Politics · Electoral Math · Reapportionment · May 26, 2026

Share

U-Hauls Are Rewriting the Electoral College. Quietly. Without a Vote.

Between the 2000 census and the 2020 census, Texas added eight Electoral College votes. Florida added five. North Carolina, Colorado, and Nevada each added two. The map of American presidential power tilted +13 electoral votes toward Republican-leaning states — not because of a single ballot, a single court ruling, or a single act of Congress, but because tens of millions of individual Americans loaded their possessions into U-Hauls and moved. The Census Bureau counted them at their new addresses. The apportionment formula did the rest.

The losing states are almost entirely Democratic-governed: New York shed five EVs, Pennsylvania four, Ohio four, Illinois three, Michigan three. California — for the first time since admission to the union in 1850 — lost a House seat in the 2020 census. The IRS Statistics of Income reports the same exodus in dollar terms: California lost a net $11.9 billion in adjusted gross income in a single filing year; New York lost $9.9 billion; Florida gained $20.65 billion. The IRS data is filed tax returns, not surveys. The money either showed up at a Palm Beach address or it didn’t.

The 2030 census, just four years away, is projected to compound the shift. Decision Desk HQ’s analysis is blunt: a Democrat running Biden’s 2020 coalition in 2032 no longer wins the Electoral Collegeunder the projected 2032 map. The “blue wall” of Pennsylvania, Michigan, and Wisconsin ceases to clear the 270-vote threshold once those states have shed additional seats. This article traces the causal chain — from individual moving decisions to presidential electoral power — and names every governor at every stage.

• +13net EVs to Republican-leaning statessince the 2000 census — locked in by 2020 reapportionment, in effect through the 2028 election— U.S. Census Bureau apportionment releases 2000–2020
• +8Texas EV gain32 EVs (2000) → 40 EVs (2024) — largest single-state gain in the post-2000 era— Census Bureau · National Archives
• -5New York EV loss33 EVs (2000) → 28 EVs (2024) — largest single-state loss, with Pennsylvania (-4) and Ohio (-4) close behind— Census Bureau
• +10additional projected R-leaning EVs in 2030on top of the +13 already locked in — would make 2032 the third consecutive census to redraw the map toward red states— Decision Desk HQ · American Redistricting Project

Last updated · May 26, 2026 · 1:00 PM ET

12 Gainers. 17 Losers. The Map Already Shifted.

The Electoral College has 538 votes. Each state’s share equals its number of U.S. House seats plus two senators. House seats are reallocated every ten years based on the decennial census. That allocation — “apportionment” — is the mechanical engine of Electoral College redistribution. It does not depend on who voted, who turned out, or which way the state leaned. It depends entirely on how many bodies the Census Bureau counted at addresses in each state on Census Day.

Between the 2000 and 2020 censuses, twelve states added House seats and Electoral College votes. Seventeen states lost them. The cumulative net shift toward states that voted Republican in 2024: +13 Electoral College votes. The two tables below show every gain and every loss, paired with the party affiliation of the governor in office as of May 2026.

Texas alone gained eight EVs across the period — more than every other state combined except Florida. Florida added five. Georgia added three across the same window. The Sunbelt and the Mountain West account for the bulk of the inflow. Notably, several Democratic-governed gainers (Colorado, Oregon, Washington, Nevada in part) sit alongside the Republican-governed ones, reflecting an absolute rule of apportionment: the formula doesn’t care why people moved, only that they did.

Sixteen of the seventeen losing states are concentrated in two geographic clusters: the Northeast (NY, PA, NJ, CT, MA) and the Rust Belt (OH, IL, MI, WI, IA, IN). Most are Democratic-governed. California, in a category of its own, held flat for two reapportionments before finally shedding one seat in 2020. Among the Republican-governed losers (OH, IN, LA, MO, MS, OK, WV) — these are mostly states bleeding population to the Sunbelt for the same cost-of-living and weather reasons that drive Democratic-state departures, but without the additional tax-flight overlay that the IRS data captures more sharply for CA, NY, IL, NJ, and MA.

