ZIP code demographic data: where it comes from.

Most teams that ask for "demographics by ZIP code" want three different things in one row: who lives there, what they earn, and what homes cost. No single public source provides all three, and the ZIP code itself is a less stable geography than it looks. Here is what each source gives you.

The Census Bureau

The American Community Survey is the reference for population, age, household composition, income, education, employment and housing tenure. It is free, it is documented and it is the denominator every other source is checked against. Its limits are timing and geography. The five-year estimates that are reliable at small geographies lag the present by years, and the Census does not publish by ZIP code at all. It publishes by ZIP Code Tabulation Area, which is a Census approximation of a ZIP built from blocks.

The ZIP versus ZCTA problem

A ZIP code is a postal routing device. It can be a set of carrier routes, a single building or a post office box, and its boundaries are not published. A ZCTA is a polygon the Census draws to approximate it. Most residential ZIPs map cleanly to one ZCTA. Thousands do not: PO box ZIPs have no ZCTA, some ZIPs span several ZCTAs, and new ZIPs appear between Census cycles. Any ZIP-level file built from Census data has quietly made decisions about these cases. Ask what they were.

HUD's crosswalk

The Department of Housing and Urban Development publishes a quarterly crosswalk between ZIP codes and tracts, counties and CBSAs, weighted by residential addresses. It is the standard way to roll tract-level Census data up to a ZIP honestly, and to roll ZIP-level data down to a county or metro. It is free. It is also the piece most home-built ZIP files skip.

Licensed consumer data

Compilers license household-level files that carry age, income bands, presence of children, homeownership and similar fields at the individual address, updated continuously. Aggregated to the ZIP, they fill the gap between Census releases and pick up change the survey has not caught. Their limits are that the fields are modeled or self-reported rather than surveyed, that coverage of some populations is thin, and that licensing terms restrict how the aggregates can be redistributed.

Parcel-level property data

Home values, appreciation, housing stock and owner-occupancy do not come from any survey. They come from the assessor and recorder files, aggregated. The median value of a ZIP computed from every parcel in it is a different and better number than a survey respondent's estimate of their home's worth, which is what the Census reports. Rents come from listing data, aggregated the same way. This is the layer that free sources cannot provide at all, and it is the reason a paid ZIP file exists.

Schools, crime and quality-of-life indices

School ratings come from state education departments and the rating services that aggregate them. Crime figures come from the FBI's reporting program and from local agencies, with uneven participation. Both are best expressed as indices normalized nationally, with the source and vintage retained, because the raw figures are not comparable across states. Treat them as context, not as precise measurements.

Choosing a source

If you need demographics on a Census schedule and can do the crosswalk work, the Census plus HUD is free and authoritative. If you need current household characteristics, license a consumer file and aggregate it. If you need home values, rents and housing stock, you need a parcel-level source, and there is no free one. If you need all of it in one flat row per ZIP that stays current, someone has to do the join and maintain it. That is what our neighborhood and ZIP intelligence file is, and this page is our answer to the question of what is inside it.

Questions