A Postal Cutback Could Sink Tens of Thousands of Midterm Ballots

FOR decades, most American post offices saw two trucks a day, one arriving in the morning with mail to deliver and another swinging back later to haul off whatever letters had piled up since. A ballot dropped at a rural post office after its only daily truck has gone now waits until the next morning to start its trip, and for a ballot mailed near the deadline, that idle day can decide whether it counts. In counties where most adults live closest to one of those one-truck offices, the share of mail ballots rejected for arriving late rose by roughly 96% in the 2026 primaries, according to a new Brookings analysis, while it held steady or fell everywhere else. Cutting the second truck was a bookkeeping decision; its side effect lands on voters.

The change is part of Regional Transportation Optimization (RTO), a Postal Service effort to run its trucks more efficiently, and it applies to post offices more than 50 miles from the plant that processes their mail. The rationale is volume (single-piece First-Class Mail fell from 57 billion pieces in 1997 to 12 billion in 2023), so by the Postal Service’s reckoning that afternoon run was carrying next to nothing.

A Day Lost Before the Ballot Even Leaves Town

Elena Patel, co-director of the Urban-Brookings Tax Policy Center, assigned every census block in the country to its nearest post office to see who actually lives beside a covered one. Roughly 121 million of the 257 million voting-age adults in the 48 contiguous states and the District of Columbia do. Among them, 49.2% live in a rural area or small town, against 8.8% of everyone else, and they skew older, poorer and more likely to have a disability. Rural America, more or less.

So how do you pull a truck schedule’s effect apart from everything else that nudges rejection rates around, from turnout to sloppy envelope signatures? Patel leaned on the fact that coverage is close to all or nothing: across California, North Carolina and Washington, 84% of counties are either under 10% covered or over 90% covered.

She split the 197 counties in those three states into two groups, those where most residents live nearest a covered office and those where most don’t, and compared late rejections across the 2022, 2024 and 2026 federal primaries, a natural experiment that economists call difference-in-differences. In the two primaries before the change the groups tracked each other closely, even though the states handle stragglers differently (California and Washington count a late ballot postmarked on time, North Carolina does not). In 2026 they came apart.

Pooled across the three states, Patel’s estimate is a 96% rise in the late-rejection rate, which works out to about one percentage point. Washington moved most in relative terms, up 142% (1.1 points), with California up 86% (0.9 points) and North Carolina up 61% (1.9 points), that last figure starting from a baseline several times higher and resting on thin data, since most of its counties returned fewer than 100 mail ballots in the primary.

Missing Signatures and Dates Did Not Budge

The obvious worry is that something else changed for rural voters between 2024 and 2026, and Patel runs two checks against it. The first swaps the outcome for ballots rejected on any other ground, a missing signature or date, say, which a truck schedule has no business touching. Those rejections moved between elections as well, but in both groups together, and the estimated effect on them is a statistically insignificant drop of 0.1 percentage points, roughly a tenth the size of the late-ballot effect and pointing the other way. The second check reaches back through five general elections, from 2016 through 2024, using the federal Election Administration and Voting Survey for 2,218 counties, and it finds the two groups’ late-rejection rates sitting within a tenth of a percentage point of each other in every single cycle, which is about as flat a before picture as this kind of analysis ever gets. Whatever else separates these places, their ballots used to show up on time at the same rate.

None of that turns the design into an experiment, and Patel says as much: covered counties were never randomly chosen, and any other shift that hit rural voters in the same window would show up in her numbers dressed as an RTO effect. The report is a Brookings analysis and has not been peer reviewed.

One wrinkle cuts the other way, though. Counties were grouped by which offices were designated for the new schedule rather than which had switched, and by primary day only 31% of designated offices in North Carolina had actually changed (against 75% in California and 98% in Washington), which ought to drag the estimate toward zero, not inflate it.

Tens of Thousands of November Ballots at Stake

Scaling up to November’s midterms means leaning on assumptions, and Patel spells hers out: that the primary effect carries over, and that it applies proportionally. Roughly 40% of the country’s mail ballots, about 16.6 million of an average 41.8 million across the 2022 and 2024 general elections, come from places whose nearest post office is covered. On that basis she projects 43,000 to 55,000 additional ballots rejected for arriving late, on top of a recent average of about 115,000. It’s a rise of 38% to 48% in the national count, from a general-election late-rejection rate nearer 0.3% than the roughly 1% seen in the primaries, and the three states were picked because they report rejection reasons in detail, not because they look like the rest of the country. The timing is awkward for other reasons too, as Patel points out: in December 2025 the Postal Service confirmed that a postmark may no longer show the day mail was sent, in June 2026 the Supreme Court left states free to count ballots arriving after Election Day provided they were postmarked on time, and a new federal ballot-mail rule has been blocked for November, which leaves the postmark (the thing that rescues a late ballot in California and Washington) doing more work just as it has become less reliable.

The remedy Patel proposes is almost boringly simple: mail the ballot a few days early, which the Postal Service already advises, at least seven days before it is due, and with Election Day on November 3, the margin for doing that is now measured in weeks. The hard part is knowing you live somewhere the change applies, because the list of covered offices sits in a regulatory filing rather than anywhere a voter would think to look, and these midterms are the first general election that will show how much that gap costs.

FAQ