The Border Tax

A ZIP code is a mail-delivery route. A municipality is a government. Americans treat the two as the same thing, and for most of the country they roughly are — but not everywhere.

We analyzed the U.S. Census Bureau's own boundary files for the 20 states where townships and boroughs are real municipal governments, looking for ZIP codes meaningfully divided between two of them. We found 2,230 such ZIP codes, containing 9,746,398 homes. Under 2020 boundaries, 2,229 of the 2,230 are still split the same way — these are not historical artifacts.

Why the smaller side is the one that gets hurt

When a ZIP is split, one municipality usually dominates it — and the postal city name tends to follow the dominant town. That leaves 3,562,780 homes whose mailing address names a town they do not legally live in, pay taxes to, or send their children to school in.

This is where automated valuation models break. An automated estimate is only as good as the sales it treats as comparable, and models select comparables partly by location labels. In a split ZIP those labels disagree across data sources — so a model can quietly build a home's value out of the other town's sales, with different tax rates, a different school district, and often a genuinely different market.

The homes most exposed are the ones on the minority side: surrounded, in the data, by a town that isn't theirs.

Where the split homes are

Twenty states qualify because their county subdivisions are functioning municipal governments. In much of the South and West, the equivalent unit is statistical rather than governmental, so the same ambiguity does not arise in the same form.

The sharpest cases: ZIP codes cut nearly in half

In 287 of these ZIP codes, neither municipality holds even 60% of the homes. These are the addresses where the postal city is close to a coin flip:

What this costs a homeowner

Three things follow the legal municipality and not the mailing label: property taxes, school district assignment, and municipal services. Two homes with identical mailing addresses in the same ZIP can pay different tax rates and attend different schools.

The fourth thing is the estimate. Zillow publishes its own accuracy figures, and even nationally its off-market Zestimates carry a median error in the mid-single-digit percentages — before you introduce a boundary that makes the comp set ambiguous. On a median-priced home, a few percentage points is tens of thousands of dollars of anchor: the number a buyer sees first, the number an agent argues against, the number that shapes what the house feels like it's worth.

We are careful about what this study does and does not show. It measures exposure — how many homes sit where the boundary problem exists. It does not claim a dollar figure for mispricing, because no public dataset supports one. Anyone who quotes you a precise national number for that is guessing.

How to check your own address

1. Look at your property tax bill. The municipality that taxes you is the municipality you legally live in — that is the definitive answer, whatever the mail says.
2. Or look up your address in your county's parcel records, which list the taxing municipality for every lot.
3. Or run the free audit below: it resolves your address against official Census boundary records, tells you which side of the line you're on, and runs the comparison against same-municipality sales only.

Methodology

Housing-unit shares come from the Census Bureau's 2010 ZCTA-to-County-Subdivision relationship file — the only file Census publishes that divides housing units across the ZIP/municipality overlap. A ZIP counts as split when its top two municipalities each hold at least 20% of its housing units and at least 300 units each.

Municipality names and legal status come from the 2020 Gazetteer; only active governments are counted, so statistical census county divisions are excluded. The still-split check uses the 2020 ZCTA-to-County-Subdivision relationship file, confirming the ZIP still overlaps both municipalities under current boundaries.

Two caveats we would rather state than have pointed out: housing-unit shares are 2010 figures, because Census has not published a housing-unit-level equivalent since, so shares have shifted at the margins even though the boundaries have not. And these are ZCTAs — the Census Bureau's statistical approximation of ZIP codes — which is the standard basis for this kind of analysis but is not byte-identical to USPS delivery routes.

Reuse: journalists and researchers are welcome to use these figures with attribution to Level (thetruelevel.com). The underlying files are public; our derived dataset is available on request.