Redistricting season is upon us again. Politicians and interest groups are pouring over proposed and finalized maps, and pundits are trying to keep score. How many seats will the Democrats pick up in California? How many will they lose in Missouri?
More important than score-keeping, however, is whether the composition of the legislature reflects the partisanship of the electorate. Will a party that wins 50 percent of the votes get 50 percent of the seats?
In most states the answer is no. Republicans can expect a sizable advantage, and not because of gerrymandering.
Consider the recently adopted districting plans in two of the states most hotly contested in the 2008 presidential election: Indiana and Missouri. In Indiana, if we overlay the new plan on precinct-level election returns, we find that Obama would have obtained a majority in only two of nine congressional districts (representing Indianapolis and Gary), in spite of winning the statewide vote. In Missouri, Obama would have won only two of eight districts (representing St. Louis and Kansas City) with almost half of the statewide vote.
In many large industrial states such as Michigan, Ohio, and Pennsylvania, where presidential and statewide elections are extremely close, a similar pattern emerges. Despite being home to roughly equal numbers of Democrats and Republicans, Republicans hold comfortable majorities in Congressional delegations and state legislatures.
Why? To frustrated Democrats, the answer seems obvious: Republicans drew favorable maps that packed Democrats into a few urban districts. GOP Congressional gains are a product of timely victories in state legislatures prior to redistricting. Moreover, Republican efforts to “pack” Democrats into homogeneous districts are enhanced by the efforts of minority-representation advocates who often join forces with Republican cartographers.
For many Democratic activists, the solution, too, seems obvious: take redistricting power away from legislators and give it to independent agencies required to draw natural districts that do not favor any party or racial group.
But Republican cartographers and minority advocates are only a small part of the problem in many states. The larger problem is the enduring legacy of American industrial development, which distributes Democratic voters in a way that is highly inefficient for Democratic candidates.
Since the New Deal, Democrats have tended to live in dense urban centers as well as smaller agglomerations (including college towns) that are spread out along the railroad tracks, rivers, canals, and lakes where industry and labor unions gathered strength in the late nineteenth and early twentieth centuries. Republicans tend to live in lower-density suburbs and exurbs surrounding these agglomerations as well as in the rural periphery.
In virtually every American city, the Democratic vote share is extremely high—often above 90 percent—in the nineteenth-century urban core. It declines gradually moving into the older suburbs, continues to decline through the newer McMansion suburbs, and then levels off in the solidly Republican countryside. This pattern characterizes not only large cities, but also smaller ones such as the Indiana cities of Anderson, Muncie, and Fort Wayne, which formed in the nineteenth century along the critical transportation routes of the time. Strings of Democrats are thus spread throughout the hinterland, such as along the Illinois and Michigan Canal and Illinois River and along railroads connecting Detroit to Chicago, via South Bend and Gary, Indiana.
Maps: Jonathan Rodden and Jowei Chen
The problem for Democrats is clear in Indiana’s new districts. They create large Democratic majorities in the biggest cities, but the smaller Democratic clusters along the rivers and railroads are overwhelmed by their solidly Republican surroundings. As a result, a state with a roughly even split of Democrats and Republicans will probably continue to have an overwhelmingly Republican congressional delegation.
An apolitical districting procedure won’t remove this geographic bias. We wrote a computer program to assign randomly each voting precinct to a congressional district without regard to that district’s partisanship or racial composition. The program only ensures that districts have equal populations and are geographically contiguous. For each of our thousands of simulated districting plans, we examined the share of districts that Republicans would win in the event of an overall tie between a Democratic and a Republican candidate for president, suggesting an electorate split along party lines. Averaging over our simulated plans, we discovered that Democrats could only expect to win a little more than three of Indiana’s nine districts—not much better than under the new redistricting plan.
The results are striking in other states as well. The graph displays the share of seats in the U.S. House and state legislatures that Republicans can expect to win in the event of a tied popular vote (we focused on states for which we had appropriate data from the 2000 presidential election). In sixteen of nineteen states, our party- and race-neutral simulations produced maps that lean Republican, sometimes overwhelmingly so.
[Click on graphic to enlarge.]
Democrats would do better if they intentionally drew wedge-shaped districts radiating from city centers—as Democrats have proposed in Chicago—and odd-shaped districts that follow rivers and nineteenth century railroad lines to string together far-flung Democratic enclaves. Even if Democrats dominate the process, however, the best they can hope for is usually zero or very slight bias in their favor. Republicans can obtain far more. So, while Democrats approve of reforms aimed at creating a party- and race-blind districting process, our analysis suggests that such reforms will be of little help in most states.
Are there other options for the Democrats? They can win and gerrymander aggressively, as they have in Illinois. But the deck is so severely stacked in states such as Florida that the prospect of unified Democratic control of the districting process seems remote. Moreover, such efforts might run into legal challenges based on the Voting Rights Act, and Democratic incumbents in safe urban seats do not like to have their districts carved up for the good of the party.
They also can push for some form of proportional representation. However, proportional representation has gained little traction in the United States, and it is also unappealing to safe Democratic incumbents.
In short, the Democrats have a geography problem with deep roots, and it is not likely to go away soon.
Jonathan Rodden is Associate Professor
of Political Science at Stanford University.
Jowei Chen is Assistant Professor of Political Science at the University of Michigan.