The Urban-Rural Divide

The Urban-Rural Divide

Why Geography Matters
Jonathan Rodden [1.16.19]

In the past, it was dispersed rural interest groups who favored free trade, and concentrated urban producers who wanted protection for their new industries. Now, in the age of the knowledge economy, the relationship has reversed. Much of manufacturing now takes place outside of city centers. Ever since the New Deal and the rise of labor unions, manufacturing has been moving away from city centers and spreading out to exurban and rural areas along interstates, especially in the South. In an era of intense global competition, these have now become the places where voters can be most easily mobilized in favor of trade protection.

Moreover, much like manufacturing in an earlier era, the knowledge economy has grown up in a very geographically concentrated way in certain city centers. These are the places that now benefit most from globalization and free trade. We’re back to debates about trade and protection that occupied Alexander Hamilton and Thomas Jefferson, although the geographic location of the interests has changed over time. Changing economic geography has shaped our political geography in important ways, and contributed to an increase in urban-rural polarization.

JONATHAN RODDEN is a professor in the Political Science Department at Stanford and a Senior Fellow at the Hoover Institution. Jonathan Rodden's Edge Bio Page

THE URBAN-RURAL DIVIDE

Over the last decade, I’ve been asking myself a couple of closely related questions. First, what explains the incredible growth in what we might call urban-rural polarization? What explains the fact that voting behavior has become so highly correlated with population density over time? This is something that’s happened not just in the United States, but also in a number of other industrialized countries. Secondly, what are the implications of this polarization for who wins and loses and whose policy agenda gets put into place?

In a political system like the United States, or Great Britain, or any of the countries that were colonized by Great Britain—including Australia, Canada and, for most of the postwar period, New Zealand—we draw single-member districts, and the way we form a legislature is by electing one individual from each of those districts. When we divide the world in that way, we end up concentrating voters for the urban parties of the left in urban districts, and end up with more dispersed support for the parties of the right elsewhere. This creates an important pattern in the transformation of votes to seats that has existed throughout the postwar period. It affects who gets represented and whose policies end up as law. In politics, geography matters in a way that we’ve been somewhat aware of for a long time, but all of these factors have become much more important over time, especially in the United States.

There was a field of classic British political geography in the 1950s, and scholars were asking these questions and noticing some of these patterns in the immediate postwar period after the Industrial Revolution, but much of this has become more pronounced over time. There’s been an increase in the concentration of electoral behavior over time in the United States. If the first question is what explains that concentration, then the second question is what happens when we draw single-member districts in the presence of that concentration?

To the first question, my inclination has been to look deep into history and try to understand when this first emerges. We can go back as far as we want and find that as long as cities have been around, there have been differences in preferences and ideas about policy between urban dwellers and rural dwellers. There were traders, artisans, and commercial elites in cities, and those living in cities tended to have different preferences than agrarians. This was the case even before the Industrial Revolution, but the Industrial Revolution ushered in a major change in the way this works.

An important moment was in the late 19th century and early 20th century, when industrialization spread to cities throughout Europe and North America. As workers were mobilized by labor unions and associated social democratic, socialist, and workers’ parties, these parties gained strength but became fundamentally urban parties. These parties mobilized workers who tended to be highly concentrated in space. Those workers were concentrated in space in the first place because manufacturing was concentrated in space. There’s a lot we know from economic geography about why that is, including the role of labor- market pooling and the importance of natural resources.

Along with that went the construction of working-class housing. This was a period in which workers did not have access to automobiles. Public transportation was in its early stages, and so cities ended up with very dense working-class housing in close proximity to factories, which is where the new left-wing parties were mobilizing people. Right away, these became urban parties. This is something that happened already in the 19th century and early 20th century in Europe with the rise of labor, social democratic, and workers’ parties. In the United States, this did not happen in the same way. We had some small Socialist parties, but there wasn’t a national party that represented the interests of workers until the late 1920s and early 1930s when the Democrats become that party, especially with FDR and the implementation of the New Deal.

If we look at county-level data, we see that there was no correlation between population density and Democratic voting in the early 20th century, outside of a couple of states like New York and Massachusetts. There was no correlation until the New Deal, and then suddenly a positive correlation emerged. The Democrats became the party of urban workers starting in the ‘30s, and that has continued to the current day.

