A farmer pulls a planter through a soybean field near Buda, Ill., on July 2. (Daniel Acker/Bloomberg)

The flow goes something like this. President Trump, eager to take a hard line against China’s trade practices, enacts new tariffs on products imported from China. The costs of those new tariffs are generally borne by the people buying the products — that is, American consumers and companies. China engages with the fight, imposing tariffs of its own. Those tariffs apply heavily to agricultural products such as soybeans, reducing demand for the crops and therefore cutting sales.

Farmers aren’t happy about this. So Trump implements a program, the Market Facilitation Program, which compensates those farmers for their losses. (The program’s website at the Department of Agriculture website says that the subsidies it provides go to those affected by “unjustified foreign retaliatory tariffs,” which is a fairly subjective use of “unjustified.”) Trump, despite his rhetoric about handouts from the Democrats, hails the program and its $16 billion price tag.

“Farmers are starting to do great again!” he tweeted last month.

The politics here aren’t complicated. Farmers are almost exclusively in rural counties, places that Trump won handily in 2016. Letting farmers suffer from his trade war simply wasn’t an option for Trump.

New data obtained by the Environmental Working Group under a Freedom of Information Act request, though, show the extent to which those subsidies overlap with Trump’s base of political support.

The data details the amount of subsidies granted to every county in the United States since it began, an allocation of more than $8 billion. More than 2,600 of the country’s 3,000-plus counties received at least some money under the program; most of it centered in the Midwest.


(Philip Bump/The Washington Post)

There are far more Trump counties in the United States than counties that voted for Hillary Clinton in 2016. That many of them aren’t very populous is why Clinton won the popular vote by 2.9 million votes. In looking at the distribution of the counties and the amount each received based on how the counties voted, though, the difference is stark.

Here are Trump-voting counties that received subsidies under the MRP. You’ll notice we’ve highlighted swing counties, ones that flipped from blue to red from 2012 to 2016.


(Philip Bump/The Washington Post)

And here are Clinton-voting ones.


(Philip Bump/The Washington Post)

Overall, more than 2,300 counties that voted for Trump got at least some money under the program, nearly 9 out of 10 Trump counties. Those counties that flipped from voting for Barack Obama in 2012 to Trump in 2016 were slightly more likely to receive money than those that were red in 2012 and 2016.

About 8 in 10 Clinton-voting counties also received some subsidy.


(Philip Bump/The Washington Post)

(A quick note on methodology: The allocation of funds goes through an office in a particular county, which is how this data is attributed to particular counties. Some of the funds may then go to farmers who own land in multiple counties.)

Some of the counties receiving the most money are ones that supported Clinton in 2016. Champaign County, Ill., for example, backed Clinton by 11 points and got nearly $29 million. Of the 50 counties receiving the most money, though, 41 voted for Trump. (The chart below shows only counties that received subsidies.)


(Philip Bump/The Washington Post)

Most of those counties aren’t very populous. The circles on the chart above are scaled to the number of votes for Trump; most of the places where Trump received the most votes received little subsidy money.

Overall, Trump counties did well. About $7.6 billion of the $8.4 billion allocated to the counties above went to places that voted for Trump. About $700 million went to Clinton counties.


(Philip Bump/The Washington Post)

But again, rural counties were a lot more likely to vote for Trump. The sheer number of Trump counties means that, on average, they received about $2.6 million in subsidies. Clinton-voting ones got an average of $1.7 million each.

Counties that flipped from 2012 are at the poles. The counties that received the least on average were ones that voted for Mitt Romney in 2012 and Clinton in 2016. The counties that received the most on average? Those that went from Obama to Trump, where the average was more than $4 million each.


(Philip Bump/The Washington Post)

While Clinton won nationally by nearly 2.9 million votes, counties that received subsidies meant to offset the effects of the trade war backed Trump by 2.2 million, a 5.1 million-vote swing in Trump’s favor.

We shouldn’t confuse the direction of causality here. Trump counties received more subsidies because they are where agriculture is, and places with a lot of agriculture are more conservative and voted for Trump more heavily. But one doesn’t need to be a political scientist to wonder how a failure to provide these subsidies might affect Trump’s reelection bid. Counties receiving subsidies in Michigan, Pennsylvania and Wisconsin — the states that handed him the White House — went to Trump by 964, 513,000 and 20,000 votes respectively. In Wisconsin, that was nearly the margin of his victory. In Pennsylvania, it’s far more than he won by.

Counties in that state which didn’t receive subsidies, by the way, preferred Clinton by 470,000 votes.