The most important factor polarizing American politics
Flashy images like this photo and the events of Black Lives Matter may give the impression to someone less familiar with the US that politics is becoming polarized along cultural lines such as race, religion. Or perhaps that the country still hasn’t got over the Civil War, still entrenched in cultural divide between the Southern States and the rest of the US. Polling data seems to confirm this, with race, religion and education all significant splits in voter patterns in the 2016 election, and the usual pattern of the “Republican stronghold” in the South. Statistical analysis of election outcomes by county also appears to confirm this theory. The “liberal” score (Green Party + Democrats - Libertarian -Republicans) for the 2016 election outcome by county can be predicted with a R2 of 0.83 using 6 variables (in order of significance) : population density, race, education, income, religion, and whether the state belongs to the “Southern states”. Age and gender are also significant factors but are more easily separable on individual surveys than county-level data.
There is a little more to the story however. The point of this article is not to downplay the significance of the culture and race in US politics, but to shift attention to another, slow-moving factor that may provide clues for how to resolve the gridlock.
Just how different is the South from the rest of the US?
Here “the South” is defined as 14 states that fought in the Confederacy and continue to vote Republican in recent elections — Alabama, Arkansas, Florida, Georgia, Kentucky, Louisiana, North Carolina, Mississippi, Oklahoma, South Carolina, Tennessee, Texas, Virginia, West Virginia.
The stereotypical culture of the South is associated with being white, rural, and evangelical Christian. So race and population density may be important places to start to compare how the South appears compared to other parts of the US.
A couple of observations show up on a quick scan of four plots of race (% white), median income, education (yrs high school + college) and age vs the county population density. The main differences from the South and everywhere else appear in the cities. Examples of Southern metro areas are Dallas, Houston, Texas ; Atlanta GA ; Tampa, Miami, Florida. Cities in the South on average are whiter, higher educated and median incomes are higher.
A second difference is that the rural South is more racially diverse than rural areas elsewhere. Rural areas by comparison to the cities are more demographically homogeneous inside and outside the South — poor, white, low education and old.
The broad story however is the pattern of population density and socioeconomic outcomes that appears inside and outside the South. The higher density of cities lowers the per-unit material and energy cost of everything — concrete, steel in apartment buildings, transport infrastructure for delivery of water, sewage, electrical lines, roads, etc.. Aside from material efficiencies, the effect of density is even more pronounced for social services. Humans are social beings and much of socioeconomic outcomes are tied to being connected socially to a community — social capital. When it comes to finding opportunities to form new connections in social networks, bigger is better. This is known as the “network effect”. A whole host of services from classrooms, medical services, legal services, broadband, delivery or even dating services all benefit from bringing people in closer contact through population density due to lower per unit cost and network effects.
Research on urban studies is uncovering the many ways that urbanization affects our experience of life and economic activity. Geoffrey West’s book “Scale” presents surprisingly universal scaling laws common to cities around the world. The theory in the book is that the scaling patterns of cities follows a similar logic of physics as metabolism in animals by economizing on resource efficiency for delivery of basic inputs of materials, food, energy. The surplus value is converted into a wide range of socioeconomic activity — bars, restaurants, theatre, patents, and also crime. The pace and frequency of everything is faster as the density of the city increases, even the pace at which people are walking.
As we learn more about ourselves in the field of Behavioral Science, the debate on “nature” vs “nurture” is becoming less of a debate. Our environment matters a lot. Where Geoffrey West explained the emergent patterns of cities, in the book “Behave” Robert Sapolsky goes deep into the psychological and behavioral effects of city life on how we see the world, treat other people and form our political attitudes.
Takeaways from Sapolsky and West —
- Cities have strong comparative advantage for human development both for material needs and the social aspects of well-being. They also have some unpleasant drawbacks such as pollution (in the city), crime and environmental costs outside the city.
- The experience of living in an urban environment influences not only our socioeconomic outcomes, but also how we think and behave, for better or for worse.
