Transit-induced gentrification and displacement: future directions in research and practice

Introduction

For the last fifty years, researchers have studied the relationship between new urban rail investments and property prices, often finding significant price premiums for houses located closer to stations (Debrezion et al., 2007). But it was not until recently that some started wondering: What does this mean for the transit-adjacent neighborhoods and particularly, their residents? Thankfully, this question has attracted substantial interest from the academic community. But we have yet to answer the most challenging portions of this question, such as the extent of household and business displacement that takes place due to loss of affordability and other changes (partially) brought by transit and transit-oriented development (TOD). In this editorial, I explore what we have yet to research, learn, and disseminate, compared to what we already know, emphasizing the need for more quantitative research that will answer these difficult questions.

What conditions lead to gentrification?

The approval and construction of a new urban rail line near a city center often comes with rezoning and private investment, transforming some station areas (especially those close to the downtown) into high-end neighborhoods of mixed-use, high-density developments, attracting residents interested in and able to afford the brand-new amenities. Beyond the metamorphosis of the areas in the immediate proximity of urban rail, surrounding neighborhoods can also become attractive and may experience a surge in new construction, renovations, and rent and house price increases (Bardaka et al., 2018).

Transit-induced gentrification studies seek to quantify the socioeconomic changes neighborhoods close to new transit systems experience because of public and private investments, the in-migration of upper-class populations, and the potential out-migration of original residents, as part of involuntary or voluntary displacement. To measure gentrification, researchers have examined the temporal changes in various socioeconomic indicators, including educational attainment, household income, occupation type, and race (Bardaka et al., 2019). San Francisco, Toronto, Denver, and Charlotte are some of the North American centers that seem to have experienced gentrification in relation to urban rail investments and TOD, while other urban areas may have not (Padeiro et al., 2019). Studies that explored changes across multiple urban areas have provided some of the contextual attributes that could be instrumental. For example, Kahn (2007) examined the changes in the census tracts close to urban rail transit stations in 14 metropolitan areas across the U.S. between 1970 and 2000. His study revealed that census tracts close to ‘walk and ride’ stations experienced a significant increase in educational attainment, income, and house prices, in comparison to census tracts that were not close to stations but were within 20 miles of the respective city center. Conversely, ‘park and ride’ stations (stations with parking facilities, located in areas of lower walkability, farther from city centers) often experienced a decline in property prices and a rise in poverty (Kahn, 2007). In a recent study, I also found larger impacts for a line traversing through the urban core compared to lines going through suburbs for the same metropolitan area (Bardaka et al., 2019).

These results highlight the significance of locational and built-environment features, such as population density, walkability, and distance from the city center, in gentrification outcomes. Still, more research is needed to help us better understand key contextual factors, and particularly whether the initial socioeconomic characteristics of the transit-adjacent neighborhoods (e.g., income and racial composition attributes prior to the announcement of a transit line) play an additional role in these processes. Moreover, the availability and frequent updates of street-view images (e.g., through Google Street-View) and the advent of machine learning algorithms that can process this information could provide more detailed information on the changes in infrastructure and real estate. This data, combined with more traditional sources, can help us get a more complete picture of neighborhood transitions around transit stations.

Studying household displacement is a priority and a challenge

As of today, researchers struggle to move past estimating neighborhood-level changes, and conclusions about household displacement have not been reached. Census data have enabled research on temporal differences in a plethora of socioeconomic indicators, but whether, for instance, the increase in median household income that one finds is due to higher-income households moving in, or lower-income households moving out, or both, remains largely unknown in such studies. As new multi-family housing is often constructed on empty land as part of TOD, it would be possible that a neighborhood depicts significant changes in aggregate terms without any of the original residents moving out. Finding evidence that supports or contradicts such scenarios has been a focal point in recent work (Baker et al., 2021). But the effort to study migration and displacement around transit stations has been met by serious data restrictions and sample limitations, mainly because of the lack of adequate information on individual households.

Gaining permission to access confidential census data requires substantial effort that may outweigh the benefits, especially given the large temporal gaps (e.g., ten years) between consecutive observations that restrict researchers from identifying when households move in and out (McKinnish et al., 2010). Household surveys in affected areas may not be able to provide concrete conclusions not only due to small sample sizes but also because they primarily capture people’s views and stated preferences but not their actions (Brown and Werner, 2011). Studies that analyze eviction records provide important findings for evicted households but naturally exclude the majority of the population who moves after a unit becomes unaffordable without legal force or the official involvement of a court (Delmelle et al., 2020). One exception is a study by Boarnet et al. (2018), who gained access to individual tax records in Los Angeles (LA) County and then geocoded one third of the population who filed income taxes and lived within 0.5 miles of LA Metro stations between 1993 and 2013. They found that households earning up to 30% of the area’s median income became less present in station areas after rail operation began, while those earning between 30% and 50% of the area’s median income had a lower likelihood of moving out of station areas compared to other households.

