Our geographic divides are central to contemporary politics, including the election of Donald Trump. Election maps show dense liberal cities in a sea of sparsely-populated Red, advantaging Republicans in our geographic electoral system. Why are Democrats concentrating in cities?

Jonathan Rodden finds increasingly concentrated left parties around the world, disadvantaging liberal cities. It started with unionized industrial railroad hubs but accelerated with the changing cultural values of the party’s new coalitions. Will Wilkinson finds urban and rural areas are becoming economically and psychologically distinct, with cities concentrating those open to new experience and working in the technology-driven economy and rural areas retaining those averse to social and economic change.

Studies: “Why Cities Lose” and “The Density Divide

Interviews: Jonathan Rodden, Stanford University; Will Wilkinson, Niskanen Center


Matt Grossmann: This week on The Science Of Politics, explaining the urban-rural divide. For the Niskanen Center, I’m Matt Grossmann.

Election maps are showing stark divides between liberal cities and conservative countrysides, advantaging Republicans in our geographic electoral system. Why are Democrats concentrating in cities and how do the US trends compare to the global patterns? Today I talk to Jonathan Rodden of Stanford University about his new book, Why Cities Lose. He finds increasingly concentrated left parties around the world, disadvantaging liberal cities in political competition. His explanation draws back to unionized industrial railroad hubs, but he finds that today’s growing divisions reflect the changing cultural values of the parties’ new coalitions.

I also talked to our own Will Wilkinson of the Niskanen Center about his new report, “The Density Divide.” He finds that U.S. geographic areas are becoming economically and psychologically distinct, with cities concentrating those open to new experience and working in the technology-driven economy and rural areas, retaining those averse to social and economic change. They both find our geographic divide central to contemporary politics, including the election of Donald Trump. Wilkinson says urbanization and geographic polarization help explain where we are.

Will Wilkinson: I think the biggest takeaway the paper is that we’ve largely overlooked the extent to which the long trend of urbanization, glacial shift of the population slowly drifting from the country to the city, sort of overlooked the extent to which that may have segregated our population along the lines of attributes that predict whether people are going to urbanize or not, things like ethnicity, personality and level of education. So consequently, I think we may have overlooked the extent to which the current era of hyperpolarization and the rise of populist nationalism is partly a consequence of this self-selected spatial segregation on these attributes. So what I call the density divide in the paper refers to the cultural and political polarization along the lines of population density. I know you’re going to be talking to Jonathan Rodden. I draw pretty heavily on his work and it’s one of the inspirations for this paper.

He shows an incredible relationship between population density and party vote shares. Just the more dense a place is, the more Democratic it is. The less dense it is, the more Republican it is. You see that pattern kind of cropping up more or less everywhere and at pretty much every scale. There’s something incredibly striking about that. I was very much struck by the returns from last election, from the 2016 presidential election, to see just how few counties Hillary Clinton had won compared to Donald Trump, but how many more people that the Clinton counties contained and how much more of national economic output they produce. There’s a puzzling pattern there that I think needs to be explained. So the linchpin of my argument is the hypothesis that the attributes that explain patterns of urbanizing migration also increasingly account for both party affiliation and growing regional economic divergence. These things are just pulling high-density cities and lower-density exurbs and rural areas further and further apart culturally and politically.

Matt Grossmann: Rodden’s book says urban-rural divides started for different reasons than they are continuing. But the trends have big implications.

Jonathan Rodden: Sure. The book has, it has two components that are reflected in the title and the subtitle of the book, the big question of why cities lose is a second part of the book. But the first part of the book is reflected in this subtitle of Deep Roots of the Urban-Rural Political Divide. So the book has these two parts and the first one is trying to understand where did this urban- rural polarization that we see in American politics today come from? Although I focus a lot on the United States, it’s really a broadly comparative book and it’s asking a similar question about a lot of other industrialized societies, focusing a lot on the US, though. I argued that this urban-rural divide really started in the industrial era, and started in the late 19th century and the early 20th century, in a time of working-class mobilization when parties that became parties of the left started forming alliances with labor unions and with workers. That activity was largely concentrated in cities.

