This is the second part in a series where I look at each of the swing states and which counties should be on the radar heading into the 2020 general election. In the first part, I covered Michigan, a Rust Belt state that flipped from President Obama in 2012 to Donald Trump in 2016. In this part, I’ll be looking at Pennsylvania, a similar state.
The Big Picture
Donald Trump- 2,970,733 (48.17 percent, up 1.58 percent from Romney in 2012)
Hillary Clinton- 2,926,441 (47.46 percent, down 4.51 percent from Obama in 2012)
Voter Turnout- 70.11 percent (up 2.62 percent from 2012)
Similar to Michigan, Donald Trump’s victory in Pennsylvania surprised political analysts on election night in 2016. While it has been friendlier to Republicans in congressional and state-level races, Pennsylvania hadn’t gone Republican in a presidential election since George H.W. Bush in 1988. FiveThirtyEight only gave Trump a 23 percent chance of winning the state four years and other forecasting models were even less confident in his chances. But similar to Michigan, much of the story here four years ago was Trump’s ability to win heavily among white voters without a college degree; in Pennsylvania, Trump won 64 percent of this demographic.
Even with this success, it was a tight race. Trump only won by 44,292 votes in a state where over 6 million people casted ballots. Trump also benefited immensely from a spike in voter turnout. Whereas Clinton received 63,833 fewer votes than Obama in 2012 despite higher turnout, Trump netted over 290,000 votes from Romney’s total. Of course, some of this is also Trump successfully flipping previously Democratic voters, but there’s strong evidence that the 373,000 additional voters that came out in 2016 heavily backed Trump over Clinton.
There’s mixed evidence on how well Biden will be able to recapture the ground that Clinton lost. In the 2018 midterm elections, Democratic Governor Tom Wolf and Senator Bob Casey Jr easily won re-election with 57.8 percent and 55.7 percent of the statewide vote respectively. On the surface, it would seem that voters that backed Trump in 2016 are dissatisfied with his presidency and are reflecting on that with increased support for Democratic candidates.
But on the other hand, while Pennsylvania has more registered Democrats than registered Republicans, Republican voter registration has grown at a faster rate than Democratic voter registration. Between 2012 and 2016, Republican registration grew by 5.8 percent while Democratic registration shrunk by 1.1 percent. Since 2016, Republican registration grew by an additional 3.2 percent while Democratic registration shrunk by an additional 1.9 percent. This reflects a shift in the state’s political composition that appears to favor Republicans, despite Democrats remaining competitive in elections.
Considering these demographics, turnout will be important for deciding the state in 2020. While it’s possible that Trump voters could flip their votes in response to the pandemic and economic recession, this won’t be clear until the votes are actually submitted. Conversely, turnout among racial minorities and Hispanics was relatively low in 2016, which presents a clear opportunity for Biden to make up lost ground. With that said, let’s consider the three subsections of counties that will be critical for deciding the statewide race.
Solid Democratic Counties
Allegheny- (Pittsburgh city proper and surrounding communities)
Delaware- (Philadelphia metropolitan area, on Delaware border)
Montgomery- (Philadelphia metropolitan area)
Philadelphia- (Philadelphia city proper)
Admittedly, I’m cheating a little bit with this grouping. Normally, I define a “solid” county as one where the winning candidate received more than 60 percent of the vote; however, in 2016, Philadelphia was the only county in the entire state where Clinton received more than 60 percent. In light of this, I decided to include counties that Clinton got between 55 and 60 percent as these counties wouldn’t otherwise be included.
As expected, the strongest bases of Democratic support are in and around Pittsburgh and Philadelphia, the state’s two largest cities. Indeed, these four counties have been Democratic strongholds for decades. The suburban Delaware and Montgomery counties have gone blue in every election since 1992. Allegheny hasn’t gone red since the 1972 Nixon landslide, even staving off similar shock waves in the two Reagan elections. And Philadelphia County, home to the nation’s sixth largest city, hasn’t gone for a Republican since Herbert Hoover in 1932.
Socioeconomically, this subsection is very affluent. In 2016, all four of these counties were in the top ten for highest per capita personal income (in fact, all they were all in the top six). Combined, this subsection has a per capita income of $59,649, placing it well above the state average. Not to mention that these four counties contributed 42 percent of the state’s GDP in 2016, making them a significant factor in the state’s economy. Granted, there remains a high degree of inequality; Philadelphia had a 25 percent poverty rate in 2016, the highest in the state. Even so, one cannot ignore these indicators when describing this subsection of counties.
