SNAP Per‑Capita by Ancestry (ACS PUMS 2023)

This repo contains a Python script to compute per‑capita SNAP participation by detailed ancestry using the ACS Public Use Microdata Sample (PUMS) 2023 1‑Year files.

What it does

• Joins household records (SNAP receipt in last 12 months) to person records by SERIALNO.
• Aggregates person weights (PWGTP) by detailed ancestry codes (ANC1P/ANC2P).
• Outputs a CSV and bar charts (Top‑N and All groups) with percent labels.

Requirements

• Python 3.9+ recommended
• Packages: pandas, numpy, matplotlib
  • Install: pip install pandas numpy matplotlib

Get the data (2023 1‑Year ACS PUMS)

• Download official CSV zips from the U.S. Census Bureau:
  • Households: https://www2.census.gov/programs-surveys/acs/data/pums/2023/1-Year/csv_hus.zip
  • Persons: https://www2.census.gov/programs-surveys/acs/data/pums/2023/1-Year/csv_pus.zip

Example using curl from repo root:

• macOS/Linux

  • curl -O https://www2.census.gov/programs-surveys/acs/data/pums/2023/1-Year/csv_hus.zip
  • curl -O https://www2.census.gov/programs-surveys/acs/data/pums/2023/1-Year/csv_pus.zip
  • unzip -j csv_hus.zip 'psam_hus*.csv' -d snap_benefits/
  • unzip -j csv_pus.zip 'psam_pus*.csv' -d snap_benefits/

• Windows (PowerShell)

  • Invoke-WebRequest -Uri https://www2.census.gov/programs-surveys/acs/data/pums/2023/1-Year/csv_hus.zip -OutFile csv_hus.zip
  • Invoke-WebRequest -Uri https://www2.census.gov/programs-surveys/acs/data/pums/2023/1-Year/csv_pus.zip -OutFile csv_pus.zip
  • Expand-Archive -Path .\csv_hus.zip -DestinationPath .\snap_benefits -Force
  • Expand-Archive -Path .\csv_pus.zip -DestinationPath .\snap_benefits -Force
  • You should see four CSVs in snap_benefits\: psam_husa.csv, psam_husb.csv, psam_pusa.csv, psam_pusb.csv.

Run the script

• From repo root, either cd into the folder or call the script with paths:
  1. Change into the work dir
     • cd snap_benefits
  2. Run with auto‑labels (downloads the 2023 PUMS data dictionary to label ancestries):
     • Top‑30 plot and CSV
       • macOS/Linux: python script.py --household psam_husa.csv psam_husb.csv --person psam_pusa.csv psam_pusb.csv --min-people 50000 --topn 30 --anc-labels auto --outdir out
       • Windows (PowerShell/CMD): python script.py --household psam_husa.csv psam_husb.csv --person psam_pusa.csv psam_pusb.csv --min-people 50000 --topn 30 --anc-labels auto --outdir out
     • All groups plot and CSV (no filtering)
       • macOS/Linux: python script.py --household psam_husa.csv psam_husb.csv --person psam_pusa.csv psam_pusb.csv --min-people 0 --topn 0 --anc-labels auto --outdir out
       • Windows (PowerShell/CMD): python script.py --household psam_husa.csv psam_husb.csv --person psam_pusa.csv psam_pusb.csv --min-people 0 --topn 0 --anc-labels auto --outdir out

Installing Python packages

• macOS/Linux: python -m pip install -U pip pandas numpy matplotlib
• Windows: py -m pip install -U pip pandas numpy matplotlib (or python -m pip ...)

Windows tips

• If you see a “path too long” error, enable long paths in Windows or work in a short path like C:\work\snap.
• If python isn’t found, try py (e.g., py script.py ...).

Outputs

• CSV: out/snap_per_capita_ancestry.csv (columns: ancestry_code, ancestry_label, people, people_in_snap_hh, rate)
• Charts:
  • Top‑N: out/snap_per_capita_topN.png
  • All groups: out/snap_per_capita_all.png
  • Optional focus: out/snap_per_capita_selected.png (use --focus "Afghan,Somali,...")

Notes

• SERIALNO in 2023 PUMS is alphanumeric (e.g., 2023GQ...), and the script handles it as a string key.
• Missing household FS is treated as not receiving SNAP in the join. If you prefer to drop missing FS households, we can add a flag.
• You can adjust memory profile with --chunksize (default 500k rows).

Versioning

• This project targets the latest public data: ACS PUMS 2023 1‑Year. The 2024 1‑Year PUMS files are embargoed (not yet public) as of this writing.