The 83 Percent County Map Counts People, Not Production

Both parties have used a mismeasured county GDP share to write off rural America.

The number is real. In 2020 Donald Trump carried 83 percent of U.S. counties; those counties produced 29 percent of gross domestic product. Joe Biden carried 17 percent of counties, producing more than 70 percent. Mark Muro and his Brookings Institution colleagues assembled the series, and the Journal’s Aaron Zitner ran it again this week. What the number does not do is what the political class has used it for.

Start with what the Brookings product actually is. The underlying data are the Bureau of Economic Analysis’s GDP-by-county estimates, part of BEA’s regional accounts. GDP is attributed to the county where the establishment that generated the output is located, not where its owner lives, not where its customers live, and not where its inputs were grown or dug. It is an establishment-location accounting exercise, and like every accounting exercise it has a boundary. Commuting flows cross county lines and are counted at the workplace. A county that hosts a corporate headquarters, a trading floor, or a federal agency captures the full value added there, whether the goods and services that value added touches were produced in the next county over or on another continent. The measure is not wrong. It answers a narrow question — where does BEA book output — and it is being cited to answer a much broader one.

The broader question has a simpler answer available in other series. BEA’s gross value added by sector shows the same national split without the geography problem. County-level payroll employment counts where the jobs are. Median household income, which BEA and the Census Bureau both publish at the county level, shows the 83 percent’s households as buyers and earners inside the same economy the 17 percent administers. Run the comparison on those measures and the “two economies” framing thins out considerably. What survives is a population-density map. Eighty-three percent of counties is roughly a sixth of the population; that is what the ratio mostly re-describes.

The time series settles it. By Brookings’ own arithmetic, Trump-supporting counties were 36 percent of GDP in 2016 and 38 percent in 2024. The 29 percent figure is a single-cycle artifact — the 2020 outcome, in the middle of a pandemic year in which the counties that host the platforms, the trading systems and the federal payroll absorbed a services boom and a goods slowdown simultaneously. The multi-cycle average sits above a third. A chart whose own author publishes the surrounding years should not be cited for one of them, and this one has been cited for one of them in every cycle since.

Now the part of the Journal’s piece that does carry weight. The same polling shows rural voters’ Republican advantage for Congress at 18 points in the Journal’s survey, against the 34-point margin AP VoteCast recorded in the 2024 House vote. Note the instruments: a Journal generic-ballot poll and a VoteCast vote-choice survey are different measurements, and a 16-point swing inferred across them is suggestive, not established. But the direction is corroborated elsewhere — the erosion of Republican economic sentiment among the party’s own voters is documented in tracking data Main Street reported earlier this week, findings that show Republican pessimism rising even as the administration holds the House map.

The mechanism is not mysterious, and it is not cultural. It is trade policy and input costs. The administration’s tariff program raised landed costs on farm equipment, on fertilizer — natural-gas intensive, and priced off the same global gas market the tariffs distort — and on the steel and aluminum that goes into grain handling and shop floors. Retaliation landed on agricultural exports. And the uncertainty is not behind anyone: this week Howard Lutnick threw the U.S.-Mexico tariff talks into further doubt by objecting to significant reductions, and Mexico is one of the largest destinations for American agricultural exports. A producer who does not know the tariff schedule in six months does not know the basis for a planting, herd or capital decision in six months. That is a policy effect on a balance sheet, not a media effect on an identity.

The consequence is visible in the seat-level map: House races in Iowa, Michigan and Wisconsin with rural precincts large enough to move, and Senate races in North Carolina, Ohio, Kansas and Nebraska. Polls showing Democrats ahead with 30 days to the vote are, at this point, measuring an anti-incumbent mood in export-exposed counties.

Which raises the question the Journal’s piece skips. If Democrats win those seats, what do they intend to pass, and who pays for it? The policy package available to a narrow Democratic majority is legible from the last Congress: making the enhanced ACA premium tax credits permanent — a cost that runs into the hundreds of billions over a ten-year window against a current-law baseline where those credits expire; extending and broadening the Inflation Reduction Act’s energy credits, which do fund rural manufacturing and grid investment but whose scored cost was contested on enactment and has since run ahead of the estimate; and tariff rollback, which lowers farm input and export costs but withdraws protection from the domestic steel, aluminum and machinery producers the tariffs were sold to. Every one of those is a distributional choice with named losers, and none of the polling shows rural voters endorsing the package. They show rural voters withdrawing consent from the current one. That is a lower bar, and a Democratic majority that reads it as a mandate will find the difference on the floor of the Senate.

The GDP share is a fine descriptive statistic and a poor political argument. The counties in the 83 percent grow, mine and manufacture; the counties in the 17 percent finance, insure, administer and platform that output, and BEA’s establishment-location convention books the second group’s contribution where the desks are. Both halves are inside one economy, which is the point the ratio exists to obscure. The policy question it should prompt is specific: which federal decisions — tariff schedules, trade-talk uncertainty, credit expiration dates — are moving against export-exposed rural producers right now, and what is the priced alternative. Those are answerable questions. The slide deck that turns a population map into a verdict on who is “the real economy” is not answering them.