We’ve seen a slow increase and a spread in that correlation starting in the manufacturing core, in the Northeast—Pennsylvania, Massachusetts, New York, Ohio. Eventually, we see this pattern spread further west to states like Missouri and Kansas. Of course, at that time the South had the opposite relationship. The Democrats were more successful in rural areas—a completely different geographic organization of politics in the South.

What’s fascinating is that this pattern of burgeoning urban-rural polarization— which was already there in the Northeast and the rest of the manufacturing core in the New Deal period—eventually took off and nationalized. It spread to the rest of the country, starting around the 1980s, and there’s been an incredible acceleration in the correlation between Democratic voting and population density since the 1980s, continuing to the present day.

Certainly, 2016 was not a sudden anomaly. We’ve seen a steady growth of geographic concentration of Democrats in cities over time. And that growth has taken place in a steady way ever since around 1980. It has also spread from the manufacturing core to the Mountain West and the South. The South has transformed, of course. It is now a place where Democrats are highly concentrated in cities like Atlanta. Georgia now has a very similar correlation between population density and Democratic voting to what we have traditionally seen in places like Pennsylvania. The nationalization of politics has been an important part of that story.

Of course, the urban-rural conflict in the United States is much older than the industrial revolution; it’s something that was there from the very beginning, and was intimately wrapped up in the controversy over slavery.

In the Hamilton-versus-Jefferson frame of thinking about early American history, it makes sense to see Hamilton as a representative of urban interests, while Jefferson was very clear about his disdain for cities. Jefferson viewed the United States as an agrarian republic. Tension between urban areas that were trying to build up industry and rural agricultural areas was there from the beginning. And the same tension was present in many other countries as well during the same period.

Marx and Engels talked about the kind of conflict that characterized European societies before the Industrial Revolution; it was a description of something similar, where there’s a battle between urban interests and rural landowners and battles over free trade. Those battles are coming full circle, although they’ve changed in an important way.

In the past, it was dispersed rural interest groups who favored free trade, and concentrated urban producers who wanted protection for their new industries. Now, in the age of the knowledge economy, the relationship has reversed. Much of manufacturing now takes place outside of city centers. Ever since the New Deal and the rise of labor unions, manufacturing has been moving away from city centers and spreading out to exurban and rural areas along interstates, especially in the South. In an era of intense global competition, these have now become the places where voters can be most easily mobilized in favor of trade protection.

Moreover, much like manufacturing in an earlier era, the knowledge economy has grown up in a very geographically concentrated way in certain city centers. These are the places that now benefit most from globalization and free trade. We’re back to debates about trade and protection that occupied Alexander Hamilton and Thomas Jefferson, although the geographic location of the interests has changed over time. Changing economic geography has shaped our political geography in important ways, and contributed to an increase in urban-rural polarization.

~ ~ ~ ~

The first part of this story involves the rise of manufacturing and the rise of the urban working class. But very quickly after unionization took place and manufacturing enterprises started leaving city centers, a lot of cities lost population. A lot of working-class housing from that early 20th- century explosion of growth—including apartment buildings, small workers' cottages, and triple-deckers—was left behind. The manufacturing was gone, but the housing generally attracted new migrants and poor people, and this marked the beginning of urban decline in American cities. Many American cities ended up with a concentration of poor voters in the city center. In the New Deal, the Democrats had been the ones to mobilize urban voters, so those city centers had come to be dominated by urban Democratic politicians. The Democrats had become the party of manufacturing cities, and then in the post-manufacturing era they simply became the party of post-manufacturing cities.

Along the way, there was another economic transformation. This story I’ve been telling is a period that many people refer to as the Second Industrial Revolution. But next, we experienced a transformation that is often referred to as the Third Industrial Revolution—the rise of the knowledge economy—which also has a similar logic of geographic concentration, just like the manufacturing revolution. There were lots of reasons why it made sense for manufacturing enterprises to colocate, even sometimes within the same industry. American cities throughout the Northeast provide a tour through the various industrial agglomerations that took off and then declined—tires in Akron, gloves in Gloversville, etc. Often the name of a place is evocative of what was produced there. We had many small, localized agglomerations based on a certain type of economic production. And now once again in the era of the knowledge economy, we have strong reasons why economic production is geographically concentrated.