Demographically, the Democratic advantage becomes more apparent. In 2016, nonwhite residents made up 34.4 percent of the population (about double the state average) and Hispanic residents made up 7.5 percent of the population (roughly the same as the state average). Philadelphia has the bulk of this diversity, with over half of its residents being nonwhite and 14 percent being Hispanic (and a big chunk of that Hispanic population has emerged since 2000). Meanwhile, other counties have seen their diversity grow in recent decades. Since 2000, Montgomery’s share of nonwhite residents grew by 7 percent and Delaware’s share grew by 10 percent, while both of their shares of Hispanic residents have doubled. These demographic shifts suggest that not only do this subsection’s current population favor Democratic candidates, but they are expected to continue moving further in that direction moving forward.
Despite the Clinton campaign’s lackluster performance across the entire Rust Belt, she didn’t do that badly in this subsection. In terms of raw votes, Clinton received 38,485 more votes than Obama in 2012 (granted, the total amount of votes cast grew by 79,894). In terms of vote share, Clinton only performed 0.7 percent worse than Obama. And voter turnout is already very high in this subsection (as it is across the rest of the state), hovering around 69 percent in 2016. Even if Clinton had gained an additional 0.7 percent to her vote share and matched Obama’s in 2012, she would have only netted 15,000 votes, not enough to flip the state. While one can argue that Clinton should have done better than Obama in 2012 considering she’s not an incumbent seeking re-election, the fact that Obama still won the state despite “under performing” indicates that this subsection wasn’t a problem.
The bigger problem is that most of the spike in turnout statewide occurred elsewhere, in regions that are more receptive to Trump’s campaign message, which resulted in this subsection making up a smaller percentage of the electorate. In 2012, this subsection contained 35 percent of all presidential ballots cast in Pennsylvania while in 2016, it only contained 34.2 percent. That may sound like a small difference, but if we hold the statewide electorate size equals, this translates to 50,812 votes that would have been cast in one of these four counties instead going somewhere else. If Clinton matches her vote share in this expanded electorate, that translates to 33,674 votes that can be added onto her total. And assuming these votes are subtracted from areas where Trump dominated (after all, the statewide electorate is fixed in this scenario), that is enough to flip Pennsylvania into Clinton’s column.
This may sound like an arbitrary exercise; however, it highlights how the Clinton campaign relied too heavily on this subsection in 2016 while lagging elsewhere in the state, where more of the votes came from in the election. This will be discussed later in the article, but for the purposes of this section, it demonstrates that while Biden will need to carry these counties by strong margins, he will also need to expand beyond the subsection in order to win Pennsylvania.
Luckily for Biden, there’s a lot of reason to believe that the campaign will carry these counties by substantial margins in 2020. In addition to the favorable demographics, this subsection has a history of supporting Democratic candidates in presidential and statewide elections. Aside from the large cities of Philadelphia and Pittsburg, this subsection contains suburbs that are experiencing similar changes as other suburbs across the country. Namely, these suburban voters are unreceptive to President Trump’s “law and order” message and his rhetoric of how Biden would “destroy” the suburbs. This shift is also seen during the 2018 midterms, when both gubernatorial candidate Tom Wolf and incumbent Democratic senator Bob Casey Jr. received over 70 percent of the vote in these four counties.
Furthermore, it should be noted that the Philadelphia metropolitan area is close to Wilmington, Delaware, where Joe Biden has lived for most of his life. This gives Biden a deep personal connection with that region, which I think will also play to his advantage in these populous counties. It should come as little surprise that his campaign headquarters is located in Philadelphia. While party affiliation still plays a large factor in voting behavior, these personal and local factors could sway certain voters that have seen or heard about what Biden is like as a person, when the cameras are off and reporters are nowhere to be found. There’s also the townies that would take immense pride in a president that’s familiar with their communities. That factor should count for at least something, even if these counties are already heavily favorable towards Democrats.
Overall, not much work will be needed from the Biden campaign in this subsection. That isn’t to say he should ignore it entirely. And he hasn’t, given his recent speech in Philadelphia on the death of Supreme Court Justice Ruth Bader Ginsburg. But generally speaking, the Biden campaign should ensure that it at least matches the vote share and turnout that Clinton received in 2016 (and preferably higher). Given that turnout is already high in these counties and there’s plenty of evidence that these voters will back Biden heavily, these four counties are likely to serve as the backbone for the campaign’s efforts to recapture Pennsylvania in the fall.
