ABSTRACT
Ghost dwellings are defined as housing units that are vacant but not abandoned nor available for rent or sale; their owners only occasionally reside in them. In big cities, they are thought to contribute to various ills such as housing unaffordability and a deadening of urban life and culture. This study provides the first systematic assessment of the spatial extent of ghost dwellings in the 50 largest cities in the United States. It quantifies their concentration at the city, neighborhood, and building scales. In keeping with the tenor of recent media reports, ghost dwellings are increasing rapidly in most big cities, albeit from a small base. They concentrate in a small number of highly privileged and well-known neighborhoods. The epicenters are a surprisingly varied collection of cities, ranging from globally renowned Miami and Las Vegas to the less well-known Austin and Atlanta. A loose test of several possible hypotheses driving ghost dwelling growth on the basis of locational patterns finds the strongest support for exclusion and retirement orientation as mechanisms. In certain cities