So this is how left parties start to become urban parties, but that’s not where the story ends. It really … just kind of where it begins. The correlation between urban residents and Democratic voting in the United States really increases a lot in the ’80s and ’90s, a time when labor unions are on the decline. So there’s a story about the rise of new issues and the reorientation of the parties around a set of new issues. And people’s preferences on those issues are correlated with population density, so that the parties kind of, as they latch onto new issues and new groups over time, the Democrats become increasingly urban. The Republicans become increasingly rural. So that’s the first section of the book, is explaining how all that works and some of the variations on that theme.

The second big part of the book tries to explain the implications of that for representation. Short version of the argument is that you end up with left voters, Democrats, in the US who are highly concentrated in space and they’re concentrated then, once we draw districts, they’re are concentrated within electoral districts. That’s something that is only furthered by gerrymandering. So that makes it hard for them to transform their votes into seats. Another thing it makes it hard for them to do is to come up with a good platform that helps them win in the pivotal districts. It also sows the seeds of a lot of internal division within left parties. I trace out some of the current divisions that are plaguing the Democratic Party today and argue that something that’s really quite common to left parties in industrialized societies with majoritarian, winner-take-all districts of the kind that the US has.

Matt Grossmann: They both started with an interest in economic patterns where cities are gaining.

Jonathan Rodden: I think a lot of it started with an interest on my part with economic geography and reading things like Alfred Marshall and then the kind of new economic geography starting with Paul Krugman and others thinking about agglomeration effects. I was really fascinated by the idea of agglomeration effects in economics and then thinking about that in the era of heavy industry and manufacturing, and then thinking about that in the current moment of agglomeration in the knowledge economy. And it just seemed like there had to be some political implications of that. There has to be when … especially if those activities start to map onto the party system then the concentration of economic activity has to have some kind of a implication for politics and trying to figure out what that might be. I think that’s really where I started, at the same time that we’re all looking at these electoral maps on election night and kind of puzzling over those, and kind of putting those things together is the origin of this book.

Matt Grossmann: Wilkinson elaborates on the economic advantages of cities.

Will Wilkinson: The divergence of urban and more rural economies is a pretty longstanding trend and it just keeps getting worse and worse and it just doesn’t seem like it’s going to get any better anytime soon. But over the long term, this shift, this increasing concentration of economic production in cities, is a function that’s a cause and an effect of urbanization. It’s part of why urbanization is happening in the first place. In the first instance, it’s just a shift from a pretty thoroughly agricultural economy to a more manufacturing-based economy, which requires that you have a large concentrated workforce. It all needs to be … The suppliers of all the inputs need to be fairly close together. They need to be near routes of shipping and transit. Manufacturing economies tend to strongly concentrate.

Over the past several decades we’ve shifted even further away from an agricultural economy and away from a manufacturing-based economy, or at least an economy that’s based in manufacturing employment, toward what people call the information economy and knowledge work. And so our economy is more and more dominated by very highly skilled, highly educated workers and not a bunch of less skilled, less educated service workers.

That has increased the concentration of economic production a little bit counterintuitively. A lot of the information and communication technology responsible for the current phase of our economy seems like … It facilitates the sort of thing we’re doing now, which is we’re talking to each other from hundreds of miles away. I work most of the time from Iowa. But these technologies didn’t really produce the death of distance. They actually amplified the advantages of clustering people closer together. And part of the reason is that there’s a skill bias to technology, so the productivity of better educated workers is augmented more by each new technological development. But the productivity of those people is enhanced yet further by being near other people with similar skill sets.

So you get efficiencies from clusters of specialized, educated workers. Economist call it agglomerative efficiencies. You get a higher rate of individual production, but you also get sort of spillovers from … Growth is driven more and more by just people coming up with new ideas, but through innovation, not just increasing the output of every worker per second, right? It’s coming up with new ideas of new products, new methods of production, everything that goes into the economy. You get those ideas faster when you have the smartest people all butting heads together day by day.

This whole kind of dynamic is explained extremely well by Enrico Moretti in his book, The New Geography Of Work. Part of that story that he tells is a sorting story, a geographic sorting story, about educated workers, that as economic output and growth depends more and more on a educated workforce but the wage bonus for a higher level of education keeps going up and up in the agglomerations of specialized workers, it draws those better educated people to those clusters and out of the rest of the economy. And so those places get a lot richer and everywhere they’ve moved away from kind of goes through a sort of negative feedback loop into stagnation and decline. So you get this big separation of these different parts of the economy.

