convex potential function, establishing the first coupled potential function model for such markets. Integrating game-theoretic reasoning, multinomial logit choice models, mirror descent dynamics, and non-convex optimization analysis, the approach designs a family of platform-ranking–induced potential functions to steer system evolution. Contribution/Results: The work yields the first optimization framework capable of uniformly explaining attention-market dynamics under diverse platform mechanisms. It provides novel theoretical guarantees on local convergence for non-convex potential landscapes and lays a rigorous foundation for analyzing content-ecosystem evolution and designing effective intervention policies.

This paper addresses the data-sharing reluctance of merchants in targeted advertising due to concerns over “customer hijacking.” To resolve this, we propose a tri-market coordination mechanism comprising a sales market (merchants sell data), an exchange market (peer-to-peer data swapping), and a purchase market (platform acquires data). Grounded in mechanism design theory, our framework models platform revenue, user engagement, and merchant welfare as a weighted optimization objective, integrating click-through-rate–driven dynamic pricing and formal data property rights assignment to mitigate trust barriers in data sharing. Unlike single-market designs, our approach achieves, for the first time under competitive settings, Pareto improvements in both data incentive compatibility and advertising allocation efficiency. Empirical evaluation on a large-scale platform demonstrates significant gains: +32% merchant participation rate and +19% advertising conversion rate. The work provides a scalable theoretical framework and practical paradigm for multi-sided data market design.

This work addresses exposure inequality, reduced diversity, and regulatory risks in online platform recommendations caused by algorithmic bias toward popular items. We propose FAIR, the first combinatorial item-selection framework explicitly enforcing *pairwise fairness*: it ensures approximately equal exposure probabilities for any pair of items via linear programming. Methodologically, we introduce a provably 1/2-approximation algorithm and a fully polynomial-time approximation scheme (FPTAS), integrating the ellipsoid method, parameterized knapsack approximation, and a dual separation oracle. Experiments on MovieLens and synthetic datasets validate FAIR’s effectiveness, quantify the “fairness cost” (i.e., the trade-off between fairness and utility), and demonstrate its ability to jointly optimize fairness, diversity, and recommendation revenue.

This work addresses the trade-off between fairness and efficiency in multi-item recommendation within heterogeneous two-sided markets. The authors propose an optimization framework that jointly accounts for disparities in consumer group utility, heterogeneity in producer capabilities, and platform business constraints. Innovatively integrating Conditional Value-at-Risk (CVaR) into consumer-side fairness modeling, the study extends formal fairness notions from soft single-item allocations to discrete multi-item recommendations. The resulting combinatorial optimization model embeds CVaR-based fairness objectives alongside practical business constraints and is paired with an efficient, scalable solver. Experimental results demonstrate that lossless fairness is unattainable in multi-item settings; however, moderate fairness constraints can enhance business metrics by diversifying exposure, while the proposed solver significantly reduces runtime without compromising solution accuracy.

This study addresses the dynamic assortment selection problem for online two-sided platforms under complete uncertainty regarding both customer and seller preferences. Within a discrete-time, heterogeneous-agent framework with incomplete information, the platform periodically displays sets of sellers, and matches are formed according to a multinomial logit model on both sides. We propose the first online algorithm that jointly learns preferences from both sides while simultaneously optimizing long-term revenue, thereby tackling the challenging setting where preference parameters on both sides are unknown. Theoretical analysis shows that the algorithm achieves worst-case regret growing at a polylogarithmic rate, which matches the established lower bound and thus attains the optimal theoretical rate.

This study addresses the challenges of extreme demand volatility, delayed pricing responses, and misalignment between short-term revenue and long-term profitability during major fashion e-commerce promotions. To tackle these issues, the authors propose a high-frequency “predict–optimize” automated pricing system that breaks away from traditional weekly decision cycles by operating at a minute-level granularity. The system achieves the first industrial-scale deployment of daily multi-objective dynamic pricing in large-scale e-commerce settings, combining gradient-boosted tree models for daily demand forecasting with a multi-objective optimization framework to generate real-time pricing strategies that jointly maximize long-term profit and net merchandise value. Evaluated across 23 A/B tests in 12 Zalando markets from 2023 to 2024, the system delivered approximately 6% higher profit while maintaining sales volume and has since been fully deployed for promotional pricing.

This study addresses how platform imitation of successful third-party products in its own marketplace may undermine sellers’ incentives to innovate. The authors develop a Stackelberg game-theoretic model in which the platform, as the leader, makes strategic entry decisions, while sellers, as followers, optimize their exploration–exploitation trade-offs. In the single-seller setting, the optimal policy is derived using the Gittins index; in the multi-seller case, deep reinforcement learning is employed to analyze equilibrium behavior. This work presents the first integration of the Gittins index with deep reinforcement learning, systematically uncovering the incentive compatibility mechanism between platform entry strategies and seller innovation. The theoretical findings align with empirical observations from real-world platforms such as Amazon and Google Play, offering rigorous support for policies aimed at preserving market innovation and diversity.

This study addresses the challenge of accurately identifying the causal effect of new supply on total transaction volume or value in heterogeneous-product two-sided markets. The authors propose a novel framework that integrates double (debiased) machine learning with hierarchical Bayesian modeling, innovatively incorporating a product-segment similarity metric from spatial economics literature as a key feature to characterize the incremental impact of supply across distinct market segments. By effectively combining causal inference techniques with domain-specific prior knowledge, the approach enhances both estimation accuracy and interpretability. Empirical results on Airbnb data demonstrate that the model not only yields substantially more precise and interpretable estimates but also exhibits robust out-of-sample predictive performance.