The emergence of longitudinal consumer databases, originally compiled for marketing purposes, has opened new avenues for research on migration around transit (Baker et al., 2021). Private companies collecting this data from a variety of public and private sources seek to cover the full population of households, which has a large advantage for research. The address of each household is verified annually, meaning that the moves of individual households can be traced at a high spatial and temporal resolution. The most noteworthy disadvantage of such a dataset is the source of the socioeconomic attributes accompanying each household record; some of these attributes, including income, are estimated using proprietary models and thus research cannot assess their validity. Nevertheless, as this data continues to improve, this information has the potential to substantially enhance our understanding of the types of households that have a higher probability of leaving the neighborhoods nearby stations and provide insights into the extent of displacement taking place during the different phases of public transit construction and operation.

What about businesses?

We lack substantial research and awareness of business impacts felt by transit. This could be one reason for the scarcity of programs and policies to support businesses in TOD environments. A few analyses have revealed that urban rail construction could be disruptive for businesses, reducing their visibility and accessibility for pedestrians and cars (Tornabene and Nilsson, 2021). However, businesses could experience impacts beyond the project construction. On one hand, businesses could benefit from the extra foot traffic brought by the operation of a new urban rail line and the increase in population locally due to new housing developments. Business starts have been shown to rise as new businesses find the area more attractive and occupy the commercial space added through TOD (Credit, 2018). On the other hand, it is unknown whether small, local businesses are disproportionately impacted by the increase in commercial property values and the potentially different preferences of the new clientele. Some commercial areas close to transit could be gentrifying, experiencing an influx of high-end stores and the closure of minority-owned and local businesses offering everyday products (Liu and Bardaka, 2023). Historical business microdata has traditionally been the main data source for testing such hypotheses, but because of limited access (often due to prohibitive costs), these topics have remained largely under-researched. Computer vision could be an appealing path forward. For example, text from storefronts acquired from street-level images can be used for identifying and classifying businesses (e.g., grocery store, pharmacy, wine store). Through a spatiotemporal analysis of street images, business openings and closures could be recognized, and questions on the impacts of transit investments on various types of businesses could be answered.

Practical implications and next steps

Initial research efforts to understand the socioeconomic implications of public transit investments used relatively simple estimation methods, such as descriptive and correlative approaches. Challenges to these methods soon appeared. First, other factors at the national, regional, and local levels impact prices and, consequently, people and businesses. Therefore, the full changes taking place in transit-adjacent neighborhoods cannot be attributed solely to transit. Second, transit lines are not randomly drawn on a map. An urban rail line may include the city’s downtown as well as other nearby areas that could be densely populated, have demonstrated development potential, or may not have strongly opposed the project. In many regions, city centers and inner suburbs are already in the process of gentrifying. These observations led to study designs that use ‘control groups’ (carefully selected areas that share common attributes with station areas) that allow us to identify the causal influence of transit and TOD (Schimdt et al., 2022).

But as analysis methods became more sophisticated, the study outcomes became less accessible and more difficult to grasp, inhibiting impactful dissemination. Furthermore, research partially overlooked that what households and businesses experience in reality is not the estimated causal impact but the total change in their expenses (e.g., rent, property tax) and the full pressures of gentrification. Although it is indeed problematic to attribute the total, social, and economic changes of a neighborhood to transit and TOD, it is equally problematic to discount people’s experience and only discuss it in comparison to potentially similar others. Simple, descriptive calculations that can portray the full magnitude of the changes faced by various population groups and businesses should still be incorporated into studies using advanced methods. Finding negligible or statistically insignificant impacts should also not be a reason for celebration or idleness, but another call for action, so that we start seeing consistent, significant benefits to those who need transit the most.

Public agencies interested in addressing issues related to gentrification have concentrated their efforts on developing income-restricted units close to transit stations (often termed as equitable TOD) (Bardaka and Hersey, 2019). However, building affordable housing is not a panacea, and additional programs may need to be included to help alleviate the pressures existing businesses and households may experience. For example, voluntary displacement could be reduced by programs that inform homeowners about the housing market and the dangers of selling their property through informal channels that offer cash fast (e.g., ‘We buy houses for cash’ signs) (Preis et al., 2023). With a better understanding of population and business displacement, researchers will be able to provide more targeted policy recommendations to the decisionmakers.

Along with more research, it is also critical that the public sector has improved capacity to act on the research findings. Instituting federal programs that can fund longer, more integrative planning activities and cross-agency collaborations is critical. Research efforts will continue to reveal how different urban processes play a role in shaping neighborhoods. But irrespective of whether transit investments are more responsible or less responsible for gentrification, ensuring that public transportation is first made accessible to those in need should be the overarching goal and a multi-agency societal obligation.