Matt Grossmann: Rodden wanted to challenge traditional explanations for the divide that are built only on movers.

Jonathan Rodden: People just sort of assume that when they see a lot of Democrats concentrated in cities and a lot of Republicans in rural areas, the word people use to describe that phenomenon, they just describe it as sorting. So when we talk about geography versus gerrymandering as an explanation for the transformation of votes to seats, people sort of used sorting as a shorthand for this geography argument. So I think a lot of that comes from a very influential book by Bill Bishop called The Big Sort. The sense is that people from rural Nebraska who had different views than everybody else in rural Nebraska kind of picked up and moved to Omaha or they moved to New York, and that kind of movement is what creates the political pattern we see.

Jonathan Rodden: I don’t dispute at all that mobility is part of the story. I think it might even be a fairly important part of the story, but I think there is something else. That can’t possibly explain everything that we see. There’s just not enough movement. Not enough people actually move, and fewer people are moving over time, and especially during the period of rapid urban-rural polarization. I just don’t think we see the kind of population movements we would need to see for that to be the explanation. So I argue that it has also to do with the kind of sorting that takes place of people into the parties, even people who don’t move. So that as rural areas have become a lot more Republican, it’s not just the case that people who have left-leaning preferences among them pick up and leave, but it’s that there’s been a general reorientation of rural places toward the Republican Party and people within cities have become more and more likely to vote for Democrats, even those that have not moved.

Matt Grossmann: Wilkinson was unsatisfied with the explanations for Trump’s election.

Will Wilkinson: The immediate inspiration for the paper was just the election of Donald Trump, which you know, came relatively closely on the heels of Brexit, which I think caught a lot of people by surprise. Those two things have inspired it, a huge amount of interest in the rise of populism, how polarization contributes to the rise of populism, the increasing salience of parties that are organized around a strong sense of ethnonationalist identity. And so I, like a lot of people, was just like, “Why the heck is this happening? How did we get this guy as our president?” That was the first thing. That was something that I wanted to have explained. I wasn’t finding a lot of the stock explanations that came out right away that persuasive, that it’s a backlash against political correctness or it’s a backlash against a demographic change and immigration or it’s people reacting to the immiseration of neoliberal globalism or whatever. None of those seem to be hitting at the phenomenon at the right level for me.

Matt Grossmann: So how did the urban rural divide get so large? Rodden starts with industrialization, where northern cities first became liberal.

Jonathan Rodden: Yeah. There’s kind of a growing literature in economics and political science looking at these long historical legacies. And this is one that I’ve been kind of puzzling over for quite a while. The correlation between the location of 19th century rail nodes and the proximity to 19th century railroads and current democratic voting is really pretty striking. So I’m viewing those railroad nodes from the early 20th century as essentially kind of a proxy for where the factories were in the era of industrialization, which then goes along with this early labor mobilization. But it also goes along with a number of other things including the built environment, so the growth of a lot of rental housing, dense affordable rental housing, and proximity to those old factories. Once you build those things, they become very resilient and they’re still there.

Jonathan Rodden: So kind of clustered along those old rail lines are a lot of a dense housing units that are still affordable and they still attract young people and immigrants and in some cases minorities. They have a long legacy of that kind. So some of the places that had this early industrialization and now have very Democratic patterns of voting, the type of population that lives there is completely different. I use the example of Reading, Pennsylvania in the book, a place that is now essentially a Spanish-speaking city.

But there are other places where the population has changed a lot less and we might think about a story more like the passing of party ID almost like religion from one family to another. I think there are some places where you can still see some of that. Some of it might actually be through institutions like labor unions, even though they’re quite a bit weakened. The places where public sector, I mean, where private-sector unions were strong in the early 20th century are places that today have especially strong public sector unions. Public-sector union members tend to be overwhelmingly Democratic. So there are a few different mechanisms there that might link the distant past with the present.

Matt Grossmann: The urban-rural divide spread from the north, Rodden says, with racial and religious change.

Jonathan Rodden: I didn’t realize how pronounced the urban-rural divide already started. It was emerging in a place like New York quite early. New York and Massachusetts, it was already around even before the New Deal and maybe had a different sort of … had roots more in ethnic politics and the nature of machine politics in those places. But in the industrialized places, once the New Deal hit, it really … The urban-rural cleavage started to emerge. But it was kind of … It was really limited to those places.

