New Benchmark for Simulating Second-Order Policy Effects
A new benchmark for simulating second-order policy effects has been detailed in an arXiv paper identified by ID 2608.15101. It features 96 public-policy cases across eight domains and four action classes: implement, modify, pilot, and block. The benchmark assesses various state variables, including benefit, capture, gaming, and implementation capacity. Notably, the simulator operates independently of the expert action target, allowing for regeneration of method outputs and aggregated results from the case table. Each case in the benchmark is also linked to its sources, enhancing its reliability and applicability.
Key facts
- The benchmark is presented in an arXiv paper with ID 2608.15101.
- It contains 96 named public-policy cases.
- Cases span eight domains and four action classes: implement, modify, pilot, block.
- State variables include benefit, capture, gaming, burden shift, instability, uncertainty, irreversibility, distributional risk, and implementation capacity.
- The runner regenerates method outputs and aggregate results from the case table.
- The simulator never reads the expert action target.
- The paper formalizes second-order policy-effect prediction.
- The benchmark is source-linked, with each case including source locators.
Entities
Institutions
- arXiv