It’s not surprising to see a partisan divide based on population density. Given what is known about cities, it would be a surprise to NOT see a divide. What is less clear is to what extent the difference in politics is rooted in conflicting needs, diverging perceptions or values or a loss of social cohesion and ability to form shared experiences between the two zones.
Another observation that is obscured by media headlines that disproportionately cover events in urban areas is just how rural the US is, even though it’s officially reported to be 80% urbanized.
US cities are known for sprawl and the pattern is mostly uniform in and out of the South with the exception of the geographically constrained areas of Miami and to a lesser extent Southern California.
Using 500 ppl/km2 as a rough cut-off (not very dense for anyone familiar with urban geography) between rural and urban, the rural population is 2x larger than the urban. Using this cut-off reveals two interesting observations. The first is that the spread of rural-urban is a stronger determinant than the Southern state designation in partisan lean. The second observation is that for 2016, the country was a near even partisan split from the cancelling out of the effect of large liberal lean in urban areas with the population advantage of rural areas. The calculation in this analysis is weighting the election results by the total county population, not the actual turnout so this could imply that the Democrats had a slight turnout edge over the Republicans in 2016. Hillary Clinton actually won the popular vote by 2% margin. The exit polls analysis identified the demographic group with the highest turnout is white, college educated (~80%) which tend to lean liberal although it’s not clear how they leaned in the 2016 election. White non college educated had similar turnout rates as minorities (50–60%).
The significance of this rural-urban divide for US politics is not new but goes back to the design of the US constitution — and in particular the design of the Senate. Of the Founding Fathers, the most progressive that were pushing for Independence were based in the manufacturing centers of the Northeast — Massachusetts whereas Loyalists less anxious for independence were in the South. To bring the Southern states on-board, a compromise was reached that would provide the smaller (non-slave) populated Southern States with a more even political standing with the densely populated Northern States. Congress was setup into two chambers — the House and the Senate. The Senate is considered as slightly more prestigious, but has a unique undemocratic design — each state gets two Senators no matter the population size. In practice this has not given any one particular region an advantage since the small number of states in the Northeast have balanced out with the small rural states of the Midwest. It does however provide the Senate with a systematic rural bias, for a country that is already 2/3 rural.
The Senate has always functioned as a brake on more radical push for changes at the national level, but it’s role as a brake has intensified in recent years. Complicating the rural-bias is the role of the controversial filabuster (which some say was created on a fluke rather than careful design intention). The filabuster was originally intended as a procedural brake to prevent rushing to pass legislation without careful debate, but has evolved over time to an effective 60% majority threshold on any controversial legislation. While the actual party control between Republicans and Democrats has been even in recent years, both parties and the actual policies are infused with a rural bias, which also tends to be conservative and resistant to change in general. Economic and political trends from globalization and urbanization emerging in the 20th and 21st century create new and unique political and economic demands (Automation, Wagner’s law, Baumol’s cost disease) that are unique to cities. While these trends present new opportunities and challenges that are existential for cities and they have spillover effects to rural areas, for the rural areas the situation appears less obviously existential.
From a political economy point of view the perception of an abundant and accessible natural resource may strengthens the resolve for the safe default option of “the way things used to work”. Proof of concept for this argument is the numerous examples of indigenous communities that continue to persist with little changes to their political economies autonomously in parallel to what is happening everywhere else. The structural rural bias, and the institution of the Senate however are ill-equipped to be able to respond to these new urban political economy demands — hence the tension. The spillover effects from urban to rural is real, despite the perceived security.
Agriculture is only one part of rural US economy, the other part is the shadow leftover from the loss of manufacturing. Some may be feeling the growing economic pressures sensing that something has to change, and willing to gamble with extremes to “shake things up” feeling that chaos is better than stalemate.
If you have chaos, then you will get change
- Rural swing voter from the 2016 election who voted for Obama in 2012 and Trump in 2016
While rural-urban is a useful dichotomy to understand the pattern, the suburban zone outside cities is large and shares characteristics of both. In the traditional Thunen model, manufacturing sits in the zone between rural and urban. The US however filled this hole with something else — The suburb.
