Where are people moving? What the parcel data shows

Migration surveys ask people where they went. Parcel data cannot ask, but it can see where somebody built a house — and a house does not get built where nobody intends to live. Across 85,061,162 American single-family homes, 4,166,289 were built since 2020, and where they landed is very unevenly distributed. All figures derived August 30, 2026.

This page also spends a section on the wrong way to answer this question, because parcel data contains a field that looks like it measures relocation and measures close to its opposite.

Where the new houses are

The share of each state’s standing single-family stock that was built between 2020 and 2026. A share rather than a count, so that Texas does not simply win for being Texas.

The spread is twelve-fold, from Idaho at 12.34% to Illinois at 0.98%. In Idaho one single-family home in eight has been built since 2020. In Illinois, Connecticut, New York and Massachusetts it is roughly one in a hundred.

Idaho at the top is the entry worth pausing on, because it is a small state that rarely leads anything. Its 72,766 new homes are a fraction of Texas’s 871,760 in absolute terms — but against a stock of 619,982 they represent a state remaking itself at a rate nowhere else matches. Texas and Florida are doing the same thing at ten times the volume and a slightly lower intensity.

The bottom of the table is not a story about desirability. Illinois, Connecticut and New York have expensive, land-constrained, heavily zoned metros where the binding limit is permission to build rather than willingness to move. A near-zero new-construction share in those states is compatible with high demand; it is a supply statistic first and a migration statistic second. That caveat runs in one direction only — Idaho building at 12% is unambiguous evidence that people are arriving, while Illinois building at 1% is ambiguous evidence about anything.

The fifteen counties absorbing the growth

Counties with at least 100,000 single-family homes, ranked by the share built since 2020.

Eight of the fifteen are Florida, four are Texas. St. Johns County — the Jacksonville suburbs — leads at 25.04%: a quarter of every single-family home in the county is less than seven years old.

The pattern underneath the list is that almost none of these are the metros themselves. Williamson and Collin and Denton are the outer rings of Austin and Dallas; Pinal is the corridor between Phoenix and Tucson; Osceola and Polk and Pasco are the collar counties of Orlando and Tampa. Horry County is Myrtle Beach, which is its own category. Growth is landing one ring out from where the jobs are, which is the same place it landed in the 2000s.

Why the obvious filter finds the wrong people

Anyone building a “recent movers” list from property data arrives quickly at a field called out-of-state owner, which compares the owner’s mailing address to the state the property sits in. It looks exactly like a relocation flag. It is not, and the mistake is worth setting out in full because it is close to universal.

A household that moves from Chicago to Dallas and lives in the house changes its mailing address to the house. It reads as an in-state owner from the day it arrives. What the flag actually finds is the people who did not move: absentee landlords, second-home owners, and out-of-state investors.

The numbers make it unambiguous. 3,996,396 single-family homes — 4.70% — have an owner whose mail goes to another state. Of those, only 308,465 are also flagged owner-occupied, which is 7.7%. Out-of-state ownership is absentee ownership almost by definition.

Combine the two filters the way an intuitive search would — recently sold and out-of-state owner, hoping for recent arrivals — and the result collapses. Of the 3,665,576 single-family homes that changed hands between August 2025 and August 2026, just 55,286 are both owner-occupied and out-of-state mailing: 1.51%. Most of those are people who have not yet updated their address with the assessor. It is a list of paperwork lag, not a list of new neighbours.

The signal that does work is far duller. Of those 3.67 million recent sales, 2,544,602 — 69.4% — are owner-occupied. Somebody moved into each of them. Recent sale plus owner-occupancy is the whole of the move-in signal that parcel data honestly supports; the out-of-state field belongs in a different query entirely, one about investors.

Turnover: 4.3% of the country moved house

3,665,576 single-family homes recorded a sale in the trailing twelve months, about 4.3% of the national stock. Set against the median tenure of 9.8 years that we measure elsewhere, it is a slow market: a stock turning over at 4.3% a year implies a far longer average holding period than American homeowners historically had, and it is the arithmetic behind every story about people being locked into low mortgage rates.

The two measures on this page answer different halves of the question. Turnover says how many households moved. New construction says where they could go. In Idaho and the Florida collar counties the second number is doing the work; in Illinois and Connecticut, whatever household formation exists is competing over a housing stock that is 1% newer than it was six years ago.

Methodology

• Source and cohort. County assessor and recorder data on Scout Data’s national parcel table, one row per property. 85,061,162 single-family homes; condominiums, multi-family, mobile homes and land excluded.
• New construction is a recorded build year of 2020 through 2026, expressed as a share of homes in that geography that carry any build year — never as a share of all parcels, which would penalise counties with poorer assessor coverage.
• Coverage gate. States need at least 200,000 single-family homes and at least 85% build-year coverage; counties need at least 100,000 homes and the same coverage floor. Seven states fall below it and are excluded: Wisconsin, Louisiana, Maine, South Dakota, Arkansas, Kentucky and Michigan.
• County names are normalised before grouping. The county field is mixed case within the same column, so grouping on the raw value splits a county in two. The first cut of this table had Pasco County, Florida at 24.9% because it was computed on 105,697 of its 203,779 homes; merged, it is 17.71% and four places lower. Every geography here is grouped case-insensitively.
• Sales are the most recent recorded sale date falling between August 1, 2025 and August 30, 2026. Recording lag means the most recent weeks are incomplete, so the 4.3% turnover figure is a floor.
• Out-of-state ownership compares the owner’s mailing state to the property’s state. It is reported here as a measure of absentee ownership, which is what it is. It is not a relocation signal and this page does not use it as one.
• What this is not. New construction is a proxy for in-migration, not a measurement of it. It undercounts growth absorbed by existing housing and by apartments, and a low share can mean constrained supply rather than absent demand — which is why the text reads the top of the state table with more confidence than the bottom. For household-level migration counts, census survey data remains the right source.

Derived August 30, 2026. Spot something that looks wrong? Tell us — we’d rather correct a number than defend it.

Companion pages: how old American homes are puts the 2020s construction in the context of every decade before it, how long people stay in their homes covers tenure and turnover in depth, corporate ownership of single-family homes covers who the absentee owners actually are, and Lead Finder opportunities covers building move-in cohorts as a repeatable query.

Frequently asked questions