The 2018 U.S. Midterms: Unstable Majorities Continue

The results of the Nov. 6 midterm elections extend a level of political instability in the United States not seen since the post-Civil War era more than a century ago. Democrats won at least 34 seats to regain control of the House of Representatives by 228-199 over the Republicans, with winners not yet declared in seven additional races. However, Republicans increased their narrow Senate majority with a net gain of two seats, bringing their majority there to 52-47 over the Democrats. (Republicans will likely hold a 53rd seat when Mississippi concludes its runoff race on November 27.) This means that, when the 116th Congress is sworn in next January, a divided legislature will share control at the national level with a Republican president. As this table demonstrates, the power-sharing arrangement will be the seventh of the eight possible configurations of institutional control of the Presidency, House and Senate the U.S. has experienced since 2001.

What explains this recurring pattern of instability? It is the culmination of a long-term process of partisan sorting and polarization, in which the two major political parties have shed their more ideologically moderate members. The result is a Congress composed of two internally homogeneous parties whose respective ideological centers of gravity are moving apart. In addition to being deeply polarized, the parties are electorally quite evenly matched. This means that when either controls the Senate or House, they see little reason to compromise, and instead seek to take advantage of their brief window as the majority to pass as much of their partisan legislative agenda as possible. Witness the Republican effort, with President Trump’s active support, to roll back Obamacare, including its politically popular coverage of pre-existing illnesses during Trump’s first two years as president. Such legislative overreach elicits the predictable response by the more moderate public: it votes the offending party out of majority control. And so the cycle perpetuates.

In addition to continuing this pattern of instability, the recent midterms also perpetuated the midterm loss phenomenon. Since 1938, the president’s party has lost seats in every House midterm election save two: 1998, when Bill Clinton was fighting an unpopular impeachment effort by Republicans, and 2002, the first midterm after the 9-11 terrorist attack, in which Americans rallied to support the Republican administration. Including these exceptions, the average House loss for the president’s party across all midterms during this period is 29 seats. A similar pattern affects the Senate – on average since 1938, the President’s party has lost four seats during the midterm.

What explains the midterm loss phenomenon? Political scientists have developed three related explanations. The first is the “surge and decline” theory, which posits that, compared to a presidential election year, the midterm turnout is smaller and less likely to contain the same proportion of voters who supported the President and his party two years earlier. A related explanation suggests that midterms often serve as a referendum on the president’s accomplishments to date. From this perspective, as the newly-elected president’s “honeymoon” with the voters inevitably erodes, his approval drops and midterm voters react by voting against his party. The third explanation is that the midterm provides Americans with an opportunity to “balance” control of the major governing institutions, by giving the non-presidential party greater representation in Congress. Of these explanations, the balancing hypothesis probably carries the most weight in an era of deeply-polarized and ideologically well-sorted parties, but there is evidence that all three factors were in play during the latest midterm. At an estimated 49% of eligible voters, turnout was the highest seen in a midterm in more than a century, and much of that was driven by increases in Democratic-leaning voters, including Latinos and younger voters. Trump’s approval rating, meanwhile, which is mired in the low 40’s, also likely contributed to Republicans’ seat loss. Moreover, it is likely that the largely moderate, centrist public sought to balance Republican extremism by handing control of the House over to Democrats.

After the tempestuous 2016 election and first two years of the Trump presidency, with many pundits and even political scientists expressing alarm at Trump’s apparent willingness to break norms of presidential behavior, it is perhaps reassuring that, at least when it comes to the midterm, the conventional electoral dynamics seem still to govern outcomes. As with presidential elections, political scientists have developed forecast models that – although simple in construction – are effective at predicting aggregate House and Senate midterm seat changes. Typically, these models focus on “fundamentals” – how long the president’s party has held on to the White House, how many seats the president’s party has exposed, the president’s approval rating and some measure, such as the change in disposable income, of how voters are doing economically. Note that the most of these variables are in place long before the events that cable news pundits proclaim as “game changers”, such as the controversial nomination of Brett Kavanaugh to the Supreme Court, or the media focus on the “caravan” of immigrants heading toward the United States’ southern border. Despite slight differences in the variables utilized, all the political science models correctly predicted the Democrat House takeover, with the median forecast predicting a 30-seat gain, and the average of the forecasts at 36 seats. These were quite close to the mark. The same models forecast the Republicans making modest 1-2 seat gains in the Senate, however, primarily because of a historical quirk that found Democrats defending 26 Senate seats, including 10 that voted for Trump in 2016. This was the most Senate seats ever defended by the “out” party since direct popular election of the Senate began in 1914, and it was enough to offset the normal seat loss experienced by the president’s party.

What will the next two years of divided government bring? Already many newly-elected Democrats, reacting to pressure from their more progressive base, are threatening to launch multiple investigations of the president and his administration. This is a potentially risky strategy. With some notable exceptions, progressive Democrats did not do well in the midterms, with most of the Democrats’ gains coming by electing relatively moderate candidates, and some Democrats believe the party would be better positioned to regain the presidency and Senate if it showed it could pass a more centrist legislative agenda, perhaps by working with Republicans in areas like immigration reform and reining in health care costs. Unfortunately, recent history suggests it is more likely that the next two years will bring more partisan bickering, legislative gridlock, and deep dissatisfaction among voters. And if Democrats win the presidency in 2020, while retaining control of the House, and Republicans hold on to the Senate, the country will have cycled through every possible permutation of government control in only two decades. Contrary to the constant claim that Americans are hopelessly divided, it seems instead that a significant number share a deep conviction that both major parties are out of step with the public’s more moderate ideological and policy preferences, and that these centrist voters trust neither party enough to let them govern for very long.

2 thoughts on “The 2018 U.S. Midterms: Unstable Majorities Continue

  1. Scott Monje

    I may be missing something, but I’m not sure I see polarization per se as contributing to the constant shifts in power-sharing arrangements. The even balance between parties, yes, because it requires so few changes to bring about a shift. (The balance also contributes to the lack of cooperation because the possibility that you might win back power in the next election puts a premium on differentiation.) The relative lack of ideology on the part of the electorate as a whole, yes, because it allows people to switch their votes despite the evident ideological differentiation between the parties. But I would say that party polarization itself (and the refusal to cooperate) contributes more to the impact of shifting power-sharing arrangements on policy and on governing than it does to the shifting per se. In that respect, however, it does contribute to instability.

  2. Matthew Dickinson


    There are probably a number of factors at work, but I think the evidence behind the midterm loss phenomenon suggests that “balancing”, whether for ideological, partisan or institutional reasons, is a prime suspect, and the clear divide between the relatively more moderate (or even inconsistent) ideological views of the electorate versus the much more partisan and uniform ideological views of the political class suggest to me that at least some of the vote switching is a backlash to what the parties have to offer when in power. See, for example,


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.