Many scholars have told the story about Silicon Valley and reasons why it can be very difficult for certain kinds of tech entrepreneurs to locate elsewhere. There are lots of reasons, including the classic story of labor market pooling—the benefits of having a lot of educated workers and employers living in the same area. This creates efficiencies that can be very powerful. The presence of venture capital and all of those things that go together in an environment like Boston, Palo Alto, and San Francisco now creates concentration in the knowledge economy sector. A fascinating and odd story, when you think about, it is that the party of the urban working class from the early 20th century has now become the party of the knowledge economy. And there is no good reason for that other than geography.

We have to understand the extent to which the Democrats became an urban party in the era of manufacturing. Once all the city mayors had become Democratic in places like Massachusetts, the Democrats started to coalesce with universities and the knowledge economy sector in its early days and became the champions of that sector. This only grew over time. Today, the Democrats are still the party of Akron and Scranton, but at the same time, they are also the party of San Francisco and Boston, and these are incredibly different constituencies. The thing they have in common is that they’re urban. The Democrats have become a party that represents an incredibly diverse coalition of urban interests.

This is related not just to the rise of the knowledge economy and the continued presence of Democrats in old manufacturing cities, but also to the rise of a set of non-economic issues in the 1980s. The increasing correlation between population density and Democratic voting that took off after the 1980s, was also driven by the rise of politics related to religion, gender, and the social transformations that came about in the ‘60s and ‘70s and then were politicized in the ‘80s. Before the 1980s, if one was an anti-abortion social conservative, it was not clear whether one should be a Democrat or a Republican. That changed in the 1980s when the parties took sharply different positions on those issues. Preferences on those issues are also highly correlated with population density.

Once we add this additional set of issues, urban-rural polarization comes into full bloom. The parties become increasingly separated in their geographies. The Democrats go from not only being a party of urban workers, but also being a party of urban social progressives, which leads to further sorting of individuals into the parties. Knowing someone’s preferences and whether they call themselves a liberal or a conservative becomes much more predictive of whether they vote for Democrats or Republicans.

There's a real geographic story to that as well. The people who were sorting into the parties in this period were often geographically clustered, which led to a further increase in the correlation between population density and Democratic voting. We ended up with two parties offering a set of policies that are no longer very accurately described as “left” and “right.” It perhaps makes more sense to refer to the two packages of policies as “urban” and “rural.”

We can see on the election-night maps that the Democrats are very clustered in places like Cleveland or Philadelphia. One of the things that people don’t often understand is that Democrats are also highly concentrated in places like Topeka and Wichita. These are overwhelmingly Democratic cities. But they are not very large cities. They’re smaller than the scale of a US congressional district. So, we don’t see Democrats being elected from any of those cities because they’re surrounded and outnumbered by their rural Republican periphery.

This pattern is fractal. We can see that there’s a correlation between density and Democratic voting at the very high level of US states, and then we can see it reappear at the level of counties, and then again at the precinct level within counties. When you go from a very rural area composed of farm fields and very small villages to the larger county seat, where there’s a little bit of rental housing along the 19th-century railroad, maybe some public-sector workers, a courthouse, some unionized schoolteachers—you often go from a majority-Republican area to a majority-Democratic area.

As we go from a rural area to a more urban area, the increase in Democratic voting can be quite steep, even in places that we think of as monolithically rural. Those places that have more rental housing, and young people, poor people, and minorities, appear as small blue dots surrounded by red on the precinct-level electoral map everywhere in the country. There are very few places where we don’t see that pattern.

Once we see these patterns in partisanship, we then have to ask ourselves what's behind them. Why are there such persistent differences in voting behavior? What brings about and sustains that type of difference? Unfortunately, there's no single answer. There’s a lot going on.

One thing political scientists know about party identification is that it can be, in some families, like religion. It can be a thing that gets passed along from one generation to another. In some places, there has probably been something like that going on ever since the New Deal, but that can’t be the entire story.

Part of the story has to do with the legacy of labor unions and institutions that persist in communities. We all know that the private-sector unions have declined tremendously, and in many post-industrial towns, they are almost completely extinct. But the public-sector unions are not. In fact, in many of these larger county seat cities and towns, a large share of the population works for the municipal government. Many of these workers are unionized: for example schoolteachers, court employees, nurses, and other municipal workers. These public sector unions are closely allied with the Democrats. That’s a big part of the story.

