Paper finds waiting identifies exactly two risk moments per period
Original title: Waiting Buys Exactly Two Moments: Endogenous Identification in Prediction-Driven Allocation
A length-τ observation prefix identifies moments through order 2τ−1, but not necessarily allocation welfare.
The paper models a planner with a fixed budget of preventive help who must choose whom to help and when under noisy signals and risk-dependent attrition. In the canonical model, each additional observation period identifies exactly two new latent risk moments. Those moments identify average next-period attrition among survivors but need not identify welfare or attrition among people prioritized for treatment. Experiments find bootstrap averaging and early stopping improve decisions, while regret weighting does not reduce pooled mean squared regret.
- Fixed observation length leaves unavoidable timing regret at any sample size.
- Exact certificates can show whether one optimal date is shared.
- Longer observation can resolve ambiguity after the optimal date has passed.
- Code is released by KurbanIntelligenceLab.
- Organizations
- KurbanIntelligenceLab
From the original article
1 min read at arxiv.orgA planner with a fixed budget of preventive help must decide whom to help and when. Waiting reveals risk but removes intervention opportunities as higher-risk individuals leave first.
By Rama AlHamidi · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.