One of the interesting things that happens is that, in the New Deal era, in order to get the new deal agenda passed, the Democrats formed this fascinating coalition between urban workers and Southern segregationists. And the Democratic Party is very different things to different people in different places. But the story that Eric Schickler tells in his book, that I think really is what starts to change this, is that urban workers in the North, many of them are African-Americans and they start pushing the Democratic Party to take up the civil rights issue, and the Democratic party slowly adopts a new set of platforms, and over a long period of time, starts to lose the South. And we have this long period of realignment in the South, at the end of which… Then there’s also this introduction of the religious dimension, which is highly correlated with population density.

And as all of that emerges in the ’80s and ’90s, the South starts to become very much like the rest of the country, that kind of unique pattern in the South where the Democrats were more rural, that really falls apart, and it falls apart in a number of other places, as well. And so, we see this convergence and it’s [inaudible 00:19:15] related to the larger nationalization of politics in the US, where it all sort of snaps together into this single dimension.

Matt Grossmann: Wilkinson points out that the rural/urban divide took a lot longer to get so strong in the US.

Will Wilkinson: Rodden’s book emphasizes the relationship between left parties and cities is very longstanding and deep and robust. And one of the peculiar things about the United States is that a rural-urban divide in partisan alignment hasn’t… took way longer than in a lot of places, to really show up for all sorts of contingent historical reasons. And so, one thing I try to emphasize, when people ask like, “Urbanization is happening everywhere. This is a pretty universal phenomenon. Are you going to see the same thing in every country, and what…” Well, you’re not, because every country has a political history.

A lot of my story, which we haven’t really touched on here, really has to do with ethnicity, has to do with the concentration of the nonwhite population in cities, and then the separation of the white population between urban areas and non-urban areas, that that population is specifically become polarized.

And so, the cities were already relatively liberal parties. And in a lot of the changes that I’m trying to describe are very fundamentally caused by the reasons that people are gravitating towards cities. The transformations in the economy, the increasing output with efficiency, of urban relative to rural economic production. These longterm trends that are causing some of these divisions are partly a function of the nature of cities and the attraction of cities. So the things that are causing the coalitions, are also causing a lot of the background economic and cultural changes that parties are aligning around.

Matt Grossmann: But today Rodden says that density matters even for very small towns.

Jonathan Rodden: I have noticed that the relationship is there really in almost every county in a place like Pennsylvania, and some of these industrialized states. One of the things I have seen is that the relationship is a lot stronger. It’s a lot sharper in places that did have a lot of industrialization. So I do think that early industrialization is part of the story, but the fact that it’s there, even in these very small places, and it’s certainly… it’s there even in places that are overwhelmingly white, so we can’t really rely on a story just about race. And certainly a lot of these places, it’s not that the urban core is somehow full of hipsters and knowledge economy professionals. That is not the way I would characterize places like Lima, Ohio, or even Muncie, Indiana, which has a university. But all of these kinds of towns, still, the urban core, if you want to call it that, is still quite democratic and you see that same pattern. Whereas, you move out from the city center and into the suburbs and out into the periphery, the Republican vote share kind of increases in this according to the kind of an exponential function.

Matt Grossmann: Rodden finds a few countries like Sweden that don’t match the global pattern, but that shows that electoral systems matter.

Jonathan Rodden: You know, I would go to Scandinavia and present this and then people would say, “Well, what about the left wing forestry workers and landless laborers in Norway? And what about the urban conservatives in Sweden?” This is something I’ve gotten very interested in and I’m continuing to work on, but I think there’s something really… This is where I think it’s important to focus on the electoral institutions and party systems. The thing that’s different about Sweden is that it has a multiparty system.

And if we just talk about the world in terms of left and right, we make this observation that much of the higher income downtown core of Stockholm votes for parties that we might call right parties. And the reason we might call them that is because they typically coalesce with other right-wing parties when they form governments in Sweden. But if you look at the ideology of the parties that people in Stockholm voted for, these are liberal parties, these are parties that have left-of-center… what in the US spectrum certainly would be considered left-of-center preferences on social issues, but they have moderate maybe center-right preferences on economic issues. And you might very well imagine if we had a multi-party system, that a liberal party like that might do pretty well in some of the high income knowledge economy city centers in the US, as well.