Narrowly Contested Counties
Bucks- (Philadelphia metropolitan area, on New Jersey border)
Centre- (central part of state and Penn State University)
Chester- (Philadelphia metropolitan area, on Delaware border)
Dauphin- (Harrisburg city proper and surrounding communities)
Erie- (northwest part of state, on Lake Erie)
Lackawanna- (Scranton city proper and Joe Biden’s birthplace)
Lehigh- (Allentown city proper and surrounding communities)
Monroe- (northeast part of state, on New Jersey border)
Northampton- (eastern part of state, on New Jersey border)
These are the “swing” counties that Clinton received between 45 and 55 percent of the vote in 2016. This is an interesting collection of counties. Two of them are Obama-Trump counties (which will be discussed more in the next section). On the flip side, one of them (Chester) is a Romney-Clinton county, which reflects the parties’ realignment despite not being widely discussed. Another three flipped to Obama in 2008 and have stayed in the Democratic column ever since. And another three have gone for the Democratic candidate in every election since 1992, reflecting staying power for the party. While the Democratic Party has a different history across these counties, the one thing they have in common is that they were all competitive in 2016. Some of that was due to unhappy Republicans refusing to back Trump, but mostly it reflected Clinton’s weaknesses as a candidate presenting opportunities for Trump to infiltrate.
Remarkably, Clinton actually won seven of these nine counties, including one that Obama didn’t carry in 2012 (Chester County). In fact, Clinton received 7,188 more votes in this subsection than Obama did in 2012. These sound like Clinton gained ground in these counties.
But upon closer inspection, it becomes evident that these “improvements” weren’t enough. Despite receiving more raw votes than Obama, Clinton’s gains were outpaced by increased turnout. Whereas Obama received 52.5 percent of the vote in this subsection, Clinton only got 49.4 percent. Also when looking at the individual counties, Clinton performed worse (by vote share) than Obama in all but one county. In four counties, Clinton’s vote share was more than 5 percentage points smaller than Obama’s 2012 performance (in fact, two of those counties had Clinton perform more than 10 points worse). And while it was impressive she was able to capture Chester County after Obama couldn’t, if we remove this county from this subsection, Clinton’s 7,188-vote gain on Obama turns into a 10,183-vote loss (in terms of vote share, removing Chester County turns a 53.2 percent performance in 2012 to 48.6 percent in 2016, adding 1.5 points to Clinton’s deficit). Chester County was still an impressive gain, but it hardly makes up for losses elsewhere.
Furthermore, voter registration grew by 4.5 percent between 2012 and 2016, considerably faster than the state average and making it more critical to win these counties. But the biggest factor is that while Republican voter registration continued to trail Democratic registration, Republican registration grew a greater margin than Democratic registration; among the 87,000 new voters added to the rolls, Republican netted over 17,000. While this alone doesn’t explain Clinton’s loss statewide, this presented another missed opportunity for the campaign. Indeed, the counties that Clinton lost the most ground from 2012 (Erie and Lackawanna) are also the counties that have the greatest gain in Republican voter registration as well as the greatest decrease in Democratic registration. To this end, this reflects changing party composition in this subsection.
In terms of educational attainment, every county in this subsection is in the top third statewide for the percentage of adults aged 25 and over with at least a bachelor’s degree. In fact, Chester County is number one in Pennslyvania, with half of its adult population having at least a bachelor’s degree. Centre County, home to Penn State University, is third and Bucks County is fifth. And this somewhat carries over to its socioeconomic standing. Five of the nine counties are in the top third statewide by per capita income, with Chester and Buck Counties being in the top three; however, there isn’t a one-to-one correlation between income and educational attainment. Centre, for example, is 28th in per capita income (in 2016, it was $42,183) despite having the third largest percentage of adults with at least a bachelor’s degree (44.7 percent). This may have to do with the county being far from the state’s largest cities of Philadelphia and Pittsburgh, meaning that a hefty chunk of the county’s high-paying jobs are concentrated in one employer (Penn State University). Even so, the subsection’s large concentration of highly-educated voters (and even young, college-age voters) is a strong advantage for the Biden campaign heading into 2020.
In terms of race and ethnicity, the subsection’s demographics present an interesting story. While it has a slightly higher Hispanic population than the rest of the state (9.5 percent), overall, this subsection’s population is slightly less racially diverse (14.4 percent of its population is non-white, versus 17.7 percent statewide). Some of these counties stand out for their diversity. In Dauphin County, home to the state’s capital of Harrisburg, over a quarter of the population is nonwhite. But in other counties, such as Lackawanna, whites make up over 90 percent of the population. This variance in racial diversity certainly complicates the political composition of these counties; however, almost all of them have seen an uptick in nonwhite and Hispanic residents in recent decades. This points towards a key shift in demographics across all these counties.
