Following the Democrats' recent win in a Republican-leaning district in upstate New York, there's been a steady drumbeat of speculation regarding the party's chance of winning back the House in 2012. Polling seems to confirm that the opportunity is there for Democrats: The RCP Average for the generic ballot shows Republicans with a slender 1.5 percent lead. I think the House is in play, but that the Democrats' road back to the majority is a steeper climb than many analysts apparently believe. There are a lot of reasons for this, but the number one reason has been receiving surprisingly little play: Democrats are getting their clocks cleaned in the 2012 redistricting.
Most redistricting analysis has focused on what you might call the "topline" number: Raw seats gained and lost. Using this metric, Stu Rothenberg estimates that Democrats are presently on pace to pick up one seat from redistricting, while Dave Wasserman of the Cook Political Report believes that Democrats will pick up two. I think Rothenberg overestimates Democratic gains in Illinois, while Wasserman overestimates Democratic gains in California, but agree overall that the topline in redistricting will be pretty close to a wash.
The topline isn't the whole story, however. As Rothenberg carefully notes (and as Wasserman has observed in interviews), there is also the question of how far Republicans are able to go in shoring up their remaining districts. By taking enough marginal districts currently held by Republicans and making them safer, Republicans can make the median congressional district -- the 218th most Republican district in the Congress -- one that is much more difficult for Democrats to win. Doing this would therefore make winning control of the House much more difficult for Democrats. This is where the "real" story is.
Let's use Ohio as an example. It is slated to lose two seats this decade due to stagnant population growth. Since Republicans control the redistricting process, we might expect them to eliminate two Democrats. Even with a 13-5 edge in the present delegation, Republicans could conceivably accomplish this. The problem is that the remainder of the state would have gone for Obama by about a 52-percent-to-47-percent margin. In other words, while you could slice this up to leave 13 Republican-leaning districts (by PVI), the median district in the state -- currently one that is roughly evenly split between Obama and McCain -- would move significantly toward the Democrats. If the national environment were to tilt even slightly toward the Democrats in a given year, the entire Republican delegation from Ohio could be wiped out. In other words, while the "topline" would show a nice Republican win in this scenario, the victory would surely become recognized as a Pyrrhic one at some point during the decade.
Now suppose that Republicans instead chose to eliminate two Republicans. Suppose they also shifted one of the Democratic districts from the Cleveland area to create a new, heavily Democratic district in central Columbus. At first blush, this would make little sense, and the topline would show a loss. But after the five Democratic districts were drawn, the remainder of the state would actually have gone for McCain by around a 54-percent-to-45-percent margin over Obama. In other words, any competent line-drawer could move the median district substantially rightward, and make it almost impossible for Democrats to expand beyond those five seats in the short-to-medium term.
The redistricting that has occurred so far looks a lot more like the second scenario above than the first one (Ohio Republicans actually look set to split the difference by eliminating both a Republican and Democratic district). The following table shows the 10 states that have either redistricted so far, or where the process is far enough along that we have a pretty good idea what the final map will look like. The most important line, of course, is the bottom line, since we don't care about the median district in any state so much as we care about the median district in the country:
Of the 163 districts in our data set (this includes single-member states where there isn't any redistricting), the median district has shifted from 52.5 percent Obama -- a touch under his national average -- to 50.6 percent Obama. In other words, if present trends continue, the median district will have barely been carried by Obama.
And it may actually get worse for Democrats. There simply aren't that many states left where Democrats can expect a favorable redistricting. Florida's "Fair Districts" law may compel the state legislature to create a few more competitive districts, and lawsuits may unravel the Texas gerrymander (on the other hand, forcing the creation of more Hispanic-majority districts in Texas would probably move the median district in that state even further rightward). Democrats control the process in only 27 of the 272 districts left to be drawn, and Democrats simply can't improve their positioning by all that much in Massachusetts, Rhode Island or Connecticut, either using the topline metric or our more nuanced metric.
In other words, once Republican legislators in North Carolina, Ohio and Pennsylvania have shown their hands, we might expect to see the median Democratic district made even more heavily Republican. I suspect the median district nationwide will move from being a little less than 52 percent Obama today to being barely more than 50 percent Obama.
There are two important, interrelated upshots to this. Under the present map, there are 60 Republicans representing districts that were carried by Obama. There will likely be many fewer such Republicans after redistricting is finished. Indeed, under the new map, it may well be impossible for Democrats to win the House without winning McCain districts. In other words, New York 26 is a huge, icing-on-the-cake win under the present map. But Democrats could have to sweep districts that are only a few points more Democratic in 2012 if they want to capture the House. That isn't an impossible task, but it is a difficult one.
More importantly, it means that the Republicans' 1.5 percent lead in the generic vote is more substantial than it looks. We may well be heading for a scenario where Democrats win the popular vote, and do so by a not-insignificant amount, but still lose the House. This hasn't happened since 1942, when Republicans won the popular vote by three points, yet narrowly lost the House. I'm willing to bet that if the House vote is as close as it was in 1996, when Republicans won the national popular vote by 77,000 votes out of 87 million cast, that they'll keep a comfortable majority.* That's what the RCP Average is pointing toward right now.
*NOTE: A few people have e-mailed about the 1996 number. The sources differ on 1996. Walter Dean Burnham and the House Clerk show the Democrats winning the popular vote in 1996 by .3 points. Ornstein, Mann and Malbin show Republicans winning by .4 points. I've taken my copy of Dubin's U.S. Congressional Elections, 1789–1997, which I consider definitive, added up the races myself, and come up with the number I shared. I suspect the difference comes from judgment calls, such as how you count votes cast for Bernie Sanders (an Independent who caucuses with Democrats), what you do about unopposed races in states that do not count votes cast in unopposed races, and how you handle open primaries conducted on Election Day.