comparative statics

This paper addresses the heterogeneity and nonlinearity of treatment effects in share-shift designs. We propose an identification framework that, under exogenous treatment shares, constructs a triangular model and incorporates a control function to correct for endogeneity, enabling systematic identification of four target parameters: (i) observable heterogeneity, (ii) a comprehensive heterogeneity measure, (iii) counterfactual policy allocation mechanisms, and (iv) their employment effects. Relative to conventional linear instrumental-variable approaches, our method uncovers pronounced heterogeneity and nonlinearity in the impact of Chinese import shocks on U.S. manufacturing employment—avoiding aggregation bias inherent in average-effect estimates. The proposed heterogeneity measure is both interpretable and policy-comparable. Empirical results demonstrate that the method substantially improves causal inference precision, overcoming a fundamental limitation of classical instruments in modeling structural heterogeneity.

The rapid proliferation of AI agent technologies renders static regulatory frameworks vulnerable to strategic manipulation, undermining market fairness. This study addresses bargaining, negotiation, and persuasion as three canonical game-theoretic settings, modeling how AI agents interact in contexts involving resource allocation, asymmetric information exchange, and strategic communication. The work proposes the “poisoned apple effect”—a phenomenon wherein participants strategically introduce unused or non-adopted technologies to influence regulatory decisions and thereby skew equilibrium payoffs in their favor. This insight reveals that technological expansion itself can function as a strategic instrument for manipulation, challenging the adequacy of conventional static regulatory paradigms. The paper advocates for a shift toward dynamic market design capable of adapting to the evolving strategic capabilities of AI agents and mitigating associated risks.

In strategic equilibrium environments, agents’ strategic behaviors induce endogenous treatment assignment, posing challenges for causal inference. This paper proposes the Strategic Doubly Robust (SDR) estimation framework, which integrates game-theoretic strategic equilibrium modeling into the doubly robust estimation paradigm. Under a “strategic ignorability” assumption, SDR simultaneously addresses strategic unobserved confounding and model misspecification risks, preserving consistency, asymptotic normality, and doubly robust protection. Theoretical analysis establishes its statistical reliability under strategic confounding; empirical evaluations demonstrate that SDR reduces estimation bias by 7.6%–29.3% across varying strategic intensities relative to baseline methods, while maintaining scalability with increasing agent populations. The core contribution is the first systematic incorporation of strategic equilibrium structure—rooted in game theory—into the doubly robust causal inference framework, thereby bridging strategic interaction modeling with robust causal estimation.

This paper identifies a fundamental bias in estimating average treatment effects (ATE) in continuous-parameter A/B tests—particularly price experiments—arising from interference among market participants. In pricing contexts, conventional estimators of profit change expectations can exhibit sign reversal, leading firms to adopt profit-damaging pricing policies. To address this, we propose a lightweight debiasing method requiring only equal partitioning of experimental units. We are the first to systematically characterize the “sign reversal” phenomenon and prove its ubiquity in two-sided markets and multi-category commission pricing. Through structural modeling and differential analysis, we derive an explicit closed-form expression for the bias and theoretically demonstrate that the classical estimator can indeed flip sign. Empirical evaluations across diverse market settings confirm that our method consistently restores correct decision directionality, thereby ensuring reliable causal inference.

This paper addresses the mechanism design problem of collaborative adaptation of general-purpose large language models (LLMs) between model providers and domain-specific adopters within industrial ecosystems. It tackles challenges concerning multi-party cost sharing, revenue allocation, and strategic interaction. Methodologically, it pioneers the integration of bargaining games and subgame-perfect equilibrium into the LLM adaptation process, establishing a dynamic negotiation framework grounded in the Nash bargaining solution and explicit cost–revenue functions. Theoretically, it proves that a Pareto-optimal revenue-sharing mechanism exists under broad parameter conditions; even high-cost parties can lead and sustain cooperation; and it rigorously characterizes necessary and sufficient conditions under which domain adopters choose among three strategic responses: active contribution, strategic free-riding, or voluntary exit. These results provide a verifiable game-theoretic foundation for incentive alignment and cooperative governance in the industrial deployment of LLMs.

