The Domestic Migration Ledger
Between 2021 and 2024, which states did Americans already living in the United States move into, and which did they move out of — measured per 1,000 residents, with international arrivals set aside?
What this ranking does not measure
This measures net moves per resident. It does not measure whether a state is a good place to live, and nothing here explains why anyone moved. Six specific limits. (1) NET, NOT GROSS: a state where 200,000 people arrive and 200,000 leave scores identically to one where nobody moves at all. This file carries no gross in- and out-flows, so churn is invisible and the comparison cannot be made. (2) A RATE FAVOURS SMALL STATES: Idaho ranks 1st on +111,459 people; California ranks 50th on -1,397,897. California lost more than twelve times as many people as Idaho gained, and the rate inverts that entirely. Read the counts alongside the ranks. (3) THE DENOMINATOR IS THE ENDPOINT: per1k divides by the 2024 population, which already contains the people who arrived during the window, so gains are divided by a larger base and losses by a smaller one. This is the file's own convention; I did not change it, but it is not a neutral choice. (4) THE FOUR-YEAR SUM HIDES ITS OWN TREND: Idaho's annual net domestic migration ran 51,978 then 28,368 then 14,730 then 16,383 — a fading flow inside a first-place total. Montana's ran 20,539 down to 5,410. New York's ran -300,058 up to -120,917 — an outflow shrinking by 60% inside a last-place total. The rank is a four-year accumulation, not a description of where things stand in 2024. (5) IT IS NOT POPULATION GROWTH: excluding international arrivals is the point of this ranking, but it means 19 of the 25 jurisdictions with negative domestic migration nevertheless grew in total population over the window — Washington, D.C. is 45th here at -20.2 per 1,000 domestic while gaining 31,727 residents overall on an international rate of +49.3 per 1,000; New Jersey is 44th while gaining 228,446. Rank position is not a growth ranking and must not be read as one. (6) NO CAUSATION AND NO CHARACTERISTICS: the file records how many people moved, not their age, income, housing situation, employer or reason, and nothing in a 51-jurisdiction count can establish that any state policy, tax, price or climate produced any of these flows.
Highest
- 1Idaho+55.7 per 1,000 (+111,459 people)
- 2South Carolina+54.7 per 1,000 (+299,962 people)
- 3Montana+44.6 per 1,000 (+50,720 people)
- 4Delaware+41.9 per 1,000 (+44,125 people)
- 5North Carolina+34.8 per 1,000 (+383,851 people)
- 6Florida+34.7 per 1,000 (+810,122 people)
- 7Maine+33.7 per 1,000 (+47,322 people)
- 8Tennessee+32.8 per 1,000 (+237,368 people)
- 9Arizona+29.3 per 1,000 (+222,465 people)
- 10South Dakota+22.8 per 1,000 (+21,045 people)
Lowest
- 47Louisiana-27.1 per 1,000 (-124,444 people)
- 48Illinois-31.1 per 1,000 (-395,721 people)
- 49Hawaii-31.8 per 1,000 (-46,018 people)
- 50California-35.5 per 1,000 (-1,397,897 people)
- 51New York-45.0 per 1,000 (-893,537 people)
All fifty-one jurisdictions carry this measure; none was excluded.
How it was computed
For each of the 51 jurisdictions I read migration.per1k.domestic from states.json and sorted descending. That field is the Census net domestic migration summed across the four annual rows in migration.years[] (2021, 2022, 2023, 2024), divided by the state's 2024 population estimate, times 1,000. I verified both halves of that arithmetic in-session across all 51: sum(migration.years[].domestic) equals migration.totals.domestic for every jurisdiction with zero exceptions, and migration.totals.domestic / population * 1000 reproduces migration.per1k.domestic to within 1e-6 for every jurisdiction. Worked example, Idaho: 111,459 net domestic movers / 2,001,619 residents * 1,000 = 55.684, displayed as +55.7 per 1,000. Worked example, New York: -893,537 / 19,867,248 * 1,000 = -44.975, displayed as -45.0 per 1,000. Displayed values are rounded to one decimal; ranks are assigned on the unrounded figure (Rhode Island at -7.424 and Kansas at -7.420 both display as -7.4 but hold distinct ranks 39 and 38). Twenty-six jurisdictions are positive, twenty-five negative. No jurisdiction has a null in this field, so all 51 are ranked and none were excluded.