We see something similar with Greens and Free Democrats and others that have taken over the political ideology spectrum in central cities in other European cities. A good example of how this works is, if we look at some recent elections in Austria, in multiparty competition in the proportional representation elections for the legislature, we see right-wing parties or parties that are kind of center-right doing pretty well in Vienna and some of the higher income urban neighborhoods. But in a recent presidential election, it went to a second round where there was a Green candidate versus a far-right candidate, that made it to the second round. And we look at that electoral map, it looks just like a US electoral map, where Vienna, Linz and Graz vote overwhelmingly for the party of the left, and the rest of the country, all of the rural parts of Austria, have vote for the party of the right.

So just by changing those electoral institutions and the menu of choices, you create something in Austria that looks just like the US, which under normal circumstances, in a multiparty competition, you don’t see in Austria. So I think that’s an important part of the difference between the US and some of these proportional systems.

But that said, this urban-rural divide, it is growing around the world in lots of different kinds of countries for some similar reasons to the United States. As the knowledge economy sector becomes more dominant and urban areas start to pull ahead and non-metro areas are left behind, the same kind of division that we’ve seen in the United States is developing in lots of other places. And I think it’s all a bit more pronounced in the countries that have majoritarian aspects to their political system, like the United States.

I think Hungary is a good example. You look at the Hungarian electoral maps and you’ll see a real concentration of support for the left in Budapest and a couple of other cities, and the support for the regime in basically the rest of the country. And an electoral system that really helps the right hold on to power. So in that respect, I don’t see the US as really completely distinctive. I think the thing that’s most distinctive about it is this combination of presidentialism and a two party system that really puts all of this maybe on steroids a bit in the US.

Matt Grossmann: Wilkinson agrees that the rural-urban divide is very widespread, with some contingency and exceptions that prove the rule.

Will Wilkinson: It’s hard in that you’re looking at… in a social sense, to find patterns as robust as the kind that Rodden is seeing. In my paper, I talk about my home town, Marshalltown, Iowa, which is a town of 27,000 people, and you see this relationship between intensity and party vote share pretty clearly, and then that’s interesting. And it defies a lot of people’s expectations about what small Midwestern towns are. Iowa went for Trump and Marshall County, went for Trump, but Marshalltown went for Clinton, partly because the densest part of town went strongly for Clinton. But some of those patterns are a contingent reflection of the history of the development of these cities. And so as Rodden points out, older cities are denser. Before you have cars, people have to be closer together in order to get the benefits of proximity. You can’t just drive to work.

So that relationship is definitely attenuated just by the date at which the city is founded, and whether or not the urban plan is based around automobiles, or not. So when you’re looking at some of these Arizona examples, or just the sprawl of the Southwest, the relationship is not as stark, and it reflects these contingencies about technology and what makes sense given certain modes of transport and certain eras of economic production.

You’re not going see incredibly strict [inaudible 00:28:17] relationship, but you still see pretty much the same relationship, and you just continue… You can see the extent to which it depends on whether the downtown was a pre-car place, or not. You know, in the United States, ethnic minorities are very, very heavily clustered in the denser, older parts of cities, because at least historically, not so much the case now, that older, depreciating housing stock was more affordable and made it possible for less advantaged communities to cluster. But you don’t see that same kind of pattern everywhere. In some European cities, the minorities tend to be arrayed around the suburbs and the cores are dominated by the ethnic majority.

Matt Grossmann: Rodden reviews several explanations for growing geographic polarization, including sorting, context effects, and changing economic interests.

Jonathan Rodden: You’re now getting to the frontier of where the research currently is, and there’s a lot that we still don’t know. I’d say there’s maybe even more… there are more possibilities than just a sorting and context. I would say these stories about sorting where people with different preferences or personality types, and then those things are connected to political behavior, and then they move for these various reasons, and that’s why we see urban-rural polarization.

I think a distinct argument is one that’s more about self-interest. It might just be that that living in a dense environment just creates certain kinds of demands for public goods that aren’t there. It just kind of changes the way public goods are provided, and the demands for regulation.

So for instance, the fire protection is something that in a rural area you can perhaps achieve through volunteer mechanisms. Whereas, in cities you end up with a government provision and public sector unions and all that. That’s just a distinct argument, I think, about… the more economic argument about the costs of providing public services and the nature of demands for public services.

