Who will control the next Congress?

AMERICANS are fixated on whether Kamala Harris or Donald Trump will win the presidency on November 5th. But what the victor will actually be able to achieve depends in large part on which party controls Congress. Most new presidents have long coattails: not since George H.W. Bush was inaugurated in 1989 has one taken office without his party also controlling both chambers of Congress. But this year the race for congressional control looks as close as the presidential one, putting the normal outcome in doubt. If the election of 2024 is as much of a nail-biter as expected, the prospect of divided government—and with it the shelving of much of the new president’s agenda—looms large.

This week we launched The Economist’s forecast of the congressional elections, to complement our presidential-election model, which we first published in June and currently gives Ms Harris a 51% chance of victory. It finds that, taken individually, each chamber is more likely than not to see a change in partisan control. The Republicans’ chance of flipping the Senate is 66% and the Democrats’ of winning the House of Representatives is 61%.

However, this does not mean that the two parties are likely simply to trade places on Capitol Hill. If Republicans do manage to win the Senate, their House candidates will probably exceed expectations as well, and vice versa. After accounting for these correlations, our model finds that the two most probable outcomes remain the typical presidency-House-Senate “trifecta” for either Republicans or Democrats, with both parties having around a one-in-four shot of securing one. The next-likeliest scenarios are those with a Republican Senate and a Democratic House, either during a Harris presidency (a 16% chance) or a Trump one (15%) (see chart).

Chamber musing

The model’s most eye-catching result is probably the relatively high chance it gives the Democrats of hanging on to the Senate, given the difficult races several of their incumbents face. The party holds 51 of the 100 seats (counting four supportive independents), but is guaranteed to lose one owing to the looming retirement of Joe Manchin, a Democrat-turned-independent from heavily Republican West Virginia. As a result, it cannot afford to lose any more seats—or the presidency, since the vice-president is the tie-breaker if the chamber is evenly divided. Of the 34 Senate seats up for grabs this year, 23 are held by Democrats.

Three factors account for the model’s insistence that Democrats have a plausible path to victory in the upper chamber. One is that the party’s candidates in many races expected to be competitive—including those in places where Mr Trump and Ms Harris are neck-and-neck, such as Arizona, Michigan, Nevada, Pennsylvania and Wisconsin—enjoy surprisingly large leads in polls. If Democrats do well enough overall that Ms Harris and the two Democratic incumbents seeking re-election in red states (Jon Tester of Montana and Sherrod Brown of Ohio) all win, the party holds on in the Senate 89% of the time.

The second reason the Democrats may hold the Senate is that the model is not yet ready to write off Mr Tester. Although recent polls have put him down by 6-8 percentage points, relatively few non-partisan surveys of the race have been published. Historically, incumbents have enjoyed unusually large advantages in states with small populations, like Montana. Moreover, Mr Tester’s fundraising within the state—a good predictor of electoral performance—far exceeds that of his opponent, Tim Sheehy, an entrepreneur and political novice. If fresh polls confirm that Mr Tester is trailing badly, the model will probably lower his chance of victory from 38% to somewhere in the 20s—and possibly drag his party’s probability of winning the Senate down into that range as well.

The final reason for hope for Democrats in the Senate is that a loss in Montana or Ohio is not necessarily the end of the story. Recent surveys put the party’s challengers to Republican incumbents in the pale-red states of Florida and Texas just a few points behind. And in a scenario that seems less outlandish with every poll, Deb Fischer, a Republican incumbent in Nebraska, appears to be in a remarkably tight race against Dan Osborn, a former union leader running as an independent.

