welfare analysis

This paper addresses the challenge of unobserved individual welfare effects under policy interventions. We propose a distributionally robust evaluation framework grounded in the superquantile (i.e., conditional value-at-risk), which characterizes tail behavior of unobserved welfare impacts without requiring individual-level heterogeneity data. Unlike conventional average treatment effect approaches, our method systematically identifies highly adversely affected subpopulations and patterns of welfare inequality. Innovatively integrating superquantile theory with marginal treatment effects (MTE), the generalized Roy model, and compensating variation analysis, we establish the first framework for *distributional causal inference* of policy welfare impacts. The approach jointly optimizes efficiency and equity considerations. Empirical validation across price-change analyses, self-selected treatment assignment, and social program cost–benefit assessments demonstrates its capacity to rigorously bound welfare gains and losses—thereby enhancing the verifiability and targeting precision of redistributive policy design.

This paper investigates how a planner can jointly intervene in agents’ marginal utilities and network link weights to maximize social welfare while characterizing distributional consequences. We model strategic interactions and externality propagation using game-theoretic tools and develop a network equilibrium framework. We formally characterize the efficiency–fairness trade-off under joint intervention—the first such analysis—and introduce a distribution-sensitive metric for intervention effectiveness. Using convex optimization and counterfactual welfare allocation analysis, we uncover an inherent paradox: aggregate welfare gains co-occur with increased inequality in welfare distribution. Our main contributions are: (1) the first theoretical model of welfare trade-offs under joint intervention; (2) Pareto-improving design principles for interventions; and (3) computationally tractable, interpretable policy guidelines for network governance that simultaneously address efficiency and equity objectives.

Standard difference-in-differences (DID) methods struggle to identify counterfactual distributions under regulatory policies—such as minimum wage laws—when confronted with mass points, distributional discontinuities, nonstationarity, or unobserved selection bias. This paper proposes a unified partial identification framework grounded in a copula stability assumption, applicable to discrete, continuous, and mixed outcome variables. Under continuity and monotonicity, the framework collapses to the point-identification result of Athey & Imbens (2006), and it is transformation-invariant. Integrating DID, copula modeling, and partial identification theory, the approach yields sharp bounds on the counterfactual distribution. Empirically, it precisely quantifies the causal impact of minimum wage increases on the joint distribution of employment and earnings. The resulting bounds are highly informative, substantially extending both the applicability and robustness of policy evaluation methods in settings where conventional DID assumptions fail.

This paper addresses the fundamental tension between “targeted assistance” and “causal effect estimation” in public resource allocation, proposing the first randomized assignment framework that jointly optimizes high-need individual identification and average treatment effect (ATE) estimation. Methodologically, it introduces the first learn–intervene dual-objective Pareto frontier, theoretically characterizes the sample complexity lower bound, and designs a computationally efficient policy optimization algorithm. Key contributions include: (i) elevating program evaluation to a systemic objective on par with need identification; (ii) providing theoretical guarantees for the optimal trade-off between intervention utility and ATE estimation accuracy; and (iii) empirical validation on Allegheny County social services data, demonstrating that the framework achieves 90% of optimal assistance utility while requiring less than twice the sample size needed by a pure randomized controlled trial (RCT) for comparable ATE precision.

This paper addresses the vulnerability of conventional utilitarian policy learning—based on the conditional average treatment effect (CATE)—to outliers and its inability to flexibly accommodate policy caution or leniency under heterogeneous individual treatment effects. We propose an optimal intervention allocation framework grounded in the conditional quantile treatment effect (QoTE). To our knowledge, this is the first work to incorporate distributionally robust welfare into policy learning, formulating a minimax strategy based on QoTE that accommodates the fundamental challenge of non-point-identification of the counterfactual joint distribution. Our method integrates causal inference, distributionally robust optimization, and decision theory, supporting both stochastic and deterministic policies under various identification assumptions. We establish an asymptotically tight upper bound on the regret and demonstrate robustness to model misspecification. The framework generalizes to any welfare objective defined as a functional of the potential outcomes’ joint distribution.

This paper addresses the identification of marginal policy effects (MPE) in centralized markets—specifically, how to nonparametrically assess the impact of marginal reforms on equilibrium outcomes without exogenous variation in policy rules. Method: We propose constructing “equilibrium-adjusted outcome variables” by modeling and estimating market-level equilibrium externalities, rendering these variables invariant to policy perturbations. This enables decomposition of the MPE into a covariance structure of observable variables, embedding equilibrium externalities directly into the structural outcome design and bridging naturally with the marginal treatment effect (MTE) framework. Contribution/Results: We establish theoretical identifiability of the MPE and validate the method via simulations and empirical applications. Our approach breaks from conventional policy evaluation’s reliance on exogenous policy shifts, offering the first nonparametric, intervention-free local policy evaluation paradigm for centralized markets—such as matching markets and platform mechanisms—where equilibrium externalities are inherent and policy interventions are often infeasible or unethical.

This study addresses income inequality from the perspective of equality of opportunity, distinguishing between inequality arising from circumstances beyond individual control and disparities stemming from voluntary risk-taking. To this end, it develops an axiomatic framework that first computes expected utilities within groups sharing identical environmental characteristics and then aggregates group welfare using a one-parameter opportunity-sensitive social welfare function. Integrating ambiguity decision theory, expected utility analysis, and mean-divergence decomposition, the framework introduces an opportunity stochastic dominance criterion and yields a decomposable measure of overall inequality into between-group opportunity inequality and within-group risk-related dispersion. The analysis derives multiple equivalent representations of the social welfare function, offering a practical toolkit for normative welfare evaluation and inequality decomposition.

This study addresses the lack of theoretical guidance in specifying the functional form—such as neighborhood radius—of spillover exposure measures in estimating economic policy spillover effects. Building on an experimental design framework that leverages the randomness in treatment assignment, the paper jointly identifies both the functional form and its parameters through orthogonal moment conditions. It further develops the first asymptotic theory tailored to spatial and network-dependent structures. The proposed design-based inference approach enables data-driven selection of the exposure function and is validated through two large-scale poverty alleviation programs. In these applications, several pre-specified radii are formally rejected, and adjusting the exposure specification leads to substantively different estimates of policy effects.

This study addresses the unresolved trade-off between investing in predictive capabilities and alternative policy instruments—such as capacity expansion or service quality improvements—in resource-scarce allocation settings. The authors propose an empirical framework integrating causal inference, counterfactual simulation, and welfare economics, and introduce rvp, the first operational open-source toolkit for quantifying the marginal welfare effects of prediction in resource allocation and enabling cross-context policy comparisons. The framework’s validity is demonstrated through two empirical applications: job placement services in Germany and poverty targeting in Ethiopia. Results reveal that the welfare value of prediction is highly context-dependent, offering policymakers a scalable benchmark for evaluating and prioritizing interventions under constrained resources.

This study addresses the growing concern that rapid advances in artificial intelligence (AI) exacerbate income inequality, underscoring the need to clarify AI’s economic effects and design effective redistribution policies. The authors develop a dynamic general equilibrium model that systematically distinguishes three labor types: AI-complementary workers, those displaced by AI, and workers employed solely in final-good production. They compare economic dynamics under competitive versus monopolistic AI production regimes. The analysis reveals that AI monopolies slow technological diffusion and intensify inequality. However, targeted taxation and regulatory interventions can achieve more equitable welfare outcomes across different market structures, offering a viable path toward Pareto improvements.