Prioritize reported recipients in take-up assignment

[MaxGhenis]

Summary

• Fix take-up assignment for SSI and SNAP to prioritize reported recipients
• Anyone with nonzero reported benefits is marked as taking up
• Adjusted rate for non-reporters calculated to hit aggregate target
• Preserves overall take-up rate while improving household-level accuracy

Problem

Previously, take-up was assigned randomly across all eligible units at the aggregate take-up rate. This meant that:

• Households who reported receiving benefits (and clearly do take them up) were sometimes randomly marked as not taking up
• Non-reporters were randomly marked as taking up

This disconnect between reported receipt and simulated take-up led to inaccurate household-level income, which matters for poverty measurement and distributional analysis.

Solution

For each benefit (SSI, SNAP), the new approach:

1. Marks all reported recipients as taking up (takes_up = True)
2. Calculates how many additional take-ups are needed among non-reporters to hit the aggregate target
3. Assigns take-up to non-reporters at the adjusted rate

This ensures:

• Aggregate take-up rates are preserved
• Household-level income better reflects who actually receives benefits
• Poverty rates are more accurate because the right households receive the right benefits

Benefits affected

• SSI (ssi_reported at person level, takes_up_ssi_if_eligible)
• SNAP (snap_reported at SPM unit level, takes_up_snap_if_eligible)

Test plan

• Verify tests pass (pre-existing failures unrelated to this change)
• Verify aggregate take-up rates are preserved
• Verify all reported recipients have takes_up = True

Generated with Claude Code