Overview

Every “cheapest places to live” list compares sticker prices and ignores the thing that actually changes when you move: what your household keeps after taxes and benefits. Homestead fixes that. You enter your real situation (marital status, each partner’s income, kids and their ages, rent or buy, current county) and it computes, for any of 3,144 US counties, the genuine money picture: federal and all-50-state income tax, FICA, the marriage penalty, EITC and CTC, SNAP, Medicaid, and ACA subsidies through the PolicyEngine-US microsimulation engine, with property tax and sales tax layered on top because PolicyEngine covers neither. Everything is cost-of-living adjusted through BEA regional price parities, so the output is “real disposable income versus where you live now.”

The second half is livability. Each county gets a household-weighted score built from national percentiles over free public data: Census ACS, FEMA hazard risk, CDC PLACES health outcomes, EPA air quality and walkability, NOAA climate normals, schools, hospital access, traffic deaths. You rank plain-language priority bundles and the map recolors instantly. Scoring runs client-side against a cached percentile matrix, so there is no submit button and no waiting. County is the spine, with a city drill-down across 32,000 places and an employer-proximity filter seeded from a ~500-company HQ catalog.

The part I care most about is the honesty rules. No crime forecasting, no single-number danger ranks. Political lean and demographics are preference filters, never quality scores. Data completeness is surfaced instead of median-filled, and veto floors fail closed. The PolicyEngine runs are vectorized so all 51 state simulations finish in about three seconds. It is AGPL on purpose: anyone hosting a tool like this should be obligated to share it.

Key Features

• Real after-tax-and-benefits income for any US county, adjusted for local prices
• PolicyEngine-US for federal + 50-state tax and benefit simulation, vectorized to ~3s
• Property tax and sales tax layers built on top (PolicyEngine omits both)
• Household-weighted livability from national percentiles with ranked priority bundles
• Map-first UI: instant client-side rescoring against a cached percentile matrix
• City drill-down across 32,000 places, employer-proximity boost, state filters
• Hard honesty rules: preference filters are not quality scores, completeness is visible

Tech Stack

• Python 3.12, PolicyEngine-US, DuckDB for the county data store
• Flask + MapLibre for the map UI
• Census ACS, BEA RPP, FEMA NRI, CDC PLACES, EPA, NOAA public data