We are still more than a year away from the 2018 election. With both parties still regrouping in the aftermath of 2016, and deeply divided, it is an open question what the major issues, messages and scandals will be. However, there is one thing we can already say with near certainty. The number of seats won by each party's candidates will not track the number of votes they received. In 2018, Democrats will win a much smaller number of House seats than their vote share suggests, up to 28 fewer seats than proportionality, making it plausible that Republicans will continue to control the House even if a majority of the votes cast go to Democratic candidates.
Bias in American elections is not uniquely in favor of Republicans. House elections showed an incumbency driven bias towards Democrats for decades, that allowed the party to maintain control of the House for 40 years from 1955 to 1995. However, the unique combination of hyper-polarization, gerrymandering and sorting make this a particularly perilous moment for anti-democratic skew in what is intended to be America's most representative body, and make it all the more necessary to implement reform now.
In the graphic below, FairVote models the relationship between America’s Democratic partisanship, represented by the National Party Preference for Democrats, and the number of U.S. House of Representatives’ seats associated with each preference level.
This graph maps the National Party Preference for Democrats on the x-axis against Democratic seats in the US House on the y-axis. The blue line represents the number of seats needed to obtain a majority, while the light purple line represents the ideal proportional, a 1% increase in House seats for Democrats for every 1% of the vote they earn.
The orange line corresponds to the number of seats that the Democratic Party stands to gain or lose under the current single winner system. The relationship between vote share and number of seats was calculated using our Monopoly Politics model. That model uses the partisanship of each district as well as a measure of the incumbent’s strength to project the results of the 2018 election. The model has proven remarkably accurate. Before the 2016 election, it identified 361 (83%) of districts safe enough to be projected, and every single such projection was correct.
The Monopoly Politics model also allows for projections based on a uniform national partisan swing and incumbency bump. That is, some years the nation as a whole leans more toward the Democrats and other years it leans more toward the Republicans, and some years the country as a whole feels more or less favorably about their current representatives (the incumbency bump has been declining since 2000). These factors tend to affect congressional districts in a more-or-less uniform way. Monopoly Politics can account for that. For example, 2014 was about a 52% Republican year with a 2.8% incumbency bump. Based on those numbers, the Monopoly Politics model would have projected about 246 seats for Republicans. In 2014, Republicans actually won 247 seats, meaning the model was 99% accurate.
This year, the model predicts that Democrats will not win a majority of seats in the House of Representatives unless 2018 is at least a 55.5% Democratic year. For context, Democrats won about 55.5% of the national congressional vote in 2008, while Republicans won about 53.4% of the congressional vote in 2010.
The dark purple line corresponds to the number of seats that the Democratic Party would gain under the multi-winner system proposed in the Fair Rep Act in our model plan for Fair Representation, created by Auto-Redistrict. The Fair Representation Act is FairVote’s solution to structural problems in the House of Representatives, and a blueprint for how to resolve fundamental issues in American Democracy. This year, we released for the first time a new Monopoly Politics super districts model for the Fair Representation Act, applying the strength of the Monopoly Politics model to a simulation of multi-winner districts with ranked choice voting. The line shows how these maps would perform under various national partisan levels; it does not incorporate the “incumbency bump,” since it is not using existing districts.
Auto-Redistrict generated a sample multi-winner district map for every state, using computer algorithms intended to roughly simulate the approach that would be taken by independent redistricting commissions under the Fair Representation Act. The maps it drew are not the actual maps that would go into effect, but they allow us to make some projections about what kind of impact the Act might have.
As we can see from the purple line in the above graph, our model Fair Rep Act plan would resolve almost completely the remarkable divide between Democratic Party vote share and political preference. Not only would this divide be minimized, but where it exists, it would be symmetrical, with Democrats winning a slightly higher share of seats than proportionality would dictate once they cross 53% of the vote, bringing the Congressional map into accordance with new standards in gerrymandering evaluation such as the efficiency gap. The Fair Rep Act would provide a higher level of proportionality, and more directly translate support into representation in the U.S. House, the house of the people. Our research shows it would cut wasted votes to half or even a third of their current value in most districts.
A higher level of Americans would have their interests represented and the monopoly politics of 1 party dominated districts would be replaced by democratic, meaningful popular choice. FairVote and Representative Don Beyer introduced the Fair Representation Act on June 26th, 2017, it’s current sponsors are Don Beyer, Ro Khanna, Jamie Raskin, and Jim Cooper. Further information can be found on our website.