And then, when you talk about contextual effects, I see that as even a separate argument about social networks and you kind of place a person in a protect particular environment and then then that shapes their beliefs through their social interactions.

I’ve offered even another argument, which is that the parties are kind of changing what they emphasize and what their issues… or what their platforms are over time, and people sort into these parties as the parties pull apart on different issues. So these things are all possibilities, and how do we disentangle these, how do we figure out… I think all these things are happening, and that’s the problem. It’s just not one or the other.

Matt Grossmann: Wilkinson thinks sorting works in tandem with social context to make areas more divergent.

Will Wilkinson: If just moving to a city makes you more liberal, or staying in a small town just makes you more conservative, and if you just swapped where those people live, their politics would accordingly change. I think that these things definitely work in tandem. There’s no doubt that people are conformist, that we very strongly gravitate toward fitting in with whatever people we’re around. Like, I have no issue with the idea that our views and our attitudes and our political opinions are heavily shaped by our social context and the places that we live. But I do think that we’re not picking up on the degree of the sorting for a number of reasons. One, is just that, as I mentioned, a lot of this literature doesn’t even think about the longterm transformative effects of urbanization, and the fact that even very small differences in the propensity of a certain kind of person to move, certain kind of person to stay will just tend to separate populations.

And because of the constancy of urbanization as a kind of background to everything that’s happening, I think it’s kind of hard to pick up in the way we tend to look at these issues. And then a lot of the studies on, specifically, on ideological migration, or whether people are sorting on partisanship, or [inaudible 00:32:41] partisanship, are really just kind of methodologically difficult because the units don’t stay fixed in the right way, geographically.

Matt Grossmann: He says basic personality traits may be driving sorting and polarization.

Will Wilkinson: I talk a lot about the Big Five personality attributes, and one of the reasons I focused on that… There’s a huge literature and political psychology that there’s all sorts of stuff, authoritarianism scales, Jonathan Haidt’s moral foundations theory, John Jost’s, just-world theory. But a lot of that stuff feels pretty ideologically loaded to me and I worry about it methodologically.

The Big Five personality theory is fairly ideologically neutral. It’s not really super theory-laden. It kind of falls out of just the factor analysis that you do to try to figure out what which traits are relatively uncorrelated than other traits. And these things very strongly predict all sorts of things. They’re just not invented to explain anything about politics, and they help you understand why some people do better in school, while some people make more income, why some people spend more time socializing, why some people are more inclined to have anxiety disorders. There’s a huge literature. This stuff predicts a lot of stuff. It’s robust, it’s replicable and the traits are fairly deep and stable.

The trait that I concentrate most on, openness to experience, is the one that has the strongest relationship to social liberalism, social conservatism. None of these traits correlate in any very interesting or significant way to positions on economic issues, but they tend to predict in a pretty robust way, people’s views on social issues.

And the openness to experience, in particular, is the one that has the deepest relationship to ideology. But also, which plays an important role in my story, to attitudes about ingroups and outgroups, warmness towards other ethnicities, tolerance of diversity, residential preferences that have to do with preferences for ethnic homogeneity or ethnic diversity. And that trait, in particular, is the most heritable of any of them, which is interesting. These things do change. You go to college, your openness goes up. If you move to a big city, your openness goes up. So you have these context effects on these traits. They’re not completely fixed, but as close to as fixed as something like IQ, which has very strong effects on lots of things.

And one of the things that really moved me looking into doing this research was this work in geographic psychology or psychological geography, whichever one comes first, that shows a pretty clear relationship between the propensity to migrate at all, and the same trait that predicts social liberalism or conservatism, as well as a… And this is something that I’m looking forward to seeing more work done on because it hasn’t been-

Will Wilkinson: That I’m looking forward to seeing more work done on because it hasn’t been done at this level in the United States. Some of Peter Renfrow’s work at Cambridge, you see the distribution of personality types, or the typical personality type, in a city like London lines up with density in an incredible way that. It looks like, the distribution of openness in London looks a lot like just the map before population density, which looks a lot like the map for Brexit or remain and I think there’s something relatively deep about that set of connections.

Matt Grossmann: [inaudible 00:36:47] open to Wilkinson’s focus on psychology as part of the broader puzzle.