Another part of the story is that in some places, the city government has been Democratic since the New Deal, so much so that if you wanted to get a city contract or have a good relationship with town government, the Democratic Party was the place to go. We see patterns of geographic political persistence like that around the world. This creates stable partisanship over time.

The question of what political preferences people have in a very specific geographic place, and how they vary a couple of miles away as we move from city to suburb, is hard to answer because we don’t have fine-grained surveys that allow us to drill down very far, say, within a county and figure that out. But from looking more broadly at the data and characterizing places according to whether they can be described as a small-town core or a medium-sized city core, or a suburb, or a large-city core, or rural, one thing we can see is that people do have very different ideas about economic policy from one type of place to another. Urban places tend to be much more progressive on economic policy, which we can imagine probably has a lot to do with demands for public goods that arise in an urban place. When one lives in a dense place, it becomes clear why public transportation makes sense, where one doesn’t necessarily have that perspective in the exurbs or in a rural area.

It’s also the case that people who live in more dense places tend to be less religious and have less traditional values on issues related to religiosity and morals.

~ ~ ~ ~

I started my academic career focusing on questions about federalism. I’ve always been interested in the relationship between higher-level and lower-level governments, and the trend toward fiscal decentralization around the world that’s been happening for several decades now. This was something at the intersection of economics and political science, and I’ve long been equally interested in both.

What makes federalism and decentralization feel like good ideas is the notion that different people in different places want different things, so if we can introduce a decentralized system of government, we can make it more likely that more of those people get what they want more of the time. And then we also introduce the idea that people can move to other places that offer the bundle of goods and services that they want. That’s a set of issues that, as a graduate student, I got very interested in.

I started studying federalism and was working with Juan Linz, who was a giant in the study of institutions. He got me interested political institutions, and suggested to me the idea of trying to understand issues related to comparative federalism—Germany and the United States in particular—so that’s how I got started.

I got more interested then in statistics, econometrics, and economic theory, so I ended up blending the studies of federalism and economics with those in political science and ended up focusing on fiscal policy. It was a time when a number of countries were experiencing debt crises among state and local governments, and that became the topic I started with. Along the way, I learned a lot about regions and the ways in which their economies are distinctive.

Like any other American, I was looking at maps on election night and noticing some patterns. The thing that piqued my interest was when I started looking at other countries and saw similar patterns. Americans don’t often pay much attention to that sort of thing. If you look at an electoral map of Britain, it has many qualities that are similar to an electoral map of the United States. Labor support is highly concentrated in the old industrial cities, the new knowledge economy cities, and in the college towns like Oxford, and votes for the conservatives are highly distributed in the rural areas. That’s not new at all. This is something that’s been true throughout the postwar period.                 

Canada has become remarkably similar to the United States on this dimension. Votes for the NDP and the Liberals are highly concentrated in city centers, and the votes for the Conservatives are spread out in rural areas. It looks just like the United States. One thing I sometimes do for fun when I’m demonstrating this to people is I make electoral maps of the US and Canada, where I just remove the boundary between the two and don’t allow the viewer to see where Ontario starts and New York or Michigan end. And you can’t tell. It’s the same pattern. Along the Great Lakes, the Democrats in the United States, and the left parties in Canada, are concentrated in the city center, and the Republicans and the Conservatives are spread out in the exurbs and the rural areas. Just seeing those patterns was purely descriptive enterprise at first.                 

I was interested in understanding those patterns, but then I started noticing related patterns in the way votes transform into seats. This is about the same time when we started noticing big differences between vote shares and seat shares of the Democrats in the United States. In Australia as well, one can see that support for the Labor Party has been highly concentrated in the urban working-class neighborhoods of Melbourne and Sydney, where they have traditionally won big supermajorities, while support for the right is more spread out in the rural areas, which allows them to win a seat share that is far beyond their vote share. It’s always been hard for the Australian Labor Party to transform its votes into seats.

The same pattern is still with us. In fact, it’s strengthened in some ways. This is going on in the US as well, but in the US, people see the same pattern and they arrive at a different explanation: gerrymandering. Because gerrymandering is such a unique and fascinating American practice that goes back to Elbridge Gerry and has been continued by people like Phil Burton and Tom DeLay, it is natural to try to understand the ways in which one can manipulate maps to manipulate outcomes. That certainly can and does happen. Gerrymandering is a very important part of the story in the United States. But there is this deeper underlying story that operates even in places where politicians don’t draw the districts— even in places like Great Britain and Canada, where independent commissions have always drawn the districts, or at least for most of recent history. And yet we see the same underrepresentation of the urban parties in those places as well.   