Five Links in the Chain. None of Them Political. All of Them Mechanical.

The translation from individual moving decision to presidential electoral-college vote is a five-stage pipeline. Each stage has its own data source. Each stage is independently measurable. Each stage moves whether or not the people moving intend any political consequence. Most don’t.

The strikingly under-appreciated point: every voter who relocated has effectively cast a structural vote that lasts a full decade. A New York family that moved to Florida in 2022 doesn’t just register and vote in Florida elections from now on. Their physical presence at a Florida address on April 1, 2030 will shift one fractional unit of New York’s congressional and electoral weight to Florida for the entire decade 2032–2042. Multiplied by hundreds of thousands of similar moves, that single census morning becomes the most consequential election of the decade — and nobody voted in it.

“People are voting with their U-Hauls.”

Gov. Ron DeSantis (R-FL) · 2026 · summarizing IRS Statistics of Income data showing Florida added more income from domestic migration than any state in the country

NY -5. PA -4. OH -4. IL -3. MI -3. The Electoral Map’s Departing Class.

Five states account for nineteen of the EVs lost since 2000. Four of the five are governed by Democrats today. Each one tells its own version of the same story.

TX +8. FL +5. GA +3. NC +2. NV +2. The Sunbelt’s Quiet Conquest.

Twenty additional electoral votes have flowed into five Sunbelt and Mountain West states since 2000. The political geography is mixed — Texas, Florida, and Georgia have Republican governors; North Carolina just elected a Democratic governor in November 2024 while keeping a Republican legislature; Nevada has a Republican governor and Democratic-trending presidential electorate. The common thread is not party. It is destination.

Four Years Out. About Ten More EVs. Democrats Lose the Blue Wall as a Path.

The 2030 census takes place April 1, 2030. The apportionment release lands April 2031. The new map takes effect for the 2032 presidential election. Forecasters using the Census Bureau’s Vintage 2024 and 2025 state population estimates have converged on a similar projection — even before any controversial citizenship-counting change is factored in.

The net shift to Republican-leaning states in 2030 is forecast at approximately +10 additional EVs, on top of the +13 the 2020 census already locked in. The compounding effect matters most for the “blue wall” geometry of recent Democratic presidential campaigns: Pennsylvania, Michigan, and Wisconsin together have lost ten EVs since 2000, and another two or three are projected to peel away in 2030. That makes the historic Pennsylvania-Michigan-Wisconsin path — the same path that delivered both Obama victories and Biden’s 2020 win — increasingly insufficient by itself.

“Democrats are running out of map.”

Decision Desk HQ · 2030 Reapportionment Analysis · December 2024 · projected 2032 Electoral College map

Decision Desk HQ’s simulation: a Democrat running the geographic coalition that delivered Biden 306 electoral votes in 2020, using the projected 2032 map, falls short of 270. The same coalition produces roughly 262 EVs once the projected seat shifts are applied. A Republican running the geographic coalition that delivered Trump 312 EVs in 2024 climbs to approximately 322 EVs under the same 2032 map. Neither scenario assumes anyone changes their mind. The candidates simply inherit a different map than they ran on the cycle before.

A separate scenario layered on top — the “citizen-only” apportionment model the Trump administration has signaled interest in pursuing for the 2030 count — would compound the shift further. Under the American Redistricting Project’s citizen-only model, California alone loses an additional four seats beyond the total-population projection (six total), because of California’s very large non-citizen population. That outcome is contested and not yet legally settled — the 14th Amendment’s “persons” language is the constitutional question — but if it survives litigation, the structural Republican advantage would deepen by an additional ~10 EVs on top of the +10 already projected.

Every Tax Hike in Albany Is Now a Vote in the 2032 Electoral College.

