Why Cities Lose

Introduction

IN MOST DEMOCRACIES, the path to victory is simple: win more votes than your competitors. For the Democratic Party in the United States, however, this is often not good enough. For example, in the 2012 election, Democratic candidates for the House of Representatives received 1.4 million more votes nationwide than Republican candidates, but the Democrats ended up with only 45 percent of the seats in the House. In 1996, the Democrats also won the popular vote—that is, the total votes cast across all of the individual races—without winning control of Congress. Democrats must win big in the overall popular vote, as they did in the “blue wave” elections of 2018 and 2006, in order to win a majority of seats in the House.

The Democrats’ problem with votes and seats is even more pronounced in state legislatures. Consider the state of Michigan, where it has become commonplace for the Democrats to win the statewide popular vote without winning a majority of seats in either chamber of the Michigan legislature. In 2012, for instance, the Democrats received around 54 percent of the total votes cast in elections for both state legislative chambers in Michigan, but they came away with only 46 percent of the seats in the Michigan House of Representatives, and 42 percent of the seats in the state senate. This has been happening over the last decade in the other states of the industrialized Midwest as well, including Minnesota, Missouri, Ohio, Wisconsin, and Pennsylvania. Most recently, it happened in Virginia in 2017, and once again in Michigan, Ohio, Wisconsin, and Pennsylvania in 2018.¹ Remarkably, as of 2019, the Republican Party has controlled the Pennsylvania Senate for almost forty consecutive years, even while losing the statewide popular vote around half of the time. The Republicans have controlled the Ohio Senate for thirty-five years, during which time Democrats won half of the state’s US Senate elections and around one-third of the gubernatorial elections.

The popular vote is also largely irrelevant, of course, in determining the composition of the US Senate. Democrats have won more votes than Republicans in elections for eleven of the fifteen Senates since 1990, but they have only held a majority of seats on six occasions.² Yet underrepresentation of Democrats in the US Senate is no mystery. It happens because, as a legacy of the bargain made at the Constitutional Convention in the eighteenth century, large Democratic states like California and New York have the same Senate representation as small Republican states such as Wyoming and the Dakotas.

But in Congress and state legislatures, districts are drawn to be as equal as possible in population. For example, Democratic California and Republican Wyoming both get two senators, but California sends fifty-three representatives to Congress while Wyoming sends only one. And within states, legislative districts are required by law to be very similar in population. It is puzzling, then, that Democrats have been able to dominate the national popular vote in presidential and Senate elections since 1990—not to mention party registration and party affiliation as expressed in opinion surveys—while only winning control of Congress for five of the last fifteen sessions. And it is puzzling that there are so many “purple” and even “blue” states like Pennsylvania where citizens routinely elect Democratic senators, governors, and attorneys general, but where Democrats have had little chance of winning a majority of the congressional delegation or state legislature.

For many frustrated Democrats, the explanation is simple: partisan gerrymandering. Republicans gained control of many state legislatures in time for the most recent round of redistricting in the early 2010s, then drew odd-shaped boundaries that packed as many Democrats as possible into a handful of districts that they easily won, leaving the remaining districts with Republican majorities. Armed with sophisticated geospatial software and a large budget, Republican operatives carefully drew maps that distributed Republicans as efficiently as possible across districts so as to win the maximum number of seats.³

There is much truth to this widely accepted account, but it provides an oversimplified and ultimately misleading answer to a complex question. Why Cities Lose demonstrates that the Democrats’ problem with votes and seats goes much deeper, and is far more intricate, than the impact of a handful of political operatives in a room with a computer. Without a doubt, gerrymandering makes things worse for the Democrats, but their underlying problem can be summed up with the old real estate maxim: location, location, location.

In most of Europe, legislators are chosen from large districts with multiple winners, and parties are represented in proportion to their share of the vote. In such a system, the geographic location of a party’s support is not of primary importance. In the United States, legislators are elected from smaller districts where there is a single winner. In such a system—known as “majoritarian” democracy—the geography of a party’s support is extremely important. In many US states, Democrats are now concentrated in cities in such a way that even when districts are drawn without regard for partisanship, their seat share will fall well short of their vote share. It matters a great deal how the districts are drawn, and by whom, but because of where Democrats live, the very existence of winner-take-all geographic districts has facilitated the systemic underrepresentation of Democrats.