While these swing counties are found in all corners of the state, most of them can be characterized as either suburban or urban, granted the “urban” counties mainly consist of medium-sized cities, such as Allentown, Harrisburg, and Scranton. These areas are a critical component of strategies by Biden and Democratic candidates across the nation. And similar to other Rust Belt states, Biden can appeal to voters in this region based on economic issues; many counties in this subsection have experienced large spikes in unemployment during 2020 and much progress can be made by Biden by centering his message around President Trump’s shortcomings in handling the economy. Furthermore, the Biden campaign can capitalize on gains experienced by Democrats on suburban voters since 2016.
In the 2018 midterms, this subsection broke heavily for Democratic candidates Tom Wolf and Bob Casey Jr. Each of these candidates won all nine counties in the subsection, receiving 59.6 percent and 57.2 percent of the vote respectively. Furthermore, while Pennsylvania’s suburban voters favored Trump in 2016 according to exit polls, they swung favorably towards both Democratic candidates in 2018. This should be encouraging for the Biden campaign.
Finally, the Biden campaign can take some solace in the fact that while Republican voter registration grew several times faster than Democratic registration in this subsection between 2012 and 2016, such growth has slowed since 2016. Since 2016, the two have grown at roughly the same rate of 1 percent. This is evidence that since President Trump took office, the subsection’s shift to the right has not outpaced shifts to the left. Not to mention the fact that putting growth aside, there are still more registered Democrats than registered Republicans.
Overall, this subsection is winnable for Biden; however, he will need to beat Hillary Clinton’s margins from 2016 in order to win statewide. Given Biden’s strengths as a candidate over Clinton, this is a very doable task. I believe that these voters are aware of the difference between the idea of a Trump presidency and the reality of it, which may change their voting behavior this time around. Given the subsection’s demographics, I believe the voters here are more likely to flip from Trump to Biden than they are from Clinton (in 2016) to Trump now, especially with the issues on the table. Even so, I must emphasize that the race is not over for either candidate.
Obama-Trump Counties
Erie- (northwest part of state, on Lake Erie)
Luzerne- (northeast part of state)
Northampton- (eastern part of state, on New Jersey border)
In contrast to Michigan’s twelve, Pennsylvania only has three Obama-Trump counties, which doesn’t appear so bad until one examines the extent of those losses. In a subsection where 28,388 more ballots were cast than in 2012, Clinton received 23,114 fewer votes than Obama. This translates to a vote share loss of almost 10 percentage points, going from 53.5 percent in 2012 to 44 percent in 2016.
It’s worth noting that two of these counties (Erie and Northampton) are also “swing” counties. Erie and Northampton share certain similarities, such as their comparable educational attainment (27.7 percent of Erie’s adult population has at least a bachelor’s degree, as with 29.2 percent for Northampton) and similar nonwhite populations (hovering around 12 percent each). Both also have a history of being manufacturing hubs, with Erie involved on steel, coal, and commercial fishing while Northampton focused more on cement. Both also have a history of voting blue in presidential elections, having supported every Democratic candidate since 1992.
Certainly, Erie and Northampton also have considerable differences, such as their location at opposite corners of the state. Northampton’s population is three times more Hispanic than Erie’s, although Hispanics remain a minority in both. Northampton is also more economically prosperous than Erie, with a lower poverty rate, lower unemployment rate, and higher per capita income. But above all else, their commonality is that they switched their allegiance from President Obama in 2012 to Donald Trump in 2016.
Luzerne is the one addition to this group and it’s a fascinating case study. Despite not containing any major cities, Luzerne has one of the fastest-growing Hispanic populations in the nation. Since 2000, Hispanics went from making up only 1 percent of Luzerne’s population to making up almost 14 percent, a dramatic change in the county’s demographics. It’s also a county that saw one of the biggest swings from Obama to Trump in the state (over 12 percentage points). And it’s also a county where since 2016, Republican registration increased by 2.6 percent while Democratic registration decreased by almost 4 percent. In 2016, Luzerne appeared to be a prime breeding ground for Trump’s rhetoric on “making America great again” could resonate with the most amount of voters, particularly those that have seen the county’s demographic and economic changes over multiple decades. And low turnout from the county’s Hispanic voters contributed to Trump’s success here.
