According to PlanScore’s basic model, the remedial plan adopted today by the Pennsylvania Supreme Court is impressively symmetric—both in absolute terms and compared to its predecessor. Its likely efficiency gap is -2% (versus its predecessor’s -10%), its likely partisan bias is -4% (versus -12%), and its likely mean-median difference is -3% (versus -4%). In my view, this is the single most important fact about the remedial plan. It’s not just comprised of aesthetically appealing districts (though its districts are, in fact, much prettier than their antecedents). Rather, it promises actually to cure the underlying constitutional violation—to eliminate the majority of the previous plan’s pro-Republican skew. This is exactly what a partisan gerrymandering remedy should do, I think.
Another notable feature of the remedial plan is its competitiveness. According to PlanScore’s basic model, five of the plan’s eighteen districts are between 45% and 55% Democratic (and a sixth is just a hair outside this range). The plan could thus plausibly elect anywhere from eleven Democrats to thirteen Republicans depending on candidate quality and the overall electoral environment. This high level of competitiveness is reflected in PlanScore’s sensitivity testing. In a good Democratic year, Democrats could win several more seats, thus tilting the plan’s efficiency gap in their favor. Conversely, in a good Republican year, Republicans could pocket even more seats, thus swinging the plan’s efficiency gap even further in their direction.
I’ve referred a couple times now to PlanScore’s “basic” model. It’s basic because it doesn’t take incumbency into account, instead relying on 2016 election results and demographic data. What happens if incumbency is incorporated into the model? (This feature is not yet available on the website, though it’s coming soon.)
For one thing, the model’s predictive power improves a bit, explaining 89% (rather than 84%) of the variance in the precinct-level congressional vote. More interestingly, it turns out that incumbency is a bigger electoral advantage in Pennsylvania than in most states today. The typical Democratic incumbent enjoys a 10% boost (compared to a Democratic candidate in an open seat), and the typical Republican incumbent benefits from a 6% lift. These figures are reminiscent of the 1980s and 1990s, and are more than double the usual contemporary levels.
Once the model includes incumbency, it becomes possible to analyze an array of electoral scenarios. For example, what if every seat in Pennsylvania was an open seat, with no incumbent running? Then Democrats would be expected to win nine seats (compared to eight using the basic model). District 17 would flip from barely Republican to narrowly Democratic.
Alternatively, what if we based the analysis on the actual incumbents’ current plans? As of today, there are twelve incumbents running for reelection in Pennsylvania: three Democrats and nine Republicans. Under this candidate configuration, Democrats would be expected to win just six seats. Three districts that would be marginally Democratic if they were open seats (1, 6, and 17) would flip to the Republican side of the ledger thanks to their Republican incumbents. This result renders particularly hollow the argument that the court’s remedial plan is “pretty close to a Democratic wet dream,” as one Republican consultant colorfully put it. In fact, given incumbents’ current plans, Republicans could keep their congressional supermajority, at least in an electoral environment similar to 2016.