To understand the roots of this phenomenon, it is important to grasp the deeper problem that gave rise to it: our highly polarized partisan geography. To see why the contemporary Democratic Party so often loses with a system of winner-take-all districts, we must first comprehend how Democrats and cities came to be synonymous. Why Cities Lose explains the rise of contemporary urban-rural polarization—a trend that is worrisome for the stability and health of American democracy regardless of one’s partisan or ideological perspective—and then reveals how this development has affected political representation.

The rise of urban-rural polarization over the last century is striking. Figure 1 is a plot based on county-level data from one hundred years of US presidential elections, excluding the states of the Deep South. The horizontal axis represents the population density of each county: higher numbers indicate that a county is more urban. The vertical axis represents the share of the presidential vote received by the Democratic candidate. The relatively flat dotted line indicates that in 1916, Woodrow Wilson’s support was no higher in urban counties than in rural counties. After the New Deal and World War II, the dashed line indicates that John F. Kennedy’s vote share in 1960 was strongly correlated with population density. He lost in most rural counties and won solid majorities in most urban counties. And the solid line indicates that by 2016, the urban concentration of Hillary Rodham Clinton’s support was astounding. She lost by large margins in rural counties and won overwhelming majorities in urban counties.

By the early part of the twenty-first century, the Democrats had become an almost exclusively urban political party. From coast to coast, their support is now concentrated in the downtown cores and inner suburbs of cities. Democrats have come to dominate not only large cities like Philadelphia and Pittsburgh, but also medium-sized cities like Reading, Pennsylvania. They have come to dominate not only knowledge-economy hubs like Seattle and San Francisco that are gaining population, but also poor postindustrial cities like Detroit and Akron that are losing population.

Because of this geographic divide, American elections have come to be seen as high-stakes sectional battles pitting the interests and identities of cities and inner suburbs against those of exurbs and the rural periphery. This urban-rural polarization is a serious problem in its own right, but in many US states, it has also created a geographic distribution of partisans that allows Republicans to win seat shares well in excess of their share of the vote. In turn, this asymmetry between votes and seats only further fans the flames of urban-rural sectionalism.

Among wealthy democracies, overt partisan gerrymandering is a uniquely American phenomenon, but urban-rural political conflict is not. While most industrialized democracies abandoned the practice in the early twentieth century, the United States is one of several former British colonies—including Canada and Australia—along with Great Britain itself, that still rely on a centuries-old English approach to representative democracy: forming a national legislature by selecting representatives from a series of geographic winner-take-all districts. Stark urban-rural polarization, along with an underrepresentation of urban parties, has emerged in those countries as well. In Britain and Australia, the support base of labor parties has been overwhelmingly urban since their formation in the industrial era. The same has been true of the New Democratic Party (NDP) in Canada for several decades. As in the United States, Canadian and British elections have become hostile, polarized battles pitting voters in cosmopolitan and postindustrial city centers against exurban and rural traditionalists.

The implications for partisan representation are strikingly similar. If we add up the popular vote for every British parliamentary election held between 1950 and 2017, the Conservative Party has received around 41 percent of the votes cast, and Labour is only slightly behind with 40 percent. Yet the Conservatives have been in power for 63 percent of that period. In recent elections, the Conservatives repeatedly have been able to form governments with only around 37 percent of the vote in part because of the highly effective geographic distribution of their support. The Australian Labor Party (ALP) and its competitors on the right have split the popular vote almost exactly down the middle since 1950, yet during that period, the right has been in power for 68 percent of the time. In Canada, in 2018, populist Doug Ford was elected premier of Ontario, with his Conservative Party receiving 60 percent of the seats in the provincial parliament, based on only 40 percent of the votes. If we look at national and provincial or state elections since World War II in Britain, New Zealand, Australia, and Canada, we find that the party winning the most seats—and forming a government—lost the popular vote on thirty-six occasions. The beneficiary was the party of the right on twenty-eight of those occasions.