In a parliamentary system, the parties have a relatively easy job when they’re trying to set their platform. There are a number of districts, and the goal is to win more districts than your opponent. What you need to do—if you think of those districts as being arranged in some ideological space from left to right—is try to win the pivotal district, which is the median district. You need to win the district in the center of the distribution. In a parliamentary system, that’s all you need to do. And that’s what the parties try to do for the most part.

This process does create battles. On the left, a challenge is that the urban ideological purists—often those who live in the old working-class neighborhoods and some of these new knowledge economy city centers—can be much more progressive than the voters in those pivotal districts, so there is always a fight on the left about what the parties’ platforms should be.

You see this fight in the Labor Party between Jeremy Corbyn and the more moderate groups. If you go back to the Blair period, this is a time when the Labor Party moderated its platform and it was able to win those crucial suburban districts. Now, they’re in a situation where they’ve gone back to a traditional urban leftist platform that makes it harder for them in the suburbs. They win big majorities in the urban districts and they find it hard to win in those pivotal districts.

The Democrats have the same problem, but there’s a different twist in the United States. A party doesn’t just need to decide how to win the pivotal district in the legislature, they also have to figure out how to win statewide elections like gubernatorial and US Senate elections, as well as presidential elections. We have multiple tiers of competition, so the best strategy for winning the Senate might not be the best strategy for winning that pivotal district in the state's US House delegation or in the state legislature.

When it comes to winning presidential elections, it’s the same thing. It’s a statewide process of counting up electoral votes. As the party strategically tries to figure out what kind of candidates to run and issues to emphasize, it might be possible to win a statewide battle for electoral votes by trying to mobilize progressive urban voters, because a vote is a vote in a statewide election. It doesn’t matter if it came from a city or if it came from a suburban or rural area. A lot of the Democrats’ supporters are concentrated in cities, so they can go and mobilize those supporters and try to win with an urban-based coalition.

We see that in states like Missouri, the Democrats are able to occasionally win Senate seats in large part by turning out their base in St. Louis and Kansas City. In Ohio, it’s quite possible for the Democrats to win Senate seats and even the gubernatorial elections, again, by trying to build from the strength of their urban voters. Trying to win the Missouri or Ohio congressional delegation is a completely different undertaking. The Democrats need to try to figure out how to craft a message that brings them to victory in that crucial, pivotal district, which will often be somewhere in the exurbs of St. Louis or somewhere outside of Columbus, Ohio. Those places tend to be quite a bit more conservative than the Democrats' urban message allows.

If they’ve crafted a message that allows them to win statewide elections, that message can be inefficient for winning in those pivotal suburban places. That’s a problem that the Democrats have to grapple with that the Labor Party doesn’t have in quite the same way. The Labor Party certainly does have a difficult challenge, but the Democrats have an especially difficult challenge because they’re being pulled in two directions because of the presence of statewide elections.

An important observation here is that the ideology of the median voter in many US states is to the left of the ideology in the median district. If the Democrats craft their message to appeal to that statewide median voter, that’s fine for winning Senate seats, but not for House seats. So, the Democrats easily come to be seen as too liberal, and they end up winning only a congressional seat in St. Louis and a congressional seat in Kansas City and nothing else, even though they are often fairly competitive in statewide elections in a place like Missouri.

There are lots of other examples of this, where the Democrats routinely win statewide elections, but they can’t dream of winning the congressional delegation or the state legislature because of the geographic concentration and dispersion of their supporters. The best example of this I can think of is the Pennsylvania State Senate. Everyone thinks of Pennsylvania as a purple state or maybe even, in recent years, a blue state. The Democrats have done very well in statewide elections, for the US Senate, for the governor, for attorney general. In all of the other statewide seats, the Democrats have done very well. They’ve almost swept recent statewide elections in Pennsylvania with the exception of the presidency in 2016 and one of the US Senate seats. But they have not won the Pennsylvania State Senate since the 1970s. A lot of it has to do with the fact that Democrats are too concentrated in their old industrial cities, and they’ve not been able to craft a message that is sufficiently appealing in those pivotal suburban districts. That kind of thing happens in lots of places. The Democrats would like to believe that redistricting reform will do away with this class of problems. There are some states where that may be true, but there are other states where changing the nature of redistricting would only make a small dent in this basic problem. One would have to go out of their way and exert some effort to draw districts that would undo the Democrats' geography problem. Above all, one would need to carve up cities. So, when Democrats do get the opportunity to draw districts, that’s what they try to do, although they have urban incumbents who don’t like it and fight against it.

