Public transit infrastructure and urban social connectedness
Cities are amazing places – the presence of many people with different backgrounds, viewpoints, occupations, and skills in close proximity allows for the speedy exchange of ideas and information, and can thus accelerate innovation and social progress. Urban economists have long pointed out that this mixing of people and ideas is at the heart of the ‘agglomeration externalities’ that drive the high productivity of cities (e.g. Jacobs 1969, Bairoch 1991, Glaeser 2011). However, for these agglomeration forces to be active and strong, it is necessary that different people living in different parts of the same city actually interact with one another.
Public transit infrastructure is one of the primary ways through which urban planners attempt to facilitate such social interactions. The theory goes that by making it faster (and cheaper) to travel between different parts of the city, individuals living in those parts are more likely to meet a broader range of other city inhabitants. However, the lack of large-scale data on the geographic scope of social connectedness within cities has made it difficult to study the extent to which public transit actually affects social connectedness.
In a recent paper, we explore the link between public transit and social connectedness in New York City (Bailey et al. 2019). We measure social networks using aggregated and anonymised data from Facebook. By the end of 2017, Facebook had 239 million monthly active users in the US and Canada, and about 2.1 billion users globally.
We observe an anonymised snapshot of all Facebook users with location history enabled as of March 2018. For these users, we observe their locations at the zip code level as well as their connections to other individuals on Facebook. We then construct a social connectedness index between all New York City zip codes as the relative probability that two individuals in these zip codes are friends with each other (see Bailey et al. 2018 for a related exercise at the county level).
We discuss a number of case studies that show that the social networks of urban zip codes are distributed along transit routes that connect these zip codes to other parts of the city. Figure 1 maps the social network of residents of Little Neck, Queens, a neighbourhood on the eastern edge of New York City with easy access to the Long Island Railroad (LIRR) into midtown Manhattan (zip code 11363). We find that Little Neck has strong social connectedness to residential areas in midtown Manhattan near the LIRR terminus.
Figure 1 Social network of Little Neck (zip code 11363)
Figure 2 shows the social network of zip code 11364, covering the neighbourhood of Oakland Gardens in Queens. While adjacent to Little Neck, which has two LIRR stops, Oakland Gardens does not itself have a LIRR stop. Its social network differs from that of Little Neck in that none of the top connected zip codes extend into Manhattan.
Figure 2 Social network of Oakland Gardens (zip code 11364)
These maps therefore provide the first suggestive evidence that New York City’s public transit system plays an important role in enabling the formation and maintenance of social ties across geographic distances. Indeed, it appears as if transit links can effectively ‘shrink’ the geographic distances between locations within the city.
To explore the relationship between social connectedness and transportation infrastructure more formally, we calculate the travel times on public transit between each pair of New York City zip codes. We find that social connectedness declines strongly with the travel time between locations. Within New York City, the elasticity of social connectedness to travel time is -1.42, which is about 60% larger in magnitude than the elasticity of social connectedness to geographic distance, which is -0.87.
This finding suggests that public transit can help the maintenance and formation of social links across individuals living in geographically distant parts of the same city. As a result, extensive public transportation infrastructure can increase agglomeration benefits as well as reduce the extent to which residential segregation leads to social segregation.
In addition to the role played by geographic distance and public transit travel time in forming and maintaining social ties between geographies, we find that zip codes that are more similar along demographic measures such as race, education, and income are more likely to be socially connected. This is consistent with previous studies that have documented that social ties are generally more common between similar individuals and regions, a feature that is often referred to as ‘homophily’.
We show that short public transit travel times are more important for connecting zip codes with different incomes than they are for connecting zip codes with similar incomes. This finding highlights that public transit investments do not just facilitate social connections between far-away zip codes in general, but do so particularly across zip codes with different demographics.
We also provide a descriptive analysis of the geographic concentration of social networks. We find substantial heterogeneity in social network concentration across New York City zip codes. For residents of the median zip code, 29.0% of US-based friends live within five miles of each other, but this number ranges from 19.5% to 39.6% between the 5th and the 95th percentiles of the zip code distribution. Similarly, for the median New York City zip code, 22.0% of US-based friends live among the nearest one million people, while the 5th–95th percentile range is 13.1% to 32.7%.
Consistent with the important role played by public transit infrastructure, this geographic concentration of social networks is highly correlated with access to public transportation infrastructure (measured, for example, by the share of a zip code’s population that lives within a quarter-mile of a rail transit station). These results hold even after conditioning on zip-code demographic and income measures.
The ease of transit also explains more of the across–zip code variation in the concentration of social networks than zip-code demographics do. Quantitatively, a 15-minute increase in the average travel time to all zip codes is associated with a 3.7-percentage-point increase in the share of friends living within the nearest 500,000 people.
We also find that the geographic concentration of social networks correlates with socioeconomic factors such as income and education levels: the share of friends living within certain distances decreases with zip-code income and increases with the fraction of population without a high school degree.
Although our data do not allow us to make statements about the causal connection between social connectedness and socioeconomic outcomes, our findings are consistent with the urban economics literature that points to social interactions as a primary channel for agglomeration externalities that can improve the economic outcomes for residents.
Bairoch, P (1991), Cities and economic development: From the dawn of history to the present, Chicago: The University of Chicago Press.
Bailey, M, R Cao, T Kuchler, J Stroebel and A Wong (2018), “Social connectedness: Measurement, determinants, and effects”, Journal of Economic Perspectives 32(3): 259–80.
Bailey, M, P Farrell, T Kuchler and J Stroebel (2019), “Social connectedness in urban areas”, NBER Working Paper w26029.
Glaeser, E (2011), Triumph of the city, London: Pan Macmillan.
Jacobs, J (1969), The Economy of cities, New York: Random House.
Please Donate To Bitcoin Address: [[address]]