In June, I authored an article for The Federalist in which I argued that Donald Trump would not win Pennsylvania because any gains he might make with white working class voters would be insufficient to overcome the steep losses he was likely to suffer in the Philadelphia area. As it transpired, Trump was able to increase white working class turnout in his core areas by a massive amount and win the state. He needed this surge of blue collar white voters because he did fare poorly in the Philadelphia suburbs. An analysis of returns at the municipal level reveals that the Philadelphia “collar” swung sharply towards Hillary Clinton as support for the GOP nominee collapsed among white college-educated voters, a collapse that in any other year would have kept the Keystone State blue.
Barack Obama won Bucks County in 2012 by 4,000 votes. Hillary Clinton won Bucks County as well but her margin shrunk to 1,000 votes, a swing of 3,000 votes to Donald Trump. These raw totals however don’t tell the full story. Bristol Township, a heavily blue collar community (12% four-year college degrees) along the Delaware River swung to Trump by almost 4,200 votes, while Falls Township (21%) and Bensalem (27%) went red by 2,200 and 2,000 votes, respectively. Trump saw more modest but perceptible gains in numerous localities whose populations have a share of college-educated residents under 35%. On the other hand, Trump’s standing eroded in those parts of Bucks County with the highest share of residents with four-year degrees. Even in such places Trump won his margins diminished compared to Romney’s. Northampton Township (49% college-educated) remained red but swung 1,100 votes towards Clinton. Mitt Romney won Buckingham Township (58%) by 1,800 votes. Trump won it by just over 400, a shift of 1,300 towards Clinton. Obama won Newtown Township (61%) by ten votes over Romney. Clinton won it by 1,200.
Three municipalities exemplify Trump’s struggles with college-educated voters. Mitt Romney won Upper Makefield (the locality with the county’s highest share of college graduates [67%]) by 1,500 votes. Trump won it by just over 150, a swing of over 1,300 votes towards Clinton. Romney won Doylestown Township (49% bachelor’s degrees) by 800 votes. Trump lost it 600, a shift of 1,400. The most dramatic pro-Clinton tilt occurred in Lower Makefield (just under 67%). Mitt Romney won it by just over 200 votes. Hillary Clinton won it by 2,900, a shift in her favor of 3,100.
A similar pattern plays out in Montgomery County, but on a much more extensive scale. When pundits forecast that Trump would lose Pennsylvania because of the Philadelphia suburbs, Montgomery County is what they had in mind: wealthy, white, and (over-)educated. They were right, too, as Trump managed to do a staggering 31,000 votes worse in MontCo than Romney did. Trump bled and bled and bled in MontCo. Compared to Trump’s 90,000-vote deficit, Romney’s arrears of 60,000 seems almost heroic. Twelve municipalities swung by a thousand votes or more towards Clinton, a baker’s dozen if you round up in a thirteenth. Lower Merion (76% college grads) stampeded towards Clinton to the tune of 7,700 votes. Upper Dublin (64%) experienced a blue shift of 2,600 votes. Whitpain (60%) swung by over 2,100 votes. The only reason there aren’t more four-figure swings towards Clinton is that there was very little left to squeeze in some places.
The most dramatic swing of suburban voters towards Clinton occurred in Chester County. Chester was the one collar county that Romney won, albeit by a slim margin of fewer than a thousand votes. Clinton took it by 25,000. Again the municipal results tell the story. Nearly 76 percent of Charlestown Township’s residents have four-year degrees. Mitt Romney won it by around 125 votes. This year it went for Hillary Clinton by almost 450, a swing of over 550 to her advantage. Romney won Birmingham Township (74%) by 600 votes. Clinton won it by over 120, a blue shift of over 700. Mitt Romney won East Bradford Township (68%) by 400, while Clinton won it by almost 650, a swing of over 1,000 votes to the Democratic nominee. Romney claimed East Goshen (57%) by 1,500 votes. Trump lost it by a handful. Romney won Easttown Township (75%) by approximately 800. Clinton took it by 1,000, a shift of 1,800 votes. The starkest example of the trend is Tredyffrin Township (76%). Romney lost it in 2012, but by a modest 600 votes. Trump, on the other hand, was crushed to the tune of 4,500 votes, for a total shift of nearly 3,900 votes in Clinton’s favor. All told, nine Chester County localities swung by four-digits towards Clinton, while over two dozen more shifted by three-digit margins. It all adds up to what was a slight GOP advantage in 2012 turning into a 25,000-vote deficit in 2016.