They also have the Voting Rights Act to contend with, which makes it potentially illegal to carve up urban districts, since that might make it less likely that minority groups can elect candidates of choice. When they are able to draw districts, Democrats might try to do what they did in Chicago, which is draw districts that start at the lake and then extend out into the suburbs in ways that make those districts less overwhelmingly Democratic. That’s something the Democrats need to do in order to get a better transformation of votes to seats, but it’s hard for them to do that. It’s not something that would emerge from a non-partisan redistricting process.

That said, there are also some states where the distribution of Democrats is much less inefficient. This all has to do with the nature of city formation in the 19th century. There are some states where there is a nice, even distribution of cities that are smaller, so they interact with the scale of districts in a way that does not undermine Democratic representation. For instance, think of small industrial cities in the Fox River Valley and around Appleton, Wisconsin. It’s not so bad for the Democrats when state legislative districts are drawn in that kind of environment.

In large cities, the Democrats are highly clustered relative to the size of districts. Then there are also smaller cities where Democratic clusters are perhaps too small to form their own districts, so the Democrats have a serious problem in transforming votes to seats. Using computer algorithms to automate redistricting, we can generate, say, 100,000 redistricting plans for a state and look at the partisanship of those plans. And we can change the criteria to specify that we’d like the districts to be more or less compact; or we’d like to hold counties together as much as possible, or then see what happens if we relax that constraint. There are different ways we’re thinking about how to treat municipal boundaries, but no matter how we do it, in many of these states, across a wide range of simulations, we still end up with an underrepresentation of Democrats because of their urban concentration.

This is a problem that the Democrats will find tricky to solve. For some, the solution will continue to be to try to win and then draw the districts themselves in some of the ways that I’ve described. Others still would prefer to disarm and promote redistricting reform. This will be an interesting battle in the years ahead. When the Republicans were in charge of the redistricting process for the last couple of cycles, especially the last one, many Democrats came to believe that gerrymandering is something that mainly advantages Republicans. That’s the case if the Republicans are the only ones who have the opportunity to redraw the districts. But if the Democrats end up finding themselves in a position to draw the districts themselves in some states, they have a lot to gain by drawing them in careful ways. It will be interesting to see what becomes of all the enthusiasm among Democrats for redistricting reform. It may be that in some states the Republicans have quite a bit to gain from redistricting reform, especially at this moment.   

The good news is that for those who do support redistricting reform, it can be a truly bipartisan endeavor. Reform is popular among Democratic, Republican, and independent voters. It's even possible to make a strong case to Republican elites that they would be very wise to join in the efforts at redistricting reform, because in some states, a neutral, nonpartisan process would be quite good for them. This geographic advantage varies a great deal from one state to another.

A deeper question for the parties is how to position themselves given the asymmetric geography I’ve described. How do the Democrats choose their platform when the efficient platform for winning a statewide race is very different than the efficient platform for winning the median district? This is not a new problem for the Democrats. They’ve been struggling with this in many states ever since the New Deal. The solution they’ve come up with is essentially not to have a platform. It’s to be a party that is many things to many people, and to allow individuals to run in these pivotal districts with very distinctive platforms from the national party platform or even the state party platform. We saw the Democrats returning to that strategy in the recent midterm election, trying to find candidates who reflect the views of local citizens in those pivotal districts, and trying to distance themselves from the national party leadership.

In the United States, we have grown accustomed to a phenomenon that one would never see in a parliamentary system. We have individuals running as Democrats who say, "If you elect me, I won’t vote with the Democrats and I won’t vote for the current leadership of the Democratic Party. I will try to change the party from within." In a presidential system, it’s credible to run on that kind of platform. This is one direction the Democrats might go, to try to find candidates who can credibly distance themselves from the national party but still vote with the national party on some issues some of the time.

