Political Gerrymandering may decimate Republicans

10/07/2013 09:34:00 PM
There's a conventional wisdom in the commentariat that political gerrymandering--where a party draws congressional districts in a way to increase the number of districts they can win--has made the republican party's control of the House of Representatives invincible for 2014. This is an argument that Thomas Friedman made in a recent column, which I summarized here. This argument is wrong.

It is true that republicans have heavily gerrymandered the electoral landscape after the 2010 census. This fact is most evident when you recognize that in 2012 a majority of Americans actually voted for Democratic congressmen, even though a majority of Americans ended up with republican congressmen. Clearly, Republicans have gerrymandered districts in a way that lets them win a greater share of the seats than votes.

But where Friedman and others go wrong is in confusing two different issues. Friedman is implicitly assuming that that the gerrymandering turned competitive districts into safe districts where republican incumbents just can't loose. The problem is that you can't gerrymander in a way that both makes all your districts safer and gives your party more districts. Political gerrymandering to gain seats is a thoroughly risky proposition.

Let's do some electoral math. Suppose there are three districts, with a total of 300 voters to be divided up between them. Lets suppose 102 of those voters typically lean republican, while 198 typically lean democratic.

Now lets consider two definitions of electoral "fairness:"
  1. Each party's share of representatives should equal its share of votes
  2. Districting should be as apolitical as possible.
These are, actually, mutually-exclusive goals. In our example, then (1) would imply two democratic representatives and one republican representative. The way to always ensure that representatives reflect the popular vote as much as possible is actually make districts as noncompetitive as possible: one district with 100 republicans, one district with 100 democrats and the third district with 98 democrats and 2 republicans. Now consider criteria (2)--the most apolitical districting scheme would involve selecting voters into districts at random, without regard to their political leanings at all. This leads to, on average, three districts each comprised of 34 republicans and 66 democrats. Far from being "fair" in the sense of (1), apolitical districting actually leads to, on average, one-party control of all the districts. Under (2), whichever party controls more than 50% of the national vote will win all the districts most of the time.

Now consider a case where the republicans in our three-district example seize control of the districting process. Even with less than a third of the national vote, they can take control of the legislature by creating one democratic district with 100 democratic-leaning voters, and two republican districts with 49 democratic and 51 republican voters each. But, republicans must pay a hefty price to win control of the legislature. Under the "fair" scenario in (1), republicans had one, totally safe, seat. They must trade that safe seat for two extremely unsafe, highly competitive districts in order to get a majority of the legislature. Even the slightest bump in public opinion--a meager 1.3 point drop in republican approval ratings--can throw all three seats to the democrats.

The lesson here is that political gerrymandering is an extremely risky strategy for the party that wins extra seats. A party can only increase the number of seats it holds by decreasing its margins in its other districts, thereby making them more competitive. This is why, despite the fact that the shutdown charades have only modestly favored democrats in national polls, it appears that democrats will retake the House in a 20 to 30 seat flop.