ABSTRACT
This article uses eviction data to test the transit-induced displacement hypothesis—that the placement of new transit stations will lead to elevated property values, gentrification, and displacement. We use a case study of four cities in the United States that built or extended rail lines between 2005 and 2009: Newark, New Jersey; San Diego, California; Seattle, Washington; and St. Louis, Missouri. We employ a combination of propensity score matching and difference-in-differences modeling to compare eviction filing rates in gentrifiable neighborhoods near new transit stations with a set of similar neighborhoods not close to the station. We find very limited evidence that new transit neighborhoods experienced heightened rates of evictions compared with the controls. In three of the four cities, the effect of the opening of the station on eviction rates was insignificant. Eviction rates did spike in St. Louis immediately following the opening of the line, but this time period also coincided with the financial crisis.
Disclosure Statement
No potential conflict of interest was reported by the authors.
Notes
1. A project directed by Matthew Desmond and designed by Ashley Gromis, Lavar Edmonds, James Hendrickson, Katie Krywokulski, Lillian Leung, and Adam Porton. The Eviction Lab is funded by the JPB, Gates, and Ford Foundations as well as the Chan Zuckerberg Initiative. More information can be found at evictionlab.org
2. Several lines that opened between 2005 and 2011 are not included in the analysis. These include the Blue Line opened in Charlotte, North Carolina, in 2007, because of missing evictions data prior to 2004; the MAX Green Line in Portland, Oregon, which opened in 2009, because of too much overlap with existing stations; the Valley Metro Rail opened in Phoenix, Arizona, in 2008, because of missing evictions data post2005; and Santa Clara Valley Transportation Authority’s Vasona line in San Jose, California, opened in 2005, because of missing evictions data between 2002 and 2013.
3. For this step, we use backward selection for our stepwise regression—meaning we start with a predictive model containing all the variables used for the matching procedure as predictors of receiving a light rail station (a dummy variable coded 1 if considered a station neighborhood). Then the algorithm iteratively eliminates predictors one at a time, at each step considering whether the model selection criterion Bayesian Information Criterion (BIC) will be improved by adding back in a variable removed at a previous step. The resulting model contains a subset of predictors that are most contributive in predicting the outcome variable (i.e., receiving a light rail station; RDocumentation (Citation2020) and references therein).
4. These results are available from the authors upon request.
Additional information
Funding
Notes on contributors
Elizabeth C. Delmelle
Elizabeth C. Delmelle is an associate professor of geography at the University of North Carolina at Charlotte. Her research interests include neighborhood dynamics, transportation, and geographic information science.
Isabelle Nilsson
Isabelle Nilsson is an assistant professor of geography at the University of North Carolina at Charlotte. Her research interests include transportation, housing, and local economic development.
Alexander Bryant
Alexander Bryant is a recent graduate of the undergraduate program in geography at the University of North Carolina at Charlotte. He is currently employed as a GIS technician for the city of Charlotte.