Jonathan Rodden: I’m fascinated by this new literature at the intersection of psychology and geography that Will is drawing upon, and I think there’s probably something to it. When you think about it, there’s probably something different about a person who picks up and moves to a new city to take a job and someone who decides to stay where they are. There’s probably a something different about those who stayed in Europe and those who moved to the new world in the early 20th century and in the 19th century, or those who stayed in the south and those who moved to Chicago during the great migration or the people that moved west in the westward migration. They’re probably some psychological characteristics about moving that are distinctive from the people who stay behind or even just at the most basic level, those who stayed on the farm in the industrial era and those who moved to the city and became wage laborers. There’s probably something about attitudes towards risk, openness to new experiences, tolerance for diversity. I think that all seems right.

But of course there are lot of other differences, more kind of in terms of economic conditions and family ties and religions and a host of other factors. So I’m certainly quite open to those arguments and I think there’s probably something to them. But again, I think it can’t be all geographic sorting just in because not enough people move and-

Matt Grossmann: However the geographic divides developed, the parties are responding to their changing coalition. Wilkinson sees a common pattern where social issues increase in importance with socioeconomic changes from urbanization.

Will Wilkinson: When countries have a relatively high rate of economic growth, you tend to get general cultural liberalization. People become less concerned with issues of material security and more concerned with issues of the individual’s self expression of identity, and my general view is that that kind of dynamic just does push politics over time toward more expressive cultural and social issues that begins to dominate people’s attention more and more. It becomes a clearer basis for elites and parties to coordinate coalitions around.

Matt Grossmann: Rodin says the constraints of the current Democratic Party are a product of their geography, and common for the left globally.

Jonathan Rodden: One of the things I discovered in the book is that this divide between AOC and the squad or the democratic socialist on the one hand and the kind of democratic leadership and these candidates who are struggling to try to win in the pivotal suburban districts. That divide is just central to majoritarian democracies ever since the rise of left parties, ever since the industrial revolution. This is nothing new at all. It’s maybe a bit new for us because our parties have been so diffuse and meaningless in the past. And so we now actually have national party labels that matter and that are hard for people to run away from. Labor parties have always had this problem.

So the true believers in the urban core who want to pull the party platform out to kind of a ownership of the means of production and things like this. The Labor Party has always had that urban kind of impetus coming from its core districts in Britain. I think something is true in Australia as well where there’s this intense divide between Labor left and the centrists, which is really a kind of an urban suburban kind of divide as well. So there is this problem if you adopt the platform preferred by the left tail of the distribution, or even if you don’t, if people believe that you do, if the perception is that that’s your platform, then you’re going to have trouble winning in the pivotal districts around the middle there you need to win in order to form a majority. And this is an insight that Shavorski and Sprague made in a classic book about electoral socialism called Paper Stones. They didn’t really focus on the geography of it but the geography really just heightens this basic problem where the the left is really kind of torn between being true to its urban base and trying to win elections.

Matt Grossmann: Both Rodin and Wilkinson see US electoral institutions as central to the story, limiting change without reforming the rules. Wilkenson sees polarization and democratic disadvantage continuing.

Will Wilkinson: Where I see the trend towards even greater polarization culturally and politically along these lines, a point that I emphasize toward the end of the paper, which is to a large extent the theme of Rodin’s book is that the political effects of these things is going to depend a lot on the relationship that that systems of representation have to geography, have to the way it aggregates populations, and our system is just uniquely bad in a lot of ways in terms of its tendency to exacerbate these problems. We have a winner take all, first past the post system that pretty much ensures a stable two party equilibrium. So the tendency is always going to be divide the population roughly in half, the United States system. Now the oldest Liberal Democratic electoral system in the world. It’s designed in the seventies eighties and nineties in a completely agricultural economy, in a political context that was obsessed with the relative power of slave and free states, and our electoral institutions reflect that. And part of that reflection is in a very strong bias in our system toward the political power of less populous places.

Will Wilkinson: There’s a big penalty for density built into our system, and as Rodin emphasizes that there’s also just a natural function of density when populations cluster really tightly together and you have a spatial system of representation and one party at the dense party and one party is the not dense party, the dense party’s voters are going to be very inefficiently distributed and they’re going to have a problem. Well that problem is exacerbated by a constitutional structure that effectively gives a bonus to places that have smaller populations and a lower population density. So I don’t think we’d be in the problem that we’re in without all of those factors lining up in the right way. Which at one level craze or reason to be optimistic. If you can just change that system even a little bit it can force realignment. The Republican Party has really strongly doubled down on being the party of older, less educated, lower density white people. That is not a majority of the population.

