Friday, June 23, 2017

Redistricting and Blue Cities

by Neil H. Buchanan

Can anything be done to make congressional and state legislative races more competitive?  The Supreme Court has taken on a case from Wisconsin that could meaningfully limit partisan gerrymandering.  Depending on Justice Kennedy's vote, that case could change the way districts are drawn, which in turn could radically alter the results of American elections.

I will surely have more to say about that case in future columns, especially the proposed formula for identifying impermissible partisanship that the plaintiffs would like the Supreme Court to endorse.  Before going there, however, it is first worth asking whether gerrymandering is as important as people like me think it is.

After all, if Republicans' recent lock on the House of Representatives and state legislatures is not a result of gerrymandering (and voter suppression, which is obviously the key part of Republicans' strategy), an awful lot of effort on Democrats' part is going to be misdirected.

While it seems rather obvious that Republicans have used redistricting to their partisan advantage -- indeed, they acknowledge it and brag about it -- there is a body of work that claims that this is somehow a mirage, or at least that gerrymandering is not the big deal that some people think it is.

Indeed, when John Oliver did one of his deep dives into this issue a few months ago on his "Last Week Tonight" show on HBO, he deliberately undermined the piece toward the end by citing research showing that Democrats have done this to themselves.

The idea is apparently that Democrats have packed themselves into cities, which has made it impossible for congressional and state legislative districts to be anything but Republican-leaning overall, because non-urban areas are now left to Republican dominance.

This "geographic clustering" idea has some logical appeal.  For example, two RAND scholars recently wrote an op-ed piece in The Washington Post describing their research into the causes of increased political polarization.  Drawing from the 2008 book The Big Sort, by Bill Bishop, they wrote:
"Bishop believed that beginning in the 1970s, Americans began clustering into communities with similar values and lifestyle preferences. He argued that this clustering had political consequences because individuals who share values and lifestyle preferences tend to share political preferences as well. Bishop speculated that these changes might have contributed to growing polarization in Congress, with like-minded communities tending to elect like-minded representatives."
So far, so good.  The op-ed, however, later goes on to announce:
"One factor we were able to rule out is gerrymandering – the process of manipulating congressional district lines to benefit whichever party is in power. We found similar clustering at the county level, suggesting that the clustering in congressional districts was not due to the redrawing of district boundaries."
The idea, then, would apparently be that Republicans have not done anything to pack liberals into districts.  We did it to ourselves.

But that is not actually what the research says -- indeed, that is not even the question that this research is trying to answer.  The issue of polarization is distinct from the issue of partisan gerrymandering, and this research simply asks how it is that politicians have become less willing to engage in bipartisan compromise.  The answer is that their constituents are more like-minded than they used to be.

In other words, we are apparently supposed to picture districts that were drawn for nonpartisan reasons but that became ever more partisan because of people's decisions about where to live.  The result of a non-gerrymandered system with clustering would be "safe seats" that allow politicians to be more partisan.

But the result of a gerrymandered system, with or without clustering, would be the same.  You can, in any case, redraw districts for partisan reasons even while geographic clustering is also happening.

Here is one way to think about this issue: It would be a rather astonishing situation if political tacticians spent as much time and effort as they do on drawing district lines if doing so had no payoff.

That is not to say that politicians are fully rational or that groupthink cannot cause politicians to deny all evidence and logic.  Thinking about the Republicans' debt ceiling insanity, or their response to the Sandy Hook massacre, or any of a number of other issues, quickly disproves the idea that politicians are incapable of making persistent errors.  But it is, at the very least, difficult to understand why politicians would be systematically wrong about political tactics.

In any case, given that clustering is a relatively slow process, we should see these changes happening slowly.  Yet what we have seen instead is that states with Republican legislatures -- most prominently Texas, Pennsylvania, and North Carolina -- have in recent years successfully shifted the partisan composition of their congressional delegations in very short order.