The striking underrepresentation of these urban parties abroad can be explained by a problem of political geography that is very similar to the one plaguing the Democrats in the United States. Left party candidates run up excessive margins of victory in their core urban bastions, but they find it difficult to craft a platform that will allow them to win in the crucial suburban constituencies. As with the Democrats in the United States—who seem perpetually at war with themselves over the soul of the party—this creates enormous tension within the parties of the left in Commonwealth countries. Above all, the urban ideologues do battle with the suburban pragmatists. And in recent years, in both the United States and the Commonwealth countries, there is an emerging fault line between successful global cities and declining postindustrial ones. However, these left-leaning parties of the Commonwealth face an additional problem that Democrats in the American two-party system do not: this tension often fuels temporary and even long-term splits between multiple parties of the left.

In short, political geography has not only disadvantaged mainstream parties of the left, including the Democrats, but it has also made them into unhappy families. One side of the family pushes the cosmopolitan agenda of the global cities and their suburbs, and the other side pushes the “old left” agenda of the struggling manufacturing towns. Moreover, urban stalwarts from safe seats butt heads with suburban pragmatists. If urban progressives are able to wrestle control of the party, they are likely to end up with a platform that allows them to win impressive victories in cities while falling short in the pivotal districts required to form legislative majorities. And if the suburban moderates gain control, the urban ideologues might revolt and form their own party.

The experience of Britain and its former colonies, where electoral districts are drawn by independent commissions, belies the notion that the Democrats’ representation problem is merely a matter of gerrymandering. In fact, throughout the postwar period, underrepresentation of the urban left in national legislatures and governments has been a basic feature of all industrialized countries that use winner-take-all districts. And these countries have produced significantly more conservative policies in the long run than the countries that long ago adopted more proportional forms of representation.

Why, then, has the underrepresentation of the Democrats in Congress only become apparent in the United States in the last three decades? This book will show that when we look at presidential elections, Democrats have been more geographically concentrated than Republicans since the New Deal. Initially, this pattern was also visible in congressional elections, but it vanished from the 1960s to the 1980s because of the highly localized nature of American electoral competition. For several decades, self-styled Democratic candidates were able to distance themselves from their party’s presidential candidates and win in Republican-leaning districts.

The seeds of the Democrats’ current problem were sown in the late nineteenth and early twentieth centuries, in the era of iron, steel, and steam power. As modern cities were taking shape, masses of peasants, immigrants, and freed slaves migrated from the countryside and from abroad to take industrial jobs, and labor unions began to mobilize urban workers, at first for socialist parties, and eventually for the Democrats. In the most industrialized states, a correlation between population density and Democratic voting first emerged during the New Deal era. By the 1940s, Democratic presidential votes in the North were concentrated in cities. And to a surprising extent, votes for Democrats today are still concentrated in the triple-deckers, apartment buildings, and workers’ cottages built in close proximity to the factories, smelters, warehouses, ports, rail hubs, and canals of the late nineteenth century. In much of the United States, a map of the nineteenth-century railroad network is a map of Democratic voting today.

But the transformation of the Democrats into a truly urban party was delayed by their long-lasting coalition with rural Southern segregationists, which began to fray during the civil rights era and was only fully severed in the 1990s. And even outside the South, for a period lasting from the 1960s to the 1990s, Democratic candidates for Congress and state legislatures were able to successfully eschew the party’s national reputation as an advocate for urban labor and socially progressive ideas. They won not only in cities, but in quite a few exurban and rural districts that consistently voted for Republicans in presidential elections. Democratic candidates—often given colorful monikers including “boll weevils” and “blue dogs”—crafted their own idiosyncratic local brands and fought to bring earmarked expenditure projects to their constituents.

Those days are over. Largely in response to their northern urban base, the Democrats have taken progressive positions as a party first on race in the 1960s, and then on social issues like abortion, gender, and sexuality in the 1980s, and more recently, immigration. Candidates who did not embrace those views were pushed aside, and voters who did not embrace them found their way to the Republican Party. Voters’ preferences on these issues are highly correlated with population density. As voters sorted into the parties on these issues, the correlation between population density and Democratic voting grew, and idiosyncratic nonurban Democratic legislators dwindled to the point of near-extinction.

