Earlier this week, the Pennsylvania Supreme Court released its much-awaited Congressional redistricting plan to replace a 2012 Republican-drawn plan which it recently ruled to be an unconstitutional partisan gerrymander. The new plan is unquestionably more fair, ensuring at least two more seats for Democratic voters who currently hold only 5 of Pennsylvania’s 18 seats, despite composing about half of the statewide electorate. However, the new plan, and others proposed over the last week, reveal a fundamental limitation of our electoral system that should concern not just the Democratic Party, but anyone concerned about political equality.
All of the non-partisan plans submitted would have been an improvement over the GOP gerrymander, but the Pennsylvania court’s requirements for the new plan included incompatible design principles. Analysis of the plans hints at a future where a majority of voters may routinely be denied a majority of seats in competitive swing states.
Figure 1: Pennsylvania’s new Congressional map: fairness wins
The new plan creates 8 districts won by Hillary Clinton in 2016 and 10 won by Donald Trump, though Democrats could pick up as many as 11 seats in 2018 if everything goes their way. It is clear that Nathaniel Persily, who designed the map, explicitly prioritized fairness in the design. Nevertheless, most initial estimates, while based on incomplete data, suggest that the plan still slightly favors the Republican Party. That is, at a 50% vote split, GOP voters have a better chance of winning 10 of 18 seats, compared to Democratic voters. Why?
As Sam Wang, director of the Princeton Gerrymandering Project, showed in a weekend analysis, most of the submitted plans yielded 6 to 8 safe seats for Democrats, 9 safe Republican seats, and more seats that lean to Republicans over Democrats. Arguably the fairest plans were those submitted by Adele Schneider and Stephen Wolf at Daily Kos, which had the two parties likely to split nine seats, and restrained partisan asymmetry (measured in terms of partisan bias) to just under statistical significance.
The crucial dynamic at play is the interaction between incompatible design principles. The court explicitly wanted a plan that cut across fewer country and municipal boundaries, with geographically compact, single-seat districts, that nevertheless treat individual voters equally, regardless of which party they vote for.
In our 2016 book Gerrymandering in America, my co-authors Anthony J. McGann, Charles Anthony Smith, Alex Keena and I demonstrated that it is generally easy to create fair districting plans, but that there are trade-offs between principles. While there is a tendency to think of the concentration of Democrats in dense, urban populations as a “natural” geographic disadvantage in districting, it is important to keep in mind that there is nothing “natural” about compactness, boundary-crossing, or the use of single-seat districts. These are political choices with political consequences.
Figure 2: Redistricting Chicago-Style
For example, the 2011 Illinois state legislature, controlled by the Democratic Party, was able to draw an unbiased plan (though not one that benefited them), but at a cost of drawing some odd looking, non-compact districts. I think of this as a “Chicago-style” plan, as Congressional districts spread out from Chicago, like slices of deep-dish pizza, to incorporate more conservative suburbs into the densely Democratic city districts. This illustrates the trade-off between the geographic goal of compactness and the individual goal of political equality.
Figure 3: Philly-Chicago-style: Less compact districts yield 8 safe Dem seats, 2 lean Dem seats, 4 lean R seats, 4 safe R seats. Do these districts look more like pizza slices or cheesesteaks?
I tried to replicate this in Philadelphia using Dave’s Redistricting App, but still could only produce a map with 8 safe Dem seats and 2 Dem-leaning seats (though with more tossups). Even without prioritizing compactness, it is difficult to assure that the party winning a majority of the vote will receive a majority of the seats. And here is the rub: if urban Democrats become increasingly segregated with increased partisan polarization, it is not just compactness that creates a fundamental restraint on how fair a plan can be. It is not just urban Democrats who waste votes under urban-rural polarization, it is all minority voters, Republicans in cities as well as rural Democrats, who still typically make up 30-35% of rural voters.
Figure 4: Multi-Seat Districts in PA; Blue (2R/3D), Yellow (2R/2D/1?), Red (2R/1D), Green (2R/3D)
Our reliance on single-seat districts puts limits on our ability to remedy discriminatory districting of any kind. To see the effect of this institutional choice, consider the possibility of multi-seat districts in Pennsylvania. I designed a plan using three 5-seat districts and one 3-seat district. The large, red district reflects the three seats allocated in the most rural part of the state. Assume that an electoral formula is used to allocate seats to party candidates proportionately. Further, assume that the Republican and Democratic parties would not have any additional competition (a risky assumption, but that’s a separate story).
The “eastern” district next to Philadelphia would lean slightly Democratic (average party vote 54% Dem, 46% GOP) while the “western” district with Pittsburgh likely yield two seats to each party, with a toss-up seat. Overall results would more proportional, and more likely to respect statewide majority rule. Further, parties would have an incentive to compete for votes throughout the entire state. In terms of racial dynamics, the Philadelphia district is approximately 25% African-American and 8% Hispanic, such that racial bloc voting would likely produce multiple racial minority seats (a candidate with 20% of the vote is guaranteed a seat). The western district, with Pittsburgh and other racially diverse cities in South Pennsylvania, would also have a good shot at electing a candidate strongly supported by people of color.
The importance of district boundary locations also begins to fade, as they are less consequential in shaping electoral outcomes. Three 6-seat districts in Pennsylvania could be designed to send equal numbers of candidates from both parties, and a single, 18-seat district would guarantee majority rule and political equality with only one district boundary, the state.
If the current trend of urban-rural partisan polarization continues, even a clear standard against unconstitutional gerrymandering and strict enforcement of voting rights may not be enough to protect majority rule and political equality. For all the work that we are putting into fixing our single-seat system, its inherent limitations could actually become more apparent.