In North Carolina, for example, the redistricting following the 2010 census was tested in the 2012 elections -- hardly a big year for Republicans, even though Mitt Romney did win back the state after John McCain lost it in 2008.  Due to that election, the state's congressional split changed from 7 Democrats and 6 Republicans to 4 Democrats and 9 Republicans, with Republicans knocking off another Democrat in the following election, leading to the current 3-10 split.

In fact, after North Carolina was recently found to have used impermissible racial factors in drawing two congressional districts (a ruling later affirmed by the Supreme Court), the Republicans in the state legislature announced a plan to redraw the state's districts.  The problem?
"But one thing would not change. According to voting statistics released for the proposed districts, three would strongly favor a Democrat, while the other 10 lean Republican. GOP lawmakers say they want to keep the existing 10-3 partisan split."
So in a state with a relatively even split of Democrats and Republicans, in which Republicans had successfully gerrymandered a shift of four out of 13 congressional seats, the state's Republicans responded to a required re-redistricting by openly reaffirming their commitment to partisan advantage.

But is the Democrats' insistence on living in cities nonetheless part of the problem?  To put it differently, does the liberal preference for city living make it impossible to unpack districts into something competitive, even if the districts were to be drawn by nonpartisans who wanted districts not to create partisan advantage?

It is hard to see how.  Imagine a state with one large city filled with Democrats, while the rest of the state has nothing but Republicans.  The overall partisan split among voters is 50-50, and the state has ten districts.  Under what we could call Plan A, it is not difficult to see how to draw districts so that we would end up with five safe seats for each party.

But if one wanted to create competitive districts, all that would be necessary would be to go to Plan B, in which the city is sliced like a pie and its ten slices are joined with ten (geographically larger) non-urban slices of the state.  This would create nothing but competitive districts.

If one draws the districts under Plan A, one creates highly polarized districts with safe seats.  Plan B results in non-polarized districts with competitive seats.

Geographic clustering does not make Plan A inevitable, even with the (rather loose) requirements of contiguity and compactness.  Plan B, after all, need not be achieved by drawing the infamous "ink-blot" districts.  In a square state with the blue city in the middle, the district map would merely look like a squared pizza that had been cut into ten equally-sized slices.

Neither Plan A nor Plan B is necessarily more small-d democratic.  And it is a separate question entirely whether a state in which one party has, say, a 60-40 advantage among voters "should" send six people to Congress from one party and four from the other.

The point is that congressional districts can easily be drawn to be competitive, even if Democrats are clustered into cities.  Indeed, in the example that I described above, the voodoo of gerrymandering is required to get something other than a 5-5 partisan split  (at least as a matter of probability, in the case of Plan B).  With gerrymandering, in fact, a guaranteed 9-1 split is possible.

It is true that "grass-roots partisan mobilization can overcome gerrymandering," as Julian Zelizer recently reminded us.  But that is very different from saying: "Gee, we really wanted not to have a Republican-friendly map, but Democrats' housing choices forced our hand!"

There are plenty of difficult questions in creating legislative districts, but this is not one of them.


Stuart McPhail said...

Agree with your point, but there does seem to be some wisdom in the packing idea. I.e., assuming one does not purposely craft districts to make them competitive or to ensure the D/R proportion of seats won roughly equates the votes cast, then you're more likely than not to end up with a system that biases towards republicans due to packing.

Which raises the question - why do districting at all? If you were a blue state that had enjoyed the benefits of gerrymandering in your favor, and were now forced to adopt some kind of neutral districting system that was likely to be biased against you, why not adopt a proportional representation system that awards seats based on the % of the total vote received?

May democracies use a proportional system and there wouldn't appear to be anything in the federal constitution to bar it. And while there are policy debates to be had about its pros and cons, it would seem like a viable solution for a blue state to adopt if it were forced to give up gerrymandering in its favor.

Joe said...

There is a claim that Democrats clustering in cities so that, not partisan gerrymandering, skewers things. As argued in the piece here, that is simplistic. I welcome the discussion since there are complex nuances on such questions & I will be upfront on my limited assurance at knowing all the details. Election Law Blog also had this report that addressed the issue:

I give John Oliver something of a pass -- those segments cover a range of issues and it is very hard to have a crisp fifteen minutes mixed in with some humor. A simple news program repeatedly gets facts wrong. Hopefully his research team is open to criticism and has a means of getting productive feedback.