This study addresses a key limitation of traditional instrumental variable (IV) models, which assume deterministic relationships between treatment selection and potential outcomes under an instrument, thereby failing to capture stochastic decision-making in real-world settings. To overcome this restriction, the paper develops a micro-founded framework in which both potential outcomes and treatment choices exhibit individual-level randomness. Response types are redefined as state-dependent treatment probabilities coupled with distributions of potential outcomes, grounded in expected utility maximization subject to information constraints. Within this framework, conventional IV estimands are reinterpreted not merely as local average treatment effects for compliers, but as population-weighted averages of treatment effects, where weights correspond to individual changes in treatment probability induced by the instrument. This approach yields a more flexible and interpretable characterization of treatment effect heterogeneity.

This study addresses a critical limitation of conventional Cross-Impact Balance (CIB) analysis, which yields only static consistent scenarios and cannot quantify dynamic structural aspects such as transition efforts, key leverage points, timing of adjustments, or responses to external shocks. To overcome this, the authors introduce linear response theory into the CIB framework, exploiting the structural isomorphism between the CIB drift matrix and the Leontief input-output matrix. This enables the derivation of four analytical constructs—Type I cross-impact multipliers, perturbation budgets, impulse response functions, and unit impulse shock profiles—each admitting closed-form solutions that characterize indirect effects, transition resistance, dynamic adjustment pathways, and network sensitivity in socio-technical systems. Applied to an energy transition case, the approach successfully computes all dynamic indicators across five structural equilibria, offering a transferable quantitative toolkit for assessing system resilience, designing transition pathways, and informing policy interventions.

This study addresses the uncertainty surrounding the real-world impacts of carbon offset policies on aggregate emissions and welfare, which stems in part from conventional carbon accounting metrics’ inability to capture general equilibrium spillovers. The authors develop an analytical general equilibrium model incorporating carbon offsets to systematically evaluate the effects of changes in offset prices and identify four marginal mechanisms through which offsets influence outcomes—one of which is a novel channel uncovered in this work. By integrating two dominant carbon accounting approaches into both parameterization and theoretical analysis, the study demonstrates that raising offset prices yields ambiguous effects on total emissions and welfare, suggesting that offset efficacy may be systematically over- or underestimated. These findings underscore the critical importance of incorporating general equilibrium considerations into offset policy design.

Existing causal models struggle to distinguish between the immediate and persistent effects of interventions in time-dynamic systems, particularly when such interventions alter the system’s equilibrium behavior. This work proposes a novel paradigm grounded in system and state representations, integrating causal directed acyclic graphs, the potential outcomes framework, and dynamic systems theory. By introducing an equilibrium-state assumption and employing state-space modeling, the study reformulates the causal inference framework to better capture temporal dynamics. It innovatively defines an equilibrium-oriented “zero effect” concept and combines it with a strategic selection of time points to enable valid identification of time-varying causal parameters. The approach establishes clear criteria for categorizing causal effects under dynamic interventions, substantially enhancing the interpretability and practical utility of causal inference in equilibrium analysis.

This study investigates the impact of machine learning algorithms on information aggregation in asset markets with dispersed information, focusing on whether price mechanisms can fully reflect the information extracted by such algorithms. The authors introduce the Chow-Liu tree into a general equilibrium framework à la Hellwig (1980), constructing an equilibrium model in which agents employ this algorithm for Bayesian inference. The analysis reveals that even when agents are initially homogeneous, they endogenously develop heterogeneous beliefs, demand functions, and utilities. More importantly, while machine learning enhances information processing in partial equilibrium, it leads to less informative prices in general equilibrium compared to the rational expectations benchmark, indicating that market prices fail to efficiently aggregate the information uncovered by machine learning.