Research map / Learning theory and algorithms / Combinatorial and search optimization
On the Meta-Design of Allocation Problems
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Formalizes meta-design of allocation problems as a welfare surface over design parameters (capacity, data coverage/feature collection, transfer size, utility) and gives empirical tools (gradients, MRS, KKT-based cost-optimality tests, expansion paths) illustrated on Ethiopian poverty targeting and German job-seeker profiling.
Problem
Planners treat design parameters of allocation systems (data collected, capacity, quality of intervention) as fixed, yet these are optimization variables that strongly affect welfare; there is no general framework or practical tools to analyze investments across them.
Approach
Define the welfare surface V*(theta) mapping design parameters to optimal-policy welfare, and use its gradients, marginal rates of substitution, KKT-based optimality tests against cost ratios, and budget-constrained expansion paths to guide investments; applied empirically on real data rather than stylized Gaussian models.
Key results
- In Ethiopia with step utility, welfare responds far more to capacity than to data coverage, but with cost ratio ~8x the typical 20% coverage / 5% capacity configuration is cost-efficient.
- Switching to CRRA utility (rho=3) flips the conclusion: prediction improvements have larger welfare impact and the typical configuration is no longer cost-efficient, with the expansion path starting with data collection.
- In German job-seeker profiling, most welfare gain from prediction comes from agency-history features; under assumed costs, returns to further data collection become small once the most informative groups are included.
Surprising findings
- Changing the welfare objective from step-utility to CRRA utility flips which investments matter most and reverses optimal strategy.
- In German profiling, a decision rule based on agency history alone may capture nearly all welfare-relevant signal.
Limitations
- Cost structures are assumed or illustrative for the German case study.
- Utility functions and cost structures must be specified, excluding unquantifiable considerations such as procedural fairness.
- German administrative data is sensitive and not publicly available, limiting replication.
Experimental setup
Datasets: Ethiopia Living Standards Measurement Survey 2015 Sample of Integrated Labour Market Biographies (SIAB) 2% sample
Benchmarks: Ethiopia LSMS 2015 poverty targeting simulation German SIAB long-term unemployment profiling
Metrics: number of correctly identified poor households normalized CRRA utility gain welfare gradients marginal rate of substitution budget-constrained optimal welfare
Baselines: prediction-access ratio analysis of Gaussian model
Builds on: prediction-access ratio (PAR) empirical welfare maximization proxy means test
Closest papers by meaning
- Evaluating Index-based Treatment Allocation in Underresourced Communities (AAAI 2025 · similarity 0.84)
- Policy Design in Long-run Welfare Dynamics (ICLR 2025 · similarity 0.84)
- The Limits of AI-Driven Allocation: Optimal Screening under Aleatoric Uncertainty (NeurIPS 2026 · similarity 0.83)
- The Hidden Cost of Waiting for Accurate Predictions (ICLR 2025 · similarity 0.83)
- The Relative Value of Prediction in Algorithmic Decision Making (ICML 2024 · similarity 0.83)
- Autoregressive Policy Optimization for Constrained Allocation Tasks (NeurIPS 2024 · similarity 0.82)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.