disclosure policy optimization

This paper investigates the structural characterization of feasible signal sets under privacy constraints. Methodologically, it introduces a unified framework that decomposes any feasible signal into two orthogonal components: a minimal-information signal and a conditional privacy-preserving signal—thereby fully characterizing the Blackwell frontier of the privacy variable. The framework integrates Blackwell’s theory of statistical information comparison, a general posterior-based privacy model (encompassing both differential privacy and inference privacy), and posterior-mean constraints on statistics. Crucially, it is agnostic to the specific privacy definition, ensuring broad applicability. The contribution is threefold: (i) it reveals the fundamental structure underlying the privacy–information trade-off; (ii) it derives a tight information-theoretic lower bound on signal informativeness under privacy constraints; and (iii) it establishes a rigorous theoretical foundation for designing and analyzing privacy mechanisms.

This study investigates how different information disclosure policies—no disclosure, full disclosure, and capped review—implemented by a third-party intermediary affect firms’ Q-learning pricing behavior under stochastic demand, with implications for algorithmic collusion and regulation. Contrary to classical theory, the findings reveal that when firms exhibit high discount factors (i.e., greater patience), no disclosure facilitates collusion more readily than full disclosure. Moreover, the capped review mechanism generally enhances firm profits across most scenarios. These results challenge the conventional assumption that restricting information sharing inherently mitigates algorithmic collusion, underscoring the necessity for regulatory frameworks to account for the learning dynamics and temporal preferences inherent in adaptive pricing algorithms.

This study addresses the challenge of maximizing utility in data disclosure under strict privacy constraints, where private data are inaccessible and each data point must satisfy a differential privacy leakage bound. The authors propose a multi-level pointwise leakage metric that transcends the conventional uniform threshold, enabling a hybrid regime combining full privacy and partial leakage. Leveraging information geometry, they derive a local quadratic approximation of mutual information under small leakage regimes, reformulating the problem as a constrained quadratic optimization solvable efficiently via singular value decomposition. Theoretically, they prove that when the leakage matrix is invertible, optimal utility can be achieved with merely binary outputs, for which a closed-form solution is provided—substantially simplifying mechanism design while preserving maximal information utility.

This paper addresses the inefficiency of standard mechanisms (e.g., VCG) in multidimensional type environments where agents hold both private preferences and shared, uncertain state information affecting common values. To restore social efficiency, we propose a novel mechanism design framework that integrates posterior behavioral data (e.g., user feedback) into incentive-compatible allocation. Our key innovation is the first incorporation of a state estimator directly within a VCG-style mechanism, yielding a theory of implementation grounded in posterior equilibrium. The framework unifies three canonical settings: full revelation, affine utilities, and consistent estimation—achieving exact social optimality in the first two, and asymptotic optimality in the third, with estimation error decaying at an explicit rate as estimator accuracy improves. Methodologically, we bridge Bayesian mechanism design, state estimation theory, and VCG extensions. We validate the framework through formal models of digital advertising auctions and LLM-based human–AI interaction.

This paper studies how a data platform monetizes state information in incomplete-information linear-quadratic games. Focusing on symmetric games with strategic complementarity or substitutability, it jointly designs information structures and incentive mechanisms. Methodologically, it fully characterizes the Gaussian mechanism family using Bayesian correlated equilibrium and the revelation principle. The key contributions are: (i) the optimal mechanism maximizes positive action correlation under complementarity and negative correlation under substitutability; (ii) under high uncertainty, the optimal recommendation is a deterministic linear function—yet not fully revealing; (iii) closed-form analytical solutions are derived for both social welfare and platform revenue maximization. Notably, this work is the first to establish that the optimal recommendation policy systematically reverses its correlation structure depending on the game type, and that, when type uncertainty is sufficiently large, the optimal policy is simultaneously linear, deterministic, and informationally parsimonious.

This study investigates the equilibrium bid-ask spread and welfare allocation in the Glosten-Milgrom market-making model when trade direction signals are perturbed by binary flipping noise—serving as a proxy for differential privacy mechanisms. By extending the concept of privacy subsidies from continuous Gaussian settings to a discrete two-state market microstructure, and combining information-theoretic modeling with Bayesian equilibrium analysis, the authors derive a closed-form solution: the equilibrium spread equals μ(1−2η)Δ, and each trade entails a privacy subsidy of μηΔ transferred from liquidity providers to informed traders. These findings demonstrate the universal impact of privacy-preserving perturbations within classical market models and provide a theoretical foundation for the joint optimization of privacy protection and market design.

This study addresses the challenge of imperfect sender control over information leakage in Bayesian persuasion by proposing an axiomatic framework for robust persuasion grounded in worst-case analysis. By characterizing the set of feasible information structures and employing axiomatic modeling, this work systematically evaluates optimal strategies across diverse leakage scenarios. The resulting general model of robust Bayesian persuasion not only advances the theoretical foundations of mechanism design under incomplete control but also examines a broad spectrum of information leakage instances. Consequently, this research establishes a novel analytical paradigm and provides rigorous theoretical support for understanding strategic communication when senders possess only partial control over information disclosure.

This study addresses strategic information sharing in a Cournot oligopoly under demand uncertainty, where firms are reluctant to disclose private signals due to fears of competitive disadvantage. Integrating game-theoretic analysis with differential privacy principles, the authors design a noisy aggregation mechanism augmented by an exogenous public signal to balance incentives for information sharing against privacy concerns. The analysis reveals that privacy protection alone is insufficient to induce truthful disclosure; a sufficiently informative public signal is essential. Firms possessing high-precision private signals demand stronger privacy guarantees. In a duopoly without public signals, information sharing never occurs, whereas in markets with three or more firms, effective sharing can be achieved through a combination of access control and public signals, highlighting the complementarity between privacy-preserving mechanisms and the informational environment.

研究在噪音信息处理下投资者如何观察公司报告，通过不同成本和噪音水平分析自愿披露行为，揭示了完全披露和部分披露的条件。

This study addresses the theoretical divergence in how private versus public transfers affect third-party beliefs and allocation outcomes within mechanism design. Drawing upon mechanism design, game theory, and information economics, it investigates preference-dependent and signaling mechanisms by contrasting equilibrium outcomes under these two transfer regimes. The central contribution establishes that equilibria achievable through hidden transfers can be arbitrarily approximated by public transfers, demonstrating that information distortion vanishes asymptotically over time. This finding indicates that transfer visibility induces only negligible informational distortion, thereby providing a rigorous theoretical foundation for assuming unobservable transfers in aftermarket models.