In addition to social issues, the story of growing urban-rural polarization in the United States depends crucially on two striking trends in economic geography. First, manufacturing activity departed long ago from the rail and water-borne transit nodes in city centers and moved to exurban and rural places along interstates, especially in the South. Democrats have maintained their strength in the places where manufacturing is a distant memory. But as part of this transition, Republican candidates have been replacing Democrats as the advocates of the embattled traditional manufacturing sector.

Second, economic dynamism and jobs have become increasingly concentrated in a relatively small number of cities. Educated workers have clustered in global cities like San Francisco, Seattle, and Boston, where incomes—and the cost of living—have soared, while much of the rest of the country has been left behind. Democratic state and local officials had been ensconced in these cities since the old manufacturing days, and it was local Democrats who, in close collaboration with universities and entrepreneurs, became advocates for the nascent globalized knowledge-economy sector.

Largely because of their preexisting urban base, the Democrats have become the party not only of poor postindustrial service workers and racial minorities, but also of social progressives, and now, incongruously, they are the party of science, technology, and the globalized knowledge economy. Responding to their exurban and rural base, the Republicans have become a party that emphasizes not only low taxes and less economic regulation but also gun rights and traditional social values along with the interests of traditional manufacturing, natural resource extraction, and agriculture. Remarkably, in the era of Donald Trump, in response to a segment of their exurban and rural manufacturing base, the Republican Party has embraced trade protection.

These odd bundles of policies came together because of economic and political geography. The Democrats, quite simply, have evolved into a diverse collection of urban interest groups, and the Republicans into an assemblage of exurban and rural interests. Soon it might be useful to dispense altogether with the terms “left” and “right,” which have lost much of the meaning they had in the early twentieth-century context of class conflict. In the United States, the party of the left is now strongly allied with well-compensated urban professionals and venture capitalists who benefit from global free trade, and the party of the right rails against trade and global capitalism. It might be more accurate to refer to the main dimension of political conflict in the United States as “urban” versus “rural.”

This transformation in the structure of democratic politics is not unique to the United States. Many other advanced industrial countries also experienced turbulent social changes related to gender and sexuality in the 1960s and 1970s, and in response, parties staked out opposing positions on a variety of noneconomic issues. The concentration of social progressives in city centers is ubiquitous, and in many countries, parties of urban labor eventually adopted socially progressive positions.

Likewise, many other countries are also seeing knowledge-economy jobs cluster in affluent global cities. In countries like Britain and Australia, urban labor parties have similarly become the parties of educated, high-income, cosmopolitan knowledge-economy workers. Like the Democrats, these parties have become diverse and deeply conflicted coalitions between cosmopolitan urban elites and the denizens of the old working-class neighborhoods. And as in the United States, the backlash against the negative side effects of globalization, including wage stagnation and the loss of manufacturing, has been exploited primarily by the exurban and rural parties of the right. Something similar has happened in countries ranging from Austria and Hungary to Italy and France.

But while urban-rural polarization is on the rise around the industrialized world, it may be especially consequential in the United States. A key argument of this book is that cities lose only when this type of geographic polarization is combined with an old-fashioned system of winner-take-all electoral districts like the one in the United States and in the British Commonwealth. Moreover, this geographic conflict is sharpened in the United States by a uniquely rigid two-party system. In countries with proportional electoral systems and multiple political parties, the introduction of issues like abortion and immigration, and the nationalist anti-elite backlash to globalization, do not encourage political elites to organize so clearly into only two mutually hostile, geographically defined camps. And in those countries, political geography does not undermine the representation of urban voters. In the United States, elections have unfortunately come to be viewed as winner-take-all battles between different sectors of the economy and dissimilar ways of life.

WHY CITIES LOSE proceeds in three steps. First, it explains the origins of urban-rural polarization in the United States and beyond, drawing on insights from urban economics and political science and data from a wide variety of sources. It brings the story to life through the example of the state of Pennsylvania, and in particular, the city of Reading—a classic case of industrial rise and decline giving birth to the geographic concentration of Democrats. The first section of the book establishes a strikingly similar pattern of political geography not only in and around American cities, but also around Canadian, British, and Australian cities. It also explores subtle differences in this pattern of political geography, distinguishing between postindustrial cities, knowledge-economy cities, and sprawling new auto-oriented cities.