Since their inception in the 1960’s elections have been decisive battlegrounds in US elections as political buffer zones between the rural and urban spheres. The suburban zone may have different definitions depending on the source in the range of 200–1,000 ppl/km2 which represents the bulk of the US population.
Mapping Urbanization in the United States for 2020, DOI: 10.1109/IGARSS.2018.8517770
The transition is not uniform, there is a distinct signature of “small town” with densities 5–100 and median 50 ppl/km2 and “surburb” with range 100–1,400 at median of 500 ppl/km2.
The south is unique from other areas for a higher mix of “small town” vs “suburb” and for not having much urban residential core. Unlike the downtown cores in New Haven or New York city with a mix of residential and commercial spaces, downtown Fayetteville, NC or Richmond, VA are void of residential activity in the evening and on weekends.
The dramatic contrast can be seen when compared with a highly urbanized country — Singapore which is composed entirely of urban planning zones with densities between 10,000 and 30,000 ppl/km2.
This skew for the Southern states of a “small town” bias results in a state-wide conservative bias because of the nonlinear effect of the partisan relationship with density. Liberal-ness is about the same between moderate density suburbs and urban core, but the same is not true for the difference between a low density suburbs and a very small town. The transition of partisan lean from conservative to liberal shows up with a clear pattern that matches with the rural-urban zone in the range of 100–1,000 ppl/km2 with an inflection point around 500 ppl/km2 — a typical suburb.
The overall density-partisan relationship in 2016 appears mostly the same inside and outside of the South.
The pattern can be seen on image maps of election outcomes by county. The same pattern appears all over the country from the Southern cities of Houston, to the state of Florida to California and the Northeast.
The electoral map in 2016 has become more polarized along rural-suburban lines than in 2008. The conditions in 2008 were on the heels of the global financial crisis and the anxiety of economic uncertainty was more uniform between the urban and rural areas. The economic environment of 2016 in contrast shows a shift to more polarization between small town and the suburbs. The shift from 2008 to 2016 was that the difference between small towns outside the South started to look more like small towns in the South, so that the country became more politically uniform by region, but more polarized by the small town-suburb divide. Low density suburbs outside the south flipped in 2016 but high density suburbs were unchanged. Inside the South the Tea party concentration of base support can be seen by a deepening shift to conservative in small towns, but unobserved elsewhere.
The gains in the recovery from 2008 to 2016 were non-uniform with highly educated white collar workers receiving a disproportionate gain of the recovery compared to unskilled workers. Compared to 2008, the share of the pie may have become a more salient question than whether the overall economy is growing or not.
While the Republican party electorate still retained the identity of rural, upper-middle income, evangelical, from 2008 to 2016 Race (white) and Education (non college educated) became a much more prominent line of contrast identifying the alliance with Trump.
Education, which is highly correlated with rural and suburbs showed up as a stronger factor in the 2016 vote compared to 2008. The sharp divide along education and race lines in 2016 was so dramatic that it was identified as one of the contributing factors to systemic polling errors in 2016. Unfortunately 2016 also saw a greater role of race in partisanship than in 2008.
The education gap however is a direct outcome of the economics of rural America. The network effect of higher populated urban areas is an advantage for supplying cost effective high quality education. The brain drain from rural areas over time draws out talent of all kinds, including teachers and the property tax revenue that schools depend on. It is in this void that religious institutions step in to provide community support both financial and emotional. For many rural counties the church is the primary center of social activity where residents can meet one another to find friends, prospective spouses, business connections and basic social interaction that is essential for psychological well-being.
Conventional images of the gains of post WWII economic growth can be seen in the incomes and standard of living increases in urban and suburban areas, but despite claims of 80% urbanization, the population distribution of the US is still mostly rural or light density suburban, and this matters in elections. Low density swing state suburbs have a disproportionate influence on election outcomes as they are large in numbers and share characteristics between the small town and the urban. While Southern cities do have different income and racial composition compared to the rest of the US, when it comes down to how they vote, population density matters a lot.