More broadly speaking, this subsection is the least racially diverse of the three discussed. In 2016, only 11 percent of the population was nonwhite and 9.4 percent Hispanic; however, these counties have seen an uptick in both of these groups in recent decades. One would think that these changes would result in these counties becoming more favorable for Democratic candidates. Certainly, the potential is there, but lower turnout in these groups was one reason why these counties flipped in 2016.
I will argue, however, that it’s not the only reason. There is evidence that voting behavior in this subsection has trended towards Republicans in recent years. Between 2012 and 2016, Republican voter registration increased by 10.5 percent while Democratic registration shrunk by 1 percent. This is similar to the trend in the swing counties during that time, where growth in Republican registration outpaced that of Democratic registration. But a critical difference between these two subsections is that since 2016, Republican registration has continued to outpace Democratic registration in the Obama-Trump counties. While the two parties’ voter registration grew at the same rate after 2016 in the swing counties, Republican registration in the Obama-Trump counties grew by an additional 8.4 percent between 2016 and now while Democratic registration decreased by an additional 0.7 percent. It’s important to mention that even after all this growth, there are still 69,000 fewer registered Republicans in this subsection than registered Democrats; however, that deficit has been cut in half from 2012. While low turnout from nonwhite voters may have deflated Clinton’s numbers, this shift helped Trump fill in the gaps.
So where does this leave Biden in 2020? On one hand, these counties don’t appear completely out of reach. In the 2018 midterms, Tom Wolf won all three counties and Bob Casey Jr won two, victories that were bolstered by high voter turnout and anti-Republican sentiment. Not to mention that there are voters that could turn out that didn’t vote in 2016 or Trump voters that are unhappy about the economy. Assuming 2016 turnout, if Biden can replicate Casey Jr’s vote share from 2018, that’s an additional 43,264 votes from this subsection (and if he replicates Wolf’s vote share, it’d be an additional 58,000 votes). So in theory, there are opportunities for Biden to make up ground.
On the other hand, one cannot ignore the effects of voter registration and Trump’s rise in the Republican Party on Democratic fortunes here. While Wolf did well here in 2018, he actually underperformed from his own previous campaign in 2014, a year known for strong anti-Democratic sentiment nationwide. This is most pronounced in Luzerne County, where Wolf did 5 percentage points better in 2014 as a challenger against a Republican incumbent than he did in 2018 where he was fending off a Republican challenger as a Democratic incumbent himself. A similar thing can be said for Senator Bob Casey Jr. In 2006 (a year with strong Democratic headwinds), Casey Jr got over 60 percent of the vote in Luzerne. Six year later in 2012, he got 54.5 percent. But in 2018, he lost 10 percentage points in Luzerne, costing him the county.
It’s difficult to say how high turnout will be in 2020 in this subsection. But even if turnout among racial minorities and Hispanics run high in these counties, I think there will still be a large contingent of Trump supporters that will back the president for another four-year term. I don’t think Biden will win back Luzerne; however, Erie and Northampton are certainly winnable given their dual status as “swing” counties.
Overall, the Biden campaign’s goal in this subsection shouldn’t be to win every county. As I mentioned earlier, there’s a sizable chunk of this subsection’s population that are unlikely to ditch Trump, even if they are disappointed with his performance in office. The goal for the Biden campaign should be to drive up turnout among the growing nonwhite and Hispanic populations in order to keep the margins narrow, preferably within 5 percentage points. While I wouldn’t be surprised if Biden flips one or two of these counties, his priority should be to convert these counties into the “swing” column.
Conclusion
When put altogether, there are some opportunities for the Biden campaign to win Pennsylvania; however, it won’t be easy. The counties that have the most solid Democratic support already have high voter turnout, leaving few opportunities for more votes to be gained. And as more votes come from other parts of the state, it would be a mistake for the campaign to rely too heavily on the solidly Democratic counties.
Meanwhile, the Obama-Trump counties are going to be challenging. While I think Biden will get more votes there than Clinton received in 2016, the uptick in registered Republicans will produce diminishing returns for higher turnout among nonwhite and Hispanic voters. Still, Biden will need that higher turnout to cancel out the immense support for Trump in these counties.
I think the biggest opportunity for Biden will be in the nine swing counties, which are sufficiently varied in their demographics to where having a general voter appeal can go a long way. These swing counties have plenty of pockets of highly-educated voters, suburban voters, nonwhite voters, and Hispanic voters that Biden has a good shot of winning. I expect that if Biden wins Pennsylvania, it will come through an increase in support across these nine counties.
So that’s what I’m looking for in Pennsylvania in November. In the next part, I’ll be looking at Wisconsin, another Midwestern state that flipped from President Obama in 2012 to Donald Trump in 2016.