Dynamic Local Average Treatment Effects

2024-05-02

arXiv.org

AI Summary

This paper addresses one-sided noncompliance in dynamic treatment regimes (DTRs), motivated by digital recommendation systems and adaptive clinical trials. We develop methods to identify, estimate, and conduct inference on multi-period dynamic local average treatment effects (Dynamic LATEs). We propose the first nonparametric identification framework for Dynamic LATEs, extending the Imbens–Angrist monotonicity assumption to sequential settings—accommodating single-step interventions, misaligned uptake, and other complex designs. Leveraging instrumental variables, sequential exogeneity, and the potential outcomes framework, we construct estimators that are consistent and asymptotically normal, with valid statistical inference. Our key contributions are threefold: (i) the first nonparametric identification of multi-period Dynamic LATEs; (ii) a unified treatment of temporal incentive structures and heterogeneous noncompliance; and (iii) robust, practical causal inference under realistic constraints—enabling reliable policy evaluation in dynamic, real-world settings.

Technology Category

• Reasoning under Uncertainty: Sequential Decision Making
• Knowledge Representation and Reasoning: Nonmonotonic Reasoning
• Multiagent Systems: Mechanism Design

Application Category

• User Modeling, Personalization and Recommendation: Studies of user behavior, including longitudinal effects of personalized systems
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

We consider Dynamic Treatment Regimes (DTRs) with One Sided Noncompliance that arise in applications such as digital recommendations and adaptive medical trials. These are settings where decision makers encourage individuals to take treatments over time, but adapt encouragements based on previous encouragements, treatments, states, and outcomes. Importantly, individuals may not comply with encouragements based on unobserved confounders. For settings with binary treatments and encouragements, we provide nonparametric identification, estimation, and inference for Dynamic Local Average Treatment Effects (LATEs), which are expected values of multiple time period treatment effect contrasts for the respective complier subpopulations. Under One Sided Noncompliance and sequential extensions of the assumptions in Imbens and Angrist (1994), we show that one can identify Dynamic LATEs that correspond to treating at single time steps. In Staggered Adoption settings, we show that the assumptions are sufficient to identify Dynamic LATEs for treating in multiple time periods. Moreover, this result extends to any setting where the effect of a treatment in one period is uncorrelated with the compliance event in a subsequent period.

Problem

• Identifies Dynamic LATEs under one-sided noncompliance in DTRs
• Estimates treatment effects for complier subpopulations over time
• Extends identification to staggered adoption and uncorrelated periods

Innovation

• Nonparametric identification of Dynamic LATEs
• Handles One Sided Noncompliance in DTRs
• Extends to Staggered Adoption settings