All of the problems that I’ve been describing are deep and old problems. The problem of regional sectionalism—the problem that different regions of the country want different things and they can’t agree—was present at the founding, and it culminated in a Civil War.

This type of problem is always with us, and we now have to decide how to solve the current manifestation of it. Federal elections feel like battles between two very different ways of life. Urban residents feel completely left without representation when the Democratic Party loses, and rural residents feel very much the same way when the Democratic Party wins. As a result, some view politics as a winner-take-all, high-stakes, almost existential battle, and that’s dangerous. It’s something that we would do well to try to unwind.

Reducing the intensity of urban-rural polarization is perhaps a more urgent policy goal than instituting small changes in the process through which we draw districts. The most obvious proposal is also the most radical, and it’s one that many Americans will reject out of hand. Some reformers would like to do what Europe did in the early 20th century and what New Zealand has done quite recently, which is give up on the whole business of winner-take-all districts and switch to a system of proportional representation.

Such a reform would, in all likelihood, generate a more diverse party system. It turns out there are lots of Americans and plenty of people in Europe who prefer less government regulation of the economy, but they also prefer less government regulation of things like abortion and gay rights. In European countries, there are liberal parties offering that mix of positions. There are also lots of people who have more traditional preferences on social issues, but who prefer more government involvement in things like healthcare. Again, the United States doesn’t have a party that offers that set of platforms.

How does this make Europe different from the United States? Educated urban voters in the US vote overwhelmingly for Democrats, but such voters have not become exclusive supporters of the old early 20th century labor, social democratic, or socialist parties in Europe. Rather, they’ve become supporters of liberal parties or in some cases, conservative parties that appeal to urban voters with more progressive positions on social issues and support for the knowledge economy.

This leads to a depolarization that would be valuable in the United States. We see that a government of the right in Sweden is not a government of the countryside, it’s not a government that locks cities out of representation altogether. It turns out that much of Stockholm, for instance, votes for a moderate party of the right. And so, a right-wing government in Sweden is basically a coalition between urban voters and rural voters. And the same thing is true of the parties of the left in Sweden. They also have significant support in rural areas. Proportional representation can unbundle the parties’ platforms in ways that would be potentially useful in the United States.

Proportional representation has a number of benefits and costs worthy of debate, but many Americans are unwilling to consider such far-reaching reform. What future might we have if we continue with the current intense two-party competition?

One possibility is that the parties themselves end up unbundling some of these issues. For instance, it is conceivable that a candidate in the Democratic Party could adopt positions related to trade that are more similar to those the Republican Party has recently adopted. Democrats could make more explicit efforts to regain some of the votes they have lost in rural areas in recent years. Such a change in tactics could begin to unwind the correlation between density and Democratic voting. We can also imagine a future in which the Republican Party tries to reposition itself, perhaps after an electoral loss, and attempts to shift its platform in a way that allows it to reengage with certain kinds of urban voters.

Such a change in party platforms is one possibility. Another possibility that is worth thinking about—and it’s a longer-term prospect—is that Americans might change their residential patterns in ways that undermine some of what I’ve described. A common belief is that Americans are always moving to ever-more homogeneous locations and that they’re sorting themselves into neighborhoods where everyone else is like them. Certainly, a lot of that has happened and some of this behavior continues, but if we look at the groups that are most likely to move and where they are moving, there are some interesting countervailing patterns that deserve a lot more attention. Many young Democrats are moving from places that are more Democratic to places that are more competitive, so a lot of the largest, fastest-growing counties are becoming more politically competitive.

Part of this story is that people are moving to places with labor market opportunities, but the presence of affordable housing is also a factor. We see a lot of growth in the populations of cities like Houston and Dallas, especially in sprawling suburban neighborhoods that are becoming more heterogeneous. So, people who are moving are often Democrats, and they are moving to places that are relatively Republican, thus making those places more competitive.

I have told a long-term story about the concentration of people—and left voters—in cities. But Americans are becoming ever less concentrated in cities. With each passing year we’re becoming a more suburbanized society. It doesn’t seem that way if you live in New York or Washington, D.C. and notice all the gentrification and the high demand for housing in the city center. But in most of the country, suburbanization is still happening. City centers are still losing population and people are spreading out in space. And the suburbs are the most politically competitive places in the United States. Urban-rural polarization need not continue to grow. If the trend of suburbanization continues, there is a possible future in which geographic polarization slowly unravels.