Add SPM threshold calculation module

[MaxGhenis]

Summary

Implements local SPM threshold calculation from source data. This enables proper SPM threshold calculation for district-level analysis where records may be reweighted to different geographic areas.

Problem

When reweighting CPS records to match district targets, records from high-cost areas (e.g., California) retain their original SPM thresholds even when assigned weight in low-cost areas (e.g., Iowa). This reduces the geographic variation in thresholds that should exist (~50% range from West Virginia to Hawaii).

Solution

Calculate SPM thresholds from components using the formula:

threshold = base_threshold[tenure] × equivalence_scale × geoadj

Key components:

• ce_threshold.py: Base thresholds from BLS by tenure type

  • Renter: $39,430
  • Owner with mortgage: $39,068
  • Owner without mortgage: $32,586 (2024 values)

• district_geoadj.py: Geographic adjustment from ACS median 2-bedroom rents

  • Formula: GEOADJ = (local_rent / national_rent) × 0.492 + 0.508
  • Range: ~0.84 (West Virginia) to ~1.27 (Hawaii)

• local_threshold.py: Calculate local thresholds for SPM units

  • SPM three-parameter equivalence scale (first adult=1.0, additional adults=0.5, children=0.3)
  • Vectorized calculation for all SPM units

Usage:

from policyengine_us_data.spm import (
    calculate_local_spm_thresholds,
    update_spm_thresholds_for_districts,
)

# Calculate thresholds for SPM units in specific districts
thresholds = calculate_local_spm_thresholds(
    district_codes=["0612", "5401"],  # CA-12, WV-01
    tenure_types=["renter", "renter"],
    num_adults=[2, 2],
    num_children=[2, 2],
    year=2024,
)

Test plan

• 13 unit tests pass for base thresholds, equivalence scale, and GEOADJ formula
• Integration test with district calibration workflow
• Validate against published Census SPM thresholds

🤖 Generated with Claude Code