Like most people who write about politics, I got the last presidential election wrong. I was pretty sure Hillary Clinton would win. But I was right about one thing: I thought she needed a big advantage in the popular vote to prevail in the Electoral College. I guessed that she needed about 5 percent, and that she would get that and become president. She ended up with 2.1 percent, and it wasn’t enough.
Why wasn’t it enough? The answer became indisputably clear in the first days after the election. Millions of Clinton votes were piled up in enormous clusters she didn’t need in states where her victory was a foregone conclusion. Clinton won California by 4.4 million votes, New York by 1.7 million, and Illinois and Massachusetts by nearly a million each. California alone dwarfed her national popular vote margin over Donald Trump of 2.87 million.
A couple of years later, it seems safe to predict that a narrow lead in the national popular vote won’t be enough to elect a Democratic president anytime in the immediate future. What doesn’t get so much attention is the existence of the same phenomenon in elections farther down the ballot, for Congress and especially for state legislatures. It is a case of clustering in cities, to the Democrats’ distinct disadvantage.
Stanford University political scientist Jonathan A. Rodden has written a book, Why Cities Lose, that documents just how important a phenomenon clustering is at virtually all levels of the American political system. When it comes to legislatures, Rodden writes, Democrats “win by excessive margins in the districts they win and fall short by relatively narrow margins in districts they lose. ... The Democratic long-term problem is with seats, not votes.”
The most striking case Rodden cites is Pennsylvania. Between 2012 and 2018, Democrats won 15 of 18 statewide elections. But they never came close to a majority in the legislature. In 2016, when Donald Trump squeaked through in Pennsylvania by 0.7 percent of the popular vote, Republicans won 60 percent of the seats in the Pennsylvania House and an even bigger share in the state Senate. They also returned to power in the lower chambers in Michigan, North Carolina and Wisconsin that year, despite taking fewer votes overall.
There is, of course, a simple explanation for this lopsided state of affairs: gerrymandering. There’s no denying that it has made a difference. Republicans have controlled the drawing of legislative districts in Pennsylvania for the past two decades, and they have drawn them to their advantage. But Rodden thinks gerrymandering is significantly overrated as an interpretation. He notes that even in the blue landslide year of 2018, the Democratic Party’s aggregate performance in winning seats in the Pennsylvania General Assembly lagged far behind its overall statewide vote.
To make sense of all this, Rodden goes back more than a century in American political history. He looks to small Pennsylvania cities such as Altoona, Johnstown and Reading. All of these grew up as factory and railroad towns settled largely by immigrants from Italy and Eastern Europe. He examines the vote along the rail lines that stretch out from the centers of each of these cities. Somewhat to his surprise -- and very much to mine -- Rodden finds that most of these factory towns, in Pennsylvania and other nearby states, still delivered for the Democrats in 2016. They were simply outvoted by Republicans in the further reaches of their counties, both for president and for seats in the legislature.
The most important reason for this, though, isn’t history. It’s demographic change. The industrial cities that have remained Democratic are not home to the same cohort of ethnic residents that provided reliable support for Franklin Roosevelt. They are increasingly the home base of more-recent immigrants, especially Hispanics. Reading is now 60 percent Hispanic. In the larger Pennsylvania and Ohio cities, Philadelphia, Pittsburgh and Cleveland, the minority vote is augmented by a growing presence of highly educated white professionals, the “creative class” that Richard Florida writes about, and by the residual strength of public employee unions. In those cities, Democrats do win their share of state legislative seats. But in most of the smaller, older, less prosperous industrial cities of middle America, this patchwork urban coalition is not enough to match the tide of Republican votes from exurban and semi-rural precincts.
Democrats would be strong enough to win more elections in states like Pennsylvania if they were distributed differently. But they huddle together in urban clusters that squander their electoral influence, much as they do at the statewide level in California and New York. They are packed together much more densely than Republicans are. In 2016, Clinton won two districts within the town of Reading by overwhelming margins. But she lost everything else in Berks County to Trump, who carried it by 18,000 votes.
“Democrats,” Rodden says, “are more likely to live in homogeneous Democratic neighborhoods, and Republicans are more likely to live in mixed neighborhoods.” This is pretty much a statewide reality in Pennsylvania. In 2008, a quarter of the Democrats there lived in neighborhoods that were at least 70 percent Democratic. It’s a recipe for winning a couple of battles decisively in each city while losing virtually everything else around it.
All of this is made worse, from the Democrats’ point of view, by the increasing ideological rigidity of their national party. A generation ago, Democrats won nearly 60 percent of the seats in state legislatures across the country -- and comfortable majorities in the U.S. House -- by fielding candidates who matched the sensibilities of their districts’ voters rather than the national party consensus. Some of those who made it to Congress were Southern conservatives who voted with Republicans more often than they did with their own Democratic colleagues; some were so-called “blue dogs” who catered to rural or small-town Midwestern preferences; some were simply clever liberals who knew how to push the right buttons at home even if they were loyal to the Democratic leadership most of the time in Washington.
It is very difficult to play that game now, either in a congressional district or in a state legislative constituency. More than ever, the Democratic Party is a monolith, in the state capitals as well as in Washington. The opportunity for mavericks or renegades to succeed is not what it was. This is not a good way for a Democrat to get elected to the state Senate from a district on the outskirts of Reading or Altoona.
Democrats don’t have to do that, of course. They could loosen up on their ideological purity and support more candidates who could tailor their message to the suburban vote. They could tell their nominees in places like suburban Berks County that if they want to be pro-life, that’s acceptable, and if they support gun rights, that’s permissible as well. We did see some of this at the legislative and congressional levels in 2018, not because of any national strategy but simply due to the circumstances of the year itself. Moderation wins elections outside the central cities, and it can win back lost legislatures. But it isn’t the direction Democratic activists are moving in, either in presidential campaigning or in contests down the ballot. It’s hard to imagine that changing anytime soon.
The most intriguing idea is a move to proportional representation, reducing or abandoning winner-take-all in small legislative districts and giving each party a share of legislative seats roughly equal to its percentage of the vote in a much larger geographic region or even an entire state. Rodden demonstrates that in all the western democracies that use proportional representation, parties of the left have been in power for most of the last half-century. They have spent heavily on health care, old-age pensions, unemployment insurance and affordable housing. Proportional representation, Rodden asserts, “is a boon not only for cities, but for the political agenda of the urban left.”
There would be some non-ideological advantages as well. A proportional representation system usually generates multiple parties offering widely different agendas. With six or seven parties appearing on the ballot, an ordinary voter would have a better chance of finding one that matched his or her own combination of preferences. Entire states would become battlegrounds -- a lonely Republican vote in Democratic Reading would still contribute to the GOP totals in the legislature.
But there are reasons why most Americans are suspicious of proportional representation. If they are old enough to remember the 1950s, they probably recall the multiparty chaos that bedeviled France and Italy in those years, with coalitions made and unmade week by week and prime ministers moving in and out of office with every political breeze that blew. Those too young to remember the ’50s might cast a glance to present-day Israel, where the squabbling of tiny ideological parties can prevent any coalition from governing at all for weeks at a time.
It’s a fascinating subject for debate, but for the United States at least, it’s also a moot point. America isn’t ready for proportional representation. Democrats -- and cities -- will have to find some other strategy for lifting themselves out of their electoral ditch.