Completing our counter-clockwise tour of the Philadelphia collar, we come to Delaware County. Obama won it by around 60,000 votes in 2012 and Hillary won it by a little under 63,000. Like Bucks, though, Delco is something of a mixed bag because here too we find four-digit swings in both directions. Obama won Upper Chichester (24% bachelor’s degree or higher) by about 1,000 votes. Trump took it by about 100, a swing of 1,100 votes towards him. Ridley Township (23%) swung by 2,000 votes towards Trump. Obama beat Romney there by 800 votes; Trump beat Clinton by 1,200. There are various other municipalities where Trump improved on Romney’s performance by a few hundred votes. But Clinton benefited from a few that stampeded in her direction. Haverford (54%), which Obama and Clinton both won, saw a net shift of 3,100 votes towards Clinton. Radnor (71%), where Romney managed to keep his deficit under a thousand, saw Trump trounced by almost 4,500, a 3,500-vote improvement for Clinton. Romney barely took 600 votes in Swarthmore (80%), losing it by 2,100. But Trump managed to do even worse, getting just over 400 votes and losing by 2,700. It’s these smaller shifts that let Clinton improve by a couple thousand votes on Obama’s 2012 performance in Delaware County.
Trump did badly in the Philly collar. From the Clinton campaign’s perspective not badly enough, perhaps; and certainly not the 20-30% debacle many polls were forecasting. Yet his margins in the Philadelphia area normally would have been enough to keep the Keystone State in the Democrats’ vaunted “blue wall.” Barack Obama won 690,000 votes in the Philly suburbs in 2012. Hillary Clinton received 729,000. Romney got 567,000. Trump dropped to 550,000. That’s a swing of 46,000 votes to Clinton. Perhaps suburban Philadelphians didn’t hate Trump, but they certainly didn’t like him much. Unfortunately for the former Secretary of State, much of the advantage she accrued in the suburbs was wiped out in Philadelphia itself, a failure Democrats will rue for many years. Obama beat Romney there by 492,000 votes. Trump did only 10,000 votes better than Romney in the City of Brotherly Love (96,000 to 106,000), but Clinton received only 563,000 votes compared to Obama’s 588,000, a net swing of 35,000 towards the Republican candidate. Trump did worse in Philadelphia than John McCain, who lost the state, while Clinton did as well as John Kerry, who won. The 2016 electorate, of course, was not that of 2012, 2008, 2004, or any other election. What Donald Trump demonstrated with this year’s electorate is that a historic turnout by white working class voters could be and in fact was – contrary to the expectations of all but a handful of pundits – sufficient to negate the Democrats’ traditional advantage in the Philadelphia region and put Pennsylvania in the GOP column for the first time since 1988. Whether this lesson can be repeated is an open question, but it is one students of American politics will be studying for many years to come.
Since this post was published, the City of Philadelphia released its final vote tallies for the 2016 election. Donald Trump’s total increased to just under 109,000 votes, while Clinton’s soared to 584,000 votes, an increase of 21,000 and a mere 4,000 short of Obama’s 2012 total. Delaware County also has updated its figures. Hillary Clinton’s margin improved to 67,000 votes, as she did 12,000 votes better than Barack Obama did in 2012. Trump’s total in Delco also rose by 4,000 votes. He now stands 5,000 votes above Mitt Romney there. It should be noted as well that Hillary Clinton’s margin in Montgomery County was 24,000, not 25,000 as I originally wrote. I apologize for the error.
Incorporating this new data yields the following updated results. Clinton’s total in the Philadelphia collar stands at 737,000 votes, an improvement of 47,000 on Barack Obama’s 2012 total of 690,000. Trump’s total rises to 554,000, which is 13,000 short of Romney’s count. This means Clinton’s improvement in the suburbs, which initially stood at 46,000, is now 59,000 votes. The updated totals in Philadelphia also reduce the swing there from 35,000 towards Trump to a more modest 17,000. Instead of coming out of the Philadelphia region only 11,000 votes ahead of Barack Obama, as my original calculations indicated, Clinton came out 32,000 votes ahead. What these updated results tell us, therefore, is that in Philadelphia Hillary Clinton performed more like Barack Obama, and Donald Trump performed even worse than Mitt Romney, than initially believed. In other words, Trump’s victory in Pennsylvania, impressive as it was, was even more impressive than it seemed at first glance.
Trump pulled off the near impossible: he made Philadelphia irrelevant. Whatever one thinks of the man, this feat is worthy of, nay demands, respect.
Varad Mehta is a historian and election analyst at the Decision Desk HQ. He lives in suburban Philadelphia and you can follow him on Twitter.