For most of American political history, state-level tax policy was understood as a state-level decision. The federal political consequences were indirect at best: voters might punish a governor at the ballot box if state taxes felt too high. Now, the federal consequence is direct, mechanical, and lagged by ten years. A New Jersey tax increase that drives 19,370 net households to Florida and Texas in a single year — as the IRS data shows happened in 2022–2023 — also reduces New Jersey’s count in the 2030 census, which reduces its House delegation, which reduces its Electoral College vote total, which makes the Democratic presidential path through New Jersey’s 14 EVs slightly less valuable in 2032 than it was in 2024.

Two consequences follow. First: the high-tax-state political coalitions that defended high tax rates as net-positive for state services now have a structural disincentive — every departing high earner shrinks both the state’s budget and its federal political power simultaneously. Second: the no-income-tax-state coalitions in Florida, Texas, Tennessee, and Nevada have a structural incentive to keep marketing themselves to wealthy refugees. The political incentives of the two clusters are now actively pulling in opposite directions, with the Electoral College as the prize.

The Honest Caveat: Most People Don’t Move for the Reasons the Map Captures.

A fair version of the counter-argument runs as follows. Most Americans who move from California to Texas do not do so to change the Electoral College. They move for jobs, housing prices, family proximity, climate, or retirement. Many continue to vote Democratic in their new state. The Brookings Institution’s research on Sunbelt migration documents that a substantial share of new Texas and Florida residents arrive with Democratic registration histories or Democratic-leaning voting patterns. Texas has not become more Republican in registration share over the past decade — if anything, slightly less so. So the gain in seats is not the same as a gain in Republican votes.

This is true and it deserves to be stated. The headline math in this article is structural — it measures Electoral College seats, not Electoral College outcomes. A Republican wins those seats only if the Republican candidate carries those states. Texas has consistently been carried by Republicans in presidential elections since 1980; Florida has tilted reliably Republican since 2016; North Carolina has tilted modestly Republican; Nevada and Arizona have been swing states. None of this is locked in.

The qualified answer to the counter-argument: the seat shift is real, mechanical, and locked in for a full decade once a census closes. The party affiliation of those seats is not locked in. But Republican performance has been strong enough in the gainer states for long enough that the floorof Republican electoral performance has risen — and the floor of Democratic electoral performance has fallen, particularly in the Rust Belt and Northeast. The shift creates structural advantage, not deterministic outcome. The shift is also cumulative: every census builds on the prior one. A party cannot simply “win back” an electoral vote that has been reapportioned to a different state — they can only try to win that state.

The split governance pattern is itself a story. Of the twelve net-gainer states, six have Republican governors and six have Democratic governors. Of the twelve, ten have either no income tax or a single low flat-tax rate. The strongest correlate of net inbound migration is not the party of the governor — it is the tax-and-cost structure of the state. That is the result the IRS, Census, and U-Haul data all converge on.

The Census Is the Most Consequential Election Americans Don’t Vote In.

American political journalism treats presidential elections, midterm elections, and Supreme Court rulings as the agenda-setting events of the political year. The decennial census almost never gets the same coverage — it is too statistical, too distant, too lacking in personalities. And yet, by the arithmetic of this article, the 2030 census will determine the shape of every presidential election from 2032 through 2040. That is three full presidential cycles, governed by a count that takes place on a single day with no national campaign, no debate stage, no candidate.

The civic-literacy takeaway is straightforward. Americans already know voting matters. They generally do not know that moving matters — and that, in a real and durable sense, the address at which they wake up on Census Day 2030 will help decide the floor and ceiling of presidential electoral power for the following decade. Tax policy, housing affordability, crime, public-school quality, and remote-work flexibility are all inputs into that address. The output is electoral.

For the governors of the losing states, the message of the IRS and Census data is the same: the policy choices that drive residents away also drive electoral power away. For the governors of the gaining states, the message is the inverse — and the implicit dare to keep the trajectory going. Whether the next decade produces another +10 EV shift, a partial reversal, or a sharper acceleration depends entirely on what individual Americans decide to do with their U-Hauls between now and April 1, 2030.