Second, the book explains how these patterns of political geography have come to undermine representation of the Democrats, as well as of labor parties in Commonwealth countries. Once again, to get a clear sense of the argument, it is useful to dwell in some detail on the state of Pennsylvania, where a high-stakes battle about political geography, gerrymandering, and representation recently took place in state court. The Pennsylvania case study sets the stage for the analysis of other states and ultimately other countries. An important conclusion is that while gerrymandering has allowed Republicans to build upon their geographic advantage, in some states, artful gerrymandering might also be the only way for Democrats to overcome that advantage. While many Democrats like the idea of a party-blind redistricting process that produces geometrically compact districts, such a process would actually be quite beneficial to the Republicans in a number of competitive states. Another lesson is that since progressives are concentrated in cities, the Democrats’ geography problem goes well beyond the mechanics of votes and seats. In many states, the pivotal voter in a statewide election is more progressive than the pivotal district in the congressional delegation or the state legislature. As a result, if the Democrats set their sights on winning statewide races and presidential elections, they are tempted to adopt platforms that will mobilize urban turnout that helps them win the popular vote, even though these same platforms weigh heavily on their candidates in crucial suburban districts. In today’s era of highly polarized and nationalized politics, it is difficult for the Democrats to craft the type of moderate policy reputation that allows them to win Congress and state legislatures absent an idiosyncratic electoral wave in their favor. Labor parties in Commonwealth countries have faced a similar problem for almost a century.

Finally, Why Cities Lose explores prospects for the future. The problem of urban-rural polarization is especially troublesome for Democrats because of its implications for representation in Congress and state legislatures, but it has high costs for everyone. When Republicans win control of the presidency and Congress, the interests of American cities—other than perhaps anticommunist Cubans in Miami’s Little Havana—are not represented. And when Democrats win control, most of rural America—outside of Native American reservations—feels frozen out. Ideally, a democratically elected national government would have incentives to consider policies that are in the national interest, rather than to fight for one geographically defined bundle of interest groups at the expense of another.

There are several avenues for the potential reduction of urban-rural polarization. The United States could follow the lead of European countries and, more recently, New Zealand, and replace the system of geographic winner-take-all districts with a more proportional electoral system. Perhaps the most attractive feature of the multiparty systems associated with proportional representation is that they combat the tendency to force all political competition into one overarching urban-rural conflict. Alternatively, even without serious reforms, a combination of demographic trends and political incentives might ultimately lead to a self-correction that lowers the temperature of urban-rural polarization. It is tempting for some Democratic candidates to double down on their progressive urban core voters in pursuit of the presidency and statewide offices. Nevertheless, to win and retain power in Congress and state legislatures, there is countervailing pressure for Democrats to return to their historical pattern of heterodox, locally tailored platforms outside of city centers.

The Republicans may also be forced to change their strategy. Educated suburban areas have moved sharply toward the Democrats in recent years. Moreover, as urban African Americans move to the suburbs and educated young people migrate in search of opportunities, the newest and fastest-growing suburban and even exurban areas—many of them in the Sun Belt—are becoming more politically competitive and trending Democratic.

In the medium term, if neither reform nor demographic change is sufficient to diminish urban-rural polarization, the United States must find ways to function as a divided society. Neither party will find it easy to shake off the influence, or reputation, of its geographic base. During periods of Republican control, many voters in cities and knowledge-economy suburbs will continue to view the federal government as a hostile power aiming to undermine their interests; voters in many exurbs and rural areas will feel the same under Democratic administrations. Perhaps it is fortunate, then, that the United States has a tradition of federalism that empowers state and local governments to cater to local majorities. Urban Democrats have discovered a new fondness for states’ rights and municipal autonomy during the recent period of unified Republican federal control. And Republicans will surely rediscover their traditional affection for federalism and local control when the Democrats return to power. If dysfunction and geographic sectionalism continue to prevail at the federal level, voters and interest groups on both sides will look increasingly to state and local governments for action.