The Democrats have not fielded a candidate in the contest, and Mr Osborn has rejected their endorsement—probably a wise move in a staunchly Republican state. But he is a fan of Bernie Sanders, a left-wing senator, and our model assumes that he would support a Democrat for majority leader. The past six polls of the race have all shown results between a slim two-point lead for Ms Fischer and a stunning five-point edge for Mr Osborn. Five of them were sponsored by Mr Osborn or groups supporting him, but an independent survey also showed him trailing by a single point. Our model gives Mr Osborn a 29% chance of victory—which is 29 percentage points more than Democrats would have expected earlier this year. That, in turn, gives Democrats a large boost in our forecast. If the model ignores partisan polling from Nebraska, the Republicans’ chance of controlling the Senate rises to 70%.

Nonetheless, our model’s most confident prediction is that some Senate race, somewhere, will deliver a surprise. Large polling errors are common in Senate races. In 2020, for instance, Susan Collins of Maine, a Republican who trailed by around three points in polling averages, won re-election by ten points. There is a good chance, therefore, that a single surprising result could determine control of the chamber. Although Democrats’ probabilities of victory in Florida, Missouri, Nebraska and Texas are fairly low, ranging from 11% to 29%, the party pulls off an upset in at least one of these states in just over half of our simulations. Moreover, they do so in 47% of the scenarios in which either Mr Tester or Mr Brown loses, compensating for what would otherwise be a fatal defeat.

The contest for the House, where Republicans currently outnumber Democrats by just 220 to 212 (with three seats vacant), is even tighter. Democrats have a narrow lead of around two percentage points in “generic-ballot” surveys, which ask respondents which party they plan to support. They have also performed extremely well in special elections to replace departing House members, a strong indicator for normal elections that is not affected by polling biases. That gives them a slight edge, but one so slender that control of the lower chamber will probably come down to the specifics of particular districts.

Gerrymandering, which used to give Republicans a big advantage in the House, is becoming more of a tie. On average, our model concludes that Democrats need to win the national popular vote only by a single point to secure a majority in the House. Although Republicans recently redrew districts to thwart three Democrats in North Carolina, they are expected to lose a seat in both Alabama and Louisiana, after the Supreme Court ordered redistricting to prevent the “packing” of black voters into a single district. Moreover, Democrats are on the offensive in New York. Even though the Democratic-controlled state legislature chose to enact only a mild redistricting following the party’s disastrous showing in the midterm elections of 2022, four Republican incumbents, all running in districts that voted Democratic in the presidential election of 2020, face tough races. On average, our model expects Democrats to win 19 of the state’s 26 seats, up from 15 two years ago.

In the past, centrist House candidates have tended to fare better than other factors would suggest and ideologically extreme ones worse. But since the start of Mr Trump’s presidency, this pattern appears to have faded. At any rate, several prominent moderates are trailing in local polls, including Jared Golden, a Democrat in a rural seat in Maine that Mr Trump carried in 2020, and Don Bacon, a Republican who represents a seat in Omaha, Nebraska, that plumped for Joe Biden. Our model is sceptical that enough voters will split their tickets for either to survive this time. In contrast, the model is more sanguine about Mary Peltola of Alaska and Marie Gluesenkamp Perez of Washington, both down-to-earth Democrats in red districts who enjoy comfortable fundraising advantages.

The battle for the House is most likely to be determined in California. Although America’s most populous state is reliably Democratic overall, a non-partisan commission draws its districts, ensuring that Republicans get a fair shake. Six of the nine House races that our model calculates are most likely to deliver the decisive 218th seat for the victorious party are in California: four in the urban sprawl surrounding Los Angeles and two in the Central Valley. Of these, five are majority-minority with large Hispanic or Asian populations, and are held by Republicans. If the gains Mr Trump has made in polls with non-white voters translate into support for Republicans lower down the ballot, House Democrats will probably meet their Waterloo in deep-blue California. But if Ms Harris, who is herself a Californian of both Jamaican and South Asian descent, can claw back some of these voters en route to the presidency, Republicans will probably lose the House, too.

A president can do a lot by fiat, without Congress’s approval. And if the president’s party controls the Senate it can push through judicial nominations, even if the House is more hostile territory. Even so, the congressional toss-up compounds the uncertainty of this year’s dead-heat presidential election. ■