Learning treatment effects while treating those in need

2024-07-10

arXiv.org

AI Summary

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.

Technology Category

• Reasoning under Uncertainty: Stochastic Optimization
• Search and Optimization: Sampling/Simulation-based Search
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Many social programs attempt to allocate scarce resources to people with the greatest need. Indeed, public services increasingly use algorithmic risk assessments motivated by this goal. However, targeting the highest-need recipients often conflicts with attempting to evaluate the causal effect of the program as a whole, as the best evaluations would be obtained by randomizing the allocation. We propose a framework to design randomized allocation rules which optimally balance targeting high-need individuals with learning treatment effects, presenting policymakers with a Pareto frontier between the two goals. We give sample complexity guarantees for the policy learning problem and provide a computationally efficient strategy to implement it. We then apply our framework to data from human services in Allegheny County, Pennsylvania. Optimized policies can substantially mitigate the tradeoff between learning and targeting. For example, it is often possible to obtain 90% of the optimal utility in targeting high-need individuals while ensuring that the average treatment effect can be estimated with less than 2 times the samples that a randomized controlled trial would require. Mechanisms for targeting public services often focus on measuring need as accurately as possible. However, our results suggest that algorithmic systems in public services can be most impactful if they incorporate program evaluation as an explicit goal alongside targeting.

Problem

• Balancing resource allocation and treatment effect evaluation
• Optimizing randomized rules for high-need targeting and learning
• Mitigating tradeoff between learning treatment effects and targeting

Innovation

• Balanced randomized allocation for learning and targeting
• Computationally efficient policy learning strategy
• Real-world evaluation with human services department