Posted by Theodore Landsman on October 19, 2016
FairVote’s model of U.S. House of Representatives elections shows that House elections are so uncompetitive and skewed that it is unlikely that even a Clinton landslide would deliver Democrats the House.
Since 1998, FairVote has projected the results of upcoming congressional elections based solely on prior congressional and presidential election results. Without using any polling or campaign spending data, we were able to make high confidence projections for 80% of all congressional elections, which were accurate in 99.7% of races in 2012 and 2014 (we got one wrong). Indeed, even when we projected all 435 U.S. House seats, our accuracy rate was 96% for these two elections.
In order to emphasize that most House elections are uncompetitive regardless of Democratic or Republican waves, we model our official projections off an assumed 50-50 split in national two-party preference,which is closely related to the national popular vote, but calculated at the district level with third party and independent votes removed so as to focus on two-party partisanship. We have 362 high confidence projections (203 safe Republican seats, 159 safe Democratic seats) using data available from 2014, that remains steadfast even after considering 2016 election polling. Still, given that projections from Fivethirtyeight, The New York Times, Larry Sabato, and the Princeton Election Consortium show Clinton is headed for a landslide in the Electoral College, it is worth asking about the likelihood that this will win the Democrats the House of Representatives.
If the election were held today (10/19/2016), polls indicate that Clinton would win 53.9 of the national two-party vote, based on fivethirtyeight’s nowcast. This is impressive, and is even higher than Obama’s national two-party vote in 2008 (53.7%), which gave Democrats 58.9% of seats in the House of Representatives. This election, however, by running our model using Clinton’s projected vote share as an estimate of national party preference, we can estimate that Obama’s 2008 margin would only yield 48.3% of the House, and Clinton’s projected margin would yield just 215 (49.4%) out of 435 seats.
These values were calculated by running our model for a spread of possible two-party vote margins from 45.0% to 56.0% for the Democratic nominee, as shown to the left. As we can see, the current US House map is daunting for Democrats, who would only win 192 (44%) seats with 50% of the two-party vote, a minor improvement from the 2014 midterms, and nowhere near the 218 seats needed to regain the majority.
According to our model, Democrats would need to win the national two-party preference 54.8% to 45.2% in order to reach even a one seat majority in the House. (Emory’s Alan Abramowitz suggests it might need to be higher.) At the presidential level, this would be a landslide that has no precedent in the last half-century in a nation that is increasingly split evenly between the parties and hesitant to split their ticket. Indeed, the last non-incumbent presidential candidate to beat this margin in the popular vote was Eisenhower in 1952.
How do FairVote’s projections stack up compared to the host of political pundits currently looking at the same question? The Cook Political Report currently estimates that Democrats will gain 16 seats, bringing their total to 204, the same number our model was predicting the last time their estimate was calculated (10/14/2016). Larry Sabato's Crystal Ball estimates Democrats will gain 10 to 15 seats, bringing their total to between 198 and 203 seats, consistent with our predictions given a Clinton landslide, but not enough to win Democrats the House. Real Clear Politics shows more uncertainty, with Democrats likely to gain at least six seats and as many as 17 if they win all the races that are too close to forecast. A 17 seat shift would be a major triumph for the Democratic Party, but it would still leave them 14 seats short of a majority.
Why are House races so unresponsive to the two-party popular vote? Gerrymandering, campaign spending and incumbency advantage play a role, but the root cause is well understood: Republicans are distributed in a more geographically advantageous way than Democrats for single-winner geographic districts. Democratic voters are mostly situated in large cities with small geographic footprints, meaning Democratic candidates win their 160 or so safe districts by much higher margins than they need to, while Republican candidates win suburban and rural districts by lesser margins.
FairVote has a number of proposed reforms aimed at combatting problems with our congressional elections which can be found on our website. For more information on partisan skew and noncompetitive districts check out our Monopoly Politics series -- with updates on the 2016 cycle soon to be published.
Image Courtesy: Lawrence Jackson