Impute American Housing Survey utility data for improved Section 8 modeling

[MaxGhenis]

Summary

Impute American Housing Survey (AHS) utility cost data onto CPS/ACS to improve modeling of housing assistance programs, particularly Section 8/Housing Choice Vouchers.

Motivation

Current limitation

• CPS ASEC and ACS provide total rent but lack detailed utility cost breakdowns
• Section 8 payment formula requires: Payment Standard - Utility Allowance - 30% of adjusted income
• Without knowing which utilities tenants pay separately, we cannot accurately apply utility allowances

What AHS provides that ACS/CPS don't

1. Utility cost separation: Contract rent vs. utilities paid by tenant
2. Utility itemization: Which utilities included (heat, electricity, water, gas)
3. Individual utility costs: Separate costs for each utility type
4. Housing quality: HQS-relevant data (plumbing, heating, structural issues)
5. Better geography: More precise metro area identifiers for matching FMRs
6. Bedroom precision: Detailed bedroom counts for payment standard matching

Implementation approach

Statistical matching/imputation from AHS to CPS/ACS using:

• Geographic location (metro area)
• Housing tenure (rent/own)
• Unit characteristics (structure type, bedroom count, building age)
• Household income
• Total housing costs (as validation)

Expected impact

More accurate modeling of:

• Section 8/Housing Choice Vouchers
• SNAP shelter deductions (which also separate shelter from utilities)
• LIHEAP eligibility and benefits
• State housing assistance programs

Data sources

• AHS microdata (most recent: 2021, biennial releases)
• Size: ~50k housing units nationally
• Public use files available at: https://www.census.gov/programs-surveys/ahs/data.html

References

• AHS vs. ACS discussion: PolicyEngine internal analysis
• Section 8 payment standards: https://www.hud.gov/program_offices/public_indian_housing/programs/hcv/pd/rent

🤖 Generated with Claude Code