I am curious of the value of a proportional system. I have read in various cases that this leads to racially discriminatory results, but would think there were means to use proportional systems in a way that honors racial and other diversity [certain nations in fact requires such and such number of women, and even places like Iran or Afghanistan thus has diversity in their government to some extent], including making sure local needs are covered. On that front, Chris Hayes noted on Twitter that he had trouble seeing how the average congressional district was much of a true "community" of interest, but then again, how exactly do we even determine that?

It's difficult, but structural matters must be addressed to deal with the current horrors of those in power now.

David Ricardo said...

Excellent post on what is probably the most significant issue with respect to democracy in the United States. That gerrymandering has produced a profound and significant bias into American politics is unquestioned. This is because of two types of gerrymandering, one accidental and one deliberate

The accidental gerrymandering is the result of the Constitutional scheme that provided for two senators per state. Lacking perfect foresight the Founders did not anticipate that the variance in the population of the 50 states would be so large and that some states would be rock solid Republican, a party that did not even exist at the beginning of the nation.

The result is a tremendous skewed Senate in favor of Republicans, and a less skewed electoral college. If for example the Dakotas were one state and California and Texas six states each the make up of the Senate and the numerical distribution of the electoral college would look much different. America would be much more of a representative democracy than it currently is. Nobody planned the state makeup this way, it just happened but the results have been highly undemocratic.

The deliberate gerrymandering is of course the deliberate drawing of House and state legislative districts to favor Republicans. Although I have not done the statistical analysis it is easy to see that basic statistics would provide an objective proof that Republican gerrymandering has succeeded. Consider the following thought experiment.

Assume that a state is approximately 50-50 divided between Democratic and Republican voters. If districts were designed in a non-partisan manner the electoral result could be but likely would not be 50-50 representation. The reason for this is that voting is motivated by more than just party affiliation. There is incumbency, personality, race and gender and other factors beside party identity that determines how one votes.

However party identification is a significant factor and it would limit the size of the party discrepancy. A 45-55 division in favor of one party could likely result, and even a 40-60 division might not be statistically significant. But the divisions that currently exist are so great in many areas that they could not be the result at say the 95% confidence limit of random results. The fact that the distribution of Republicans to Democrats in states like North Carolina and Virginia in state legislatures and congressional districts is so far removed from the distribution of voters by party identity that it can only be the result of deliberate bias in the districting process.

Mr. Buchanan mentions North Carolina. I am a resident of western North Carolina. Until 2012 we had a Congressional District that was relatively homogenous and somewhat evenly split between Democrats and Republicans. The seat alternated between the two parties based on personality and a desire to change out the party in power. In 2012 the Republicans openly stated their desire to rig the district, and split it into a Republican dominated district with the Democratic voters being pushed into a heavily Republican district in the center of the state. Democratic Buncombe County was split to disenfranchise Democratic voters with respect to representation in the House. This is not spin, this is not conjecture, it is pure fact. You can look it up.

Joe said...

The Senate was less of a problem when you had a limited number of states and the population differences was much less stark. But, there was some concern from the start and James Madison himself wanted the Senate to be apportioned by population.

Slavery was one reason why certain states began to be firmly one party. Then, the creation of certain states -- like the Dakotas -- were purposely formulated with party concerns in mind. The thinly populated states these days benefit one party more than the other, and even when small states like Vermont benefit the Democrats, there is room there for Republicans to win.

And, yes, Texas and California only having two senators is really bad. This is why the filibuster, which might help the Democrats some short term, is a problematic thing. The minority that blocks is even more of a minority than their mere numbers might suggest. At this time, last I checked, it isn't THAT bad -- simply counting Democratic and Republican voters nation-wide, Democrats would probably control the Senate, but only by a few votes. But, yeah, a 52-48 Democratic Senate (or whatever, if the Senate had more members) would not only be good policy-wise, but democratic-wise.