Inference on Optimal Dynamic Policies via Softmax Approximation
This work addresses a fundamental issue in causal inference and dynamic decision-making by providing a method for reliable statistical inference on optimal policies, which is incremental but improves upon prior sub-sample approaches.
The paper tackles the problem of constructing confidence intervals for optimal dynamic treatment regimes from offline data, which is challenging due to non-linear and non-differentiable functionals. It shows that a softmax approximation with a fast-growing temperature parameter achieves valid inference, as demonstrated for a two-period regime with potential extension to finite horizons.
Estimating optimal dynamic policies from offline data is a fundamental problem in dynamic decision making. In the context of causal inference, the problem is known as estimating the optimal dynamic treatment regime. Even though there exists a plethora of methods for estimation, constructing confidence intervals for the value of the optimal regime and structural parameters associated with it is inherently harder, as it involves non-linear and non-differentiable functionals of unknown quantities that need to be estimated. Prior work resorted to sub-sample approaches that can deteriorate the quality of the estimate. We show that a simple soft-max approximation to the optimal treatment regime, for an appropriately fast growing temperature parameter, can achieve valid inference on the truly optimal regime. We illustrate our result for a two-period optimal dynamic regime, though our approach should directly extend to the finite horizon case. Our work combines techniques from semi-parametric inference and $g$-estimation, together with an appropriate triangular array central limit theorem, as well as a novel analysis of the asymptotic influence and asymptotic bias of softmax approximations.