Yesterday at the Washington Post’s Fix blog, Aaron Blake observed that the Democratic Party appears likely to have won more raw votes in U.S. House of Representatives races, and yet fell well short of winning a majority of the seats. Redistricting at work? Blake thinks so:
The numbers seem to back up what we’ve been talking about on this blog for a while: Redistricting drew such a GOP-friendly map that, in a neutral environment, Republicans have an inherent advantage.
But partisan redistricting isn’t the only source of bias in the translation of votes to seats. Consider a paper by Jowei Chen and Jonathan Rodden:
We show that in many states, Democrats are inefficiently concentrated in large cities and smaller industrial agglomerations such that they can expect to win fewer than 50 percent of the seats when they win 50 percent of the votes. To measure this “unintentional gerrymandering,” we use automated districting simulations based on precinct-level 2000 presidential election results in several states. Our results illustrate a strong relationship between the geographic concentration of Democratic voters and electoral bias favoring Republicans.
It’s not hard to find districts in which the Democratic House candidate takes more than 85% of the vote, even in states like California or New York where the GOP had no advantages in drawing district boundaries. But it’s much harder to find contested races where the GOP candidate wins by those margins, suggesting that the Republican House vote is distributed more efficiently across space. It’s not just about who draws the districts, but also about where Democrats and Republicans live in the first place. Much more is in their paper.
{ 7 comments… read them below or add one }
Very interesting, but I think the framing of the finding in the final sentence (“It’s not just about who draws the districts, but also about where Democrats and Republicans live in the first place”) is slightly off. The way the sentence is written suggests that this mismatch stems from ‘social’ factors that are mainly disconnected from politics (e.g. housing preferences) and not for political reasons (e.g. partisan redistricting). However, there would seem to be a political reason behind this mismatch that goes beyond redistricting: the decision by Congress to limit the size of the House rather than to continue growing with the country itself. Were the House to expand, some of these populous Democratic districts would presumably be subdivided, leading to an increase in Democratic representatives and better matching the overall vote. It’s interesting that reforming this aspect of the House never gets brought up when discussing various institutional reforms (adding PR or DC as a state, for instance).
Josh R., I am afraid you’re mistaken. The House seats are evenly apportioned by population. Adding more would mean more rural and suburban Republican districts as well as more urban Democratic districts. The situation works like this:
Suppose a state has four congressional districts, and it is 50% Democratic and 50% Republican. Then they should elect 2 Dems and 2 Reps, right? Or they should have four very competitive seats:
Scenerio 1:
District 1: 50% Dem
D2: 50% Dem
D3: 50% Dem
D4: 50% Dem
–>All tossups; this election leaned Democratic so we’d expect narrow Dem wins in 3 or 4 of those districts, but possibly less.
Scenario 2:
D1: 70% R, 30% D
D2: 70% R, 30% D
D3: 30% R, 70% D
D4: 30% R, 70% D
–>The Reps will always win 1 & 2 and the Dems will always win 3 & 4
But there are tons of other ways the votes can be distributed, such as:
Scenario 3:
D1: 65% R, 35% D
D2: 65% R, 35% D
D3: 60% R, 40% D
D4: 90% D, 10% R
–>Now you have 3 safe Republican seats and only one safe Democratic seat, even though the state overall is exactly 50-50. The Democratic district is really really really safe, but that doesn’t help anyone: that Congressman still only gets 1 vote.
If every state were like scenario 3, then the Republicans would win a 75% supermajority in the House in every election. In order to pick up District 3, the Dems would have to win by more than 20 points statewide, which is extremely rare. It has nothing to do with the size of the districts, just in how people are distributed within them.
Dan’s post is that the United States overall is set up like a milder version of scenario 3. The only way to fix this imbalance is switching to proportional representation.
I would deny that “The only way to fix this imbalance is by switching to proportional representation” is true. It should always be possible to “reverse gerrymander” even geographical areas that are monolithically one party or the other, if one wants to. For example, most people in AZ live in Maricopa County. Of the 9 congressional districts in AZ, 8 of them include some portion of Maricopa County (the other one has a portion of Tucson, the 2nd biggest city in the state). This year, in those 8 districts, 4 have gone R, 3 D, and the eighth one is leaning D. (While AZ2, the district with no Maricopa County voters,is currently 36 voters apart with 250,000 counted!).
Many folk may not want that kind of reverse gerrymandering – do Queens Democrats each want to have a long thin strip of Long Island added to their district to make it more Republican? Peter King (R-Long Island), certainly doesn’t – but that doesn’t mean it couldn’t be done. And the results wouldn’t look any crazier than the NC districts that are currently gerrymandered into long thin strips to maximize the percentage of Democratic party members packed into them.
Yeah, you can do a very aggressive gerrymander to solve this problem. I’m hoping that Democratic states will start doing this more. But the deck’s stacked against Democrats for three reasons: 1) the Republicans are more aggressive about this seemingly by their nature 2) the Dems are packed by default as discussed here, and 3) the majority-minority districts, designed to elect minority representatives are by definition lopsided districts that hurt the Democrats overall, a 3a) lots of minority members actually like those minority districts, and will fight for them, because they want to guarantee those areas elect black/Hispanic reps.
In any case, I really meant PR is necessary to avoid any nationwide structural disadvantage for either side. Even if you can overcome Democrats’ reluctance to do high-quality gerrymandering (and here in CA it’s basically impossible–a nonpartisan commission is now giving us districts that elect more Dems than our Dem-dominated legislature was willing to make, because the dingbats wouldn’t give up their 30 point margins!), if you leave the dominant party in each state to set the districts the nationwide sum will end up skewed one way or another.
I must admit I haven’t read the new paper, but isn’t this also explained in Rodden 2006? The idea I thought was that the process of industrialization leads to geographically concentrated areas of left-wing voters, while right-wing voters are less geographically concentrated. This creates an advantage for right-wing parties in SMDP style electoral systems like the US, but no advantage, and possibly a left-wing advantage, in PR systems.
Matt, it is the same argument but tested in a really cool way with redistricting simulations.
Interestingly, the House disadvantage faced by democrats reverts in the electoral college. Nate Silver wrote: “The problem for Republicans is that in states like these, and others like Tennessee, Kentucky and Arkansas, they are now winning by such large margins there that their vote is distributed inefficiently in terms of the Electoral College”.
You can go too far with the argument about unintentional gerrymandering. It is certainly true that many “packed” districts are naturally occurring — such as districts 1 and 2 in Philadelphia. However, that was as true in 2008 and 2010 as in 2012. And it is clear that something has changed. Previously when the Democrats won the raw House vote in Pennsylvania, they got a majority of the seats. Now they get 5 seats out of 18. Same story in Ohio and Michigan.
What did change was Vieth v. Jubelirer (2004) in which the Supreme Court decided that partisan gerrymandering was not judiciable. The previous standards for judicial gerrymandering (Davis v. Bandemer 1986) were ambiguous, but at least provided some deterrence to obvious gerrymandering. This is the first election with the post Vieth districts, where there is no threat of judicial recourse.