Public transit access and income segregation

Low-income households in US cities tend to be segregated in low-opportunity neighbourhoods with worse access to good schools and desirable urban amenities. Public transportation is often expected to bridge inequalities, but sparsely distributed public transit networks may end up concentrating poverty further if high- and low-income households sort differently between driving-accessible and transit-accessible residential neighbourhoods. This column presents new insights on the relationship between transit expansions and income segregation within cities, and when segregation arises as an outcome of planners trying to maximise public transit ridership.

Residential segregation by income is increasingly common within cities (Gallagher 2021). Segregation can limit access to good schools, urban amenities, and opportunities for disadvantaged households, and has been associated with low intergenerational mobility (Saez et al. 2014, Fogli et al. 2024) and persistent racial inequality (Sethi and Somanathan 2004). Public transportation is frequently touted as a tool in addressing these inequalities by improving urban mobility and social connectedness (Ströbel and Kuchler 2019) disproportionately for those with poor private transport alternatives.

At the same time, slow public transit networks in US cities have been held responsible for spatially concentrating poverty (Glaeser et al. 2008). The most common modes of public transportation in US cities tend to be fixed-route bus services that can be considered locational amenities of the neighbourhoods they connect. They are typically much slower than private travel alternatives and thus more attractive to travellers with relatively low opportunity costs of time. So, as a neighbourhood amenity, public transit access is more attractive to low-income households and often deemed a poverty magnet. Indeed, there is a large literature showing higher poverty rates or lower household incomes closer to mass transit stops (e.g. Pathak et al. 2017, Barton and Gibbons 2017) that complement the conclusion by Glaeser et al. (2008) that “the urbanization of poverty comes mainly from better access to public transportation in central cities”.

Yet, ironically, in a recent paper (Akbar 2024), I find that it’s also when transit authorities expand accessibility on these ‘slow’ public transit networks that they reduce income segregation across neighbourhoods. On the other hand, I find that if the transit network were instead ‘fast’ relative to driving (such as comprised of dense networks of subways and trains as opposed to buses), transit authorities would expand accessibility in ways that exacerbate segregation.

To reach this conclusion, I document six empirical regularities that characterise how households sort across travel modes and residential neighbourhoods within US metropolitan areas. Then I use these regularities to guide a stylised transit planner who chooses where to improve transit accessibility to maximise ridership and to characterise the implications of their choices for income segregation.

• Fact 1: The incomes of transit riders differ notably by type of public transit (as shown in Figure 1). Buses and other road-based transit are used more by low-income commuters, whereas subways and rail-based transit are used more by high-income commuters.

Figure 1 Share of commuters by household income whose primary means of commute is either road- or rail-based public transit

• Fact 2: Commuters who reside closer to transit stops are more likely to ride transit.
• Fact 3: Higher-income households reside closer to rail transit stops whereas lower-income households reside closer to road transit stops (as shown in Figure 2).

Figure 2 Median annual household incomes in census tracts by distance to nearest mass transit (either road or rail transit) stop for the average household in the tract

• Fact 4: Cities with more expansive and rail-based transit networks have higher transit ridership among commuters as well as a higher share of high-income commuters using mass transit (as shown in Figure 3). The additional transit riders in cities with faster transit networks tend to be higher income commuters who ride rail transit more.

Figure 3 For the three largest cities, the share of commuters by household income whose primary means of commute is public transit

• Fact 5: Housing prices are higher near both road and rail transit stops. As such, both types of transit are locational amenities that nearby residents pay a premium for better access to.  And while low-income households are more likely to pay a premium for access to bus transit, higher-income households are more likely to do so for access to rail transit.
• Fact 6: Neighbourhoods where transit ridership is low-income, changes in transit ridership are negatively correlated with changes in incomes (potentially due to poorer transit riders relocating to the neighbourhood when it is more accessible by transit and relocating away when it is less accessible by transit). In contrast, in neighbourhoods where transit ridership is high-income, increases in transit ridership coincide with increases in neighbourhood incomes (potentially driven by the relocation of high-income transit riders).

Figure 4 Change in median household incomes versus change in transit ridership in census tracts between 2000 and 2010

Now, consider a hypothetical city where travellers either drive or ride public transit and the only neighbourhood amenity is the neighbourhood’s public transit accessibility. When the city’s transit network is slow relative to the driving network, public transit is more attractive to low-income households who are more likely to sort into the more transit-accessible residential neighbourhoods and pay higher housing prices for it. High-income households, who ride transit less, are more likely to sort into the less transit-accessible neighbourhoods.

Consider a transit planner who is debating where to expand transit accessibility to improve public transit ridership in the city. They know that while the average transit rider in the city is low-income, the marginal transit riders (who are indifferent between driving and riding transit) are relatively high-income. So, the planner would expand transit accessibility to the neighbourhoods where the marginal high-income riders can benefit from them (assuming constant marginal cost of transit expansions). In other words, they would expand the transit network in relatively high-income neighbourhoods. Since doing so would make these neighbourhoods more attractive to all transit riders (not just the marginal riders), it would also attract disproportionately more low-income households (the average transit riders) from other neighbourhoods to relocate here.

As a result, in our hypothetical city with slow transit, the planner’s transit expansion choices lead to lower income segregation across neighbourhoods. This is because the planner expands public transit in areas where it is relatively less accessible. By reducing disparities in transit accessibility across space, they also minimise segregation by income.

Note, however, that the planner’s choices need not always lead to this outcome. Consider a city where the transit network is fast relative to the driving network. In this case, both the average and the marginal transit riders are high-income households. To improve ridership, the planner still expands public transit accessibility in relatively high-income neighbourhoods (that are already more transit-accessible than other areas), but doing so leads to even more high-income households relocating to these neighbourhoods and outbidding incumbent low-income households in willingness to pay the higher housing costs. In this scenario, the planner’s transit expansion choices lead to greater residential segregation by income.

In Akbar (2024), I also consider richer alternatives to this stylised hypothetical city, but the key insights remain the same. The relationship between public transit access and income segregation differ for ‘fast’  versus ‘slow’ transit networks and relies importantly on the choices of transit planners. Trying to improve transit ridership may inadvertently lead to greater segregation in cities with fast transit networks. However, public transit networks are slow in most US cities. The average transit riders and transit-accessible neighbourhoods are low-income. And expanding public transit accessibility to more neighbourhoods is likely to reduce income segregation and the urbanisation of poverty.

Source : VOXeu