Primary Submission Category: Dynamic Treatment Regimes
Dynamic Local Average Treatment Effects
Authors: Ravi Sojitra, Vasilis Syrgkanis,
Presenting Author: Ravi Sojitra*
We enable identification, estimation, and inference for Local Average Treatment Effect (LATE) estimands in multiple time period settings with noncompliance and allow for treatment dynamics. Dynamics occur when treatment is encouraged in each time period depending on previous encouragements, treatments, and states (e.g. short term outcomes and time varying covariates). Although one may hope to leverage estimates of the effects of switching from one sequence of treatments to another, such quantities are not estimable under noncompliance and standard identifying assumptions for dynamic settings and LATEs. We introduce two conditions to enable identification of Dynamic LATEs that quantify effects for subpopulations who would comply with encouragements. First, we show that One Sided Noncompliance enables identification of all Dynamic LATEs corresponding to treating in a single period only. Second, further adding Staggered Adoption enables identification of effects of treating in multiple time periods. In general, the second result holds when the treatment effect of not continuing to comply is uncorrelated with whether one continues complying. We also show that a sequential extension of Monotonicity in Imbens and Angrist (1994) is not sufficient for the first result and an additional assumption is necessary for the second result. Finally, we use the automatic debiased machine learning framework to perform plug-in estimation and inference based on our identification results.