Pennsylvania, Pennsylvania, Pennsylvania
It's really important for both parties to win PA this fall
I haven't written anything in the past three weeks, mostly because there haven't been any significant events or changes in the race that warrant a full article. After a wild summer, this has been a nice, calm break before things heat up again on September 10th, when Trump and Harris (likely) take the debate stage together for the first time and we enter the home stretch of the 2024 Presidential Cycle.With that said, I’ll dive into a brief overview of the current situation before we move on to the main event of this newsletter, which is more focused on electoral college math.
Recent Updates
Summary: The Presidential Election was a tossup 3 weeks ago and it remains so now.
There was definitely a “reset period” in the race during the few weeks after Biden dropped out, with Trump’s betting odds falling from a high of 72% on July 16th to a low of 44% on August 16th. Since then, Trump has mostly stabilized in the polls, trailing Kamala by about 3% depending on the aggregator and the day. The 44% was likely pricing in Kamala’s “momentum,” as she’s actually polling slightly better now than a couple of weeks ago, but Trump has rebounded to around 51% on Polymarket as of this writing. I think 51% is a bit high compared to what the current polls suggest (if the election were held today, I believe Kamala would have a slight edge), so it now seems the market is pricing in a slight reversion of her recent rise, expecting Kamala to drop by around 0.5% from now until election day as her honeymoon period fades.
Democrats have been polling well in recent generic ballot polls in the past week or so (still slightly trailing the national presidential topline), and are favorites to win the House. My latest forecast gives them a 63% chance of taking the chamber, which aligns with betting markets. I estimate Republicans have a 30% chance of winning a trifecta, while Democrats have a 23% chance, meaning we are now up to a 47% chance that Trump or Harris could be the first president since Bush in 1989 to not have a trifecta on their Inauguration Day.
What State is Most Important this Fall?
The main thesis of this writeup is just how crucial Pennsylvania is this fall—almost laughably so. There’s nearly a 50% chance that Pennsylvania will be the tipping point state, not only because it’s highly competitive and quite large, but also due to specific quirks in the electoral map this year that make it exceptionally tough for either side to win without it.
But moreso, I want to explain how I think about state importance and the electoral college more generally, so I can refer back to this in all my October articles when we dive into the real “Road to 270” phase of the year.
In analyzing the impact of winning or losing key states on a candidate’s chances of winning the Electoral College, we want to consider both direct and indirect effects. The direct effect refers to the increase in the probability of winning the Electoral College that is directly attributable to winning the state's electoral votes. The indirect effect, on the other hand, captures the broader influence of winning a state on the overall election dynamics—how a strong performance in one state signals and potentially drives similar success in other states. For example, if Kamala wins Michigan by 10 points, it greatly raises expectations for her success in closely related states like Wisconsin and Pennsylvania, and even in less demographically similar states like Arizona, as outcomes in these states are positively correlated.
To quantify these effects, we can employ a regression discontinuity approach. This method leverages the Democratic vote percentage as a running variable to analyze the discontinuity in the probability of winning the Electoral College. The "discontinuity difference" refers to the change in a candidate's chances of winning the Electoral College when a state's probability of victory shifts from just below 50% to just above 50%. This difference represents the direct effect, as the RD analysis controls for the running variable.
Looking at the results of simulations right now in the table below, Pennsylvania (PA) stands out as particularly important. The table shows that if Harris wins Pennsylvania, her chance of winning the Electoral College jumps to 89%, but it drops to about 20% if she loses the state. Pennsylvania is so crucial that I like to center my electoral college analysis around it. I focus on the scenarios conditional on the outcome in Pennsylvania—considering how Trump could possibly win if he loses there, and what Kamala would need if she doesn’t carry the state.
This chart gives a pretty good tier list for how important *actually winning* each swing state is this November if it is very close:
Tier 1: PA
Tier 2a: MI, NC, GA; Tier 2b: AZ, WI
Tier 3: NV
It’s surprising that Pennsylvania (19 EV) has a greater Discontinuity Difference than Georgia and North Carolina combined (32 EV). It’s pretty hard to wrap your head around, but its the current situation. If a state so likely to be the tipping point ends up being extremely close, a significant portion of the election could hinge on its outcome.
Really In-Depth Electoral Map Analysis (State-by-State Correlations)
While Nevada is effectively irrelevant, I think a lot of people might be surprised to see how high North Carolina ranks in importance. I was puzzled at first too, but it all ties back to Pennsylvania. Despite Nevada and Arizona being slightly more competitive, the most common path to 270 for Harris is through Michigan, Wisconsin, North Carolina, and Georgia—more often than paths involving Arizona or Nevada.
It’s important to recognize the strong correlation between North Carolina and Georgia due to their similar demographics, making them a more influential pair than Arizona and Nevada, which, despite being highly correlated, lack sufficient electoral votes to get her to 270 without GA, NC, or PA. Democrats have been polling better in North Carolina than fundamentals suggest. For example, while Trump is polling just 0.4% ahead, my fundamentals calculation gives him a 2.6% advantage.
Consider Situation A: Trump wins Florida, Texas, and Pennsylvania but loses Wisconsin, Michigan, and Georgia. In this scenario, Kamala's chances of winning North Carolina are higher than her chances in Arizona or Nevada. A victory in Georgia is a strong positive signal for North Carolina, while losing Texas and Florida slightly dampens her prospects in Arizona and Nevada. I realize I'm getting into the nitty-gritty here—when I say there's a difference, I mean p(Harris wins AZ) - p(Harris wins NC) = 7%, and p(Harris wins NC | Situation A) - p(Harris wins AZ | Situation A) = 5%. Because North Carolina and Georgia are highly correlated, powerful as a pair in terms of electoral votes, and Georgia is easier for Harris to win, it’s actually more important for her to secure North Carolina if it’s close.
Incredible analysis!