Matt Grossmann: But Rodin says partisan segregation may not continue increasing given and has a little room to grow. There are even some mechanisms for a potential reversal.

Jonathan Rodden: When you look at the just scatterplots of population density and voting, we’re almost getting to the point where it’s hard to imagine that the scatterplot could be any cleaner than it already is and the correlation is getting pretty close to one. So I’m not sure we can go much higher, but there are some interesting countervailing forces that push in the direction of greater partisan mixing rather than segregation.

One of them is the suburbanization of minorities. That there’s this big movement of minorities from the central city to suburbs that that looks a lot like the movement of whites from cities to suburbs a generation ago.

Even in cities like my hometown of St Louis is actually leading to a more heterogeneous, suburban communities. And at the same time, people who are moving for opportunity, if we look at where they’re moving to there, they’re moving to places like Houston and Orlando and Phoenix and Austin. These are places that are politically competitive and heterogeneous places and they’re becoming even more competitive and heterogeneous. So I don’t envision us as constantly moving to increasingly segregated environments, in fact; as we just become more and more suburbanized and continue to spread out, you can see the trend really moving in in the other direction. And in terms of the party’s coalitions, an argument that Will makes that I think also just that sounds sounds right to me is that it could could very well be the case that the kind of takeover of the Republican Party by Trump and Trump’s allies is, in 2018, created the appearance that that was not a very good development for the Republicans in their desire to hold on to the suburbs that have kind of really been, when we talk about those pivotal districts in the middle of the distribution that have given this advantage to them over the last couple of decades, if they start losing those places, they’re in trouble.

And so one could imagine the Republican party trying to change its strategy if it loses this kind of goose that has been laying the golden egg in these suburban areas.

Matt Grossmann: Both emphasized the need for more research. Roden says, “We need to understand the mechanisms of polarization and the implications for the reversal of the party’s economic basis.”

Jonathan Rodden: Disentangling treatment versus selection effects. I think there’s a lot of good research to do there and a lot of room for creativity and understanding those things. Trying to get a better handle on mapping preferences at have very low level of geography is something that we can improve on.

I think many people have pointed out that the support for Democrats is increasingly concentrated in counties that are relatively wealthy. And if we look at data from the IRS on taxes, we see that something like two thirds of of IRS income comes from places that the very small number of counties that voted for Hillary Clinton in 2016.

So essentially we’re creating a system in which the party of the left is the party that is paying into the system of fiscal transfers. And this is true within states as well. So the party whose position is essentially anti-government is the party whose voters, at least when it comes to spatial fiscal flows, tend to be relatively dependent on government. And that creates an interesting dynamic and an interesting set of dilemmas for the Republican Party in states like Kansas and Kentucky, that might be interesting to keep an eye on in the years ahead. How do they manage this desire to simultaneously to cut government but also to avoid angering places that are important to their base where government employment is actually a pretty large share of the employment base.

Matt Grossmann: And Wilkinson is focused on the understudied implications of urbanization and migration for politics.

Will Wilkinson: Urbanization is slow, but it is a really transformative social force. I don’t really remake the nature of economic and social life all over the world. Just the movement of people from the countryside to cities everywhere over the past two centuries really. But mostly in the last century. It’s just been this massive, just titanic force that’s just changed everything, everywhere. And it seems like it would figure in more centrally to a lot of accounts of the things that are puzzling us about our politics. So I’ve been long interested in the reasons that people migrate, as well as an ideas and social theory about how even little differences in individual preferences and abilities can lead to the clustering of like with like, and a corresponding segregation of populations on those differences.

So knowing that even small differences in case and preferences can separate populations on those differences, it seemed natural to surmise that tens of millions of individual moves from lower to higher densities over multiple generations, could have disproportionately filtered certain types of people out of the country and led to patterns of settlement that produce the striking relationship between population density and partisanship that has emerged.

Matt Grossmann: There’s a lot more to learn. The Science of Politics is available biweekly from the Niskanen Center. I’m your host, Matt Grossmann. Thanks to Jonathan Roden, Will Wilkinson for joining me. Please check out why cities lose and the density divide and then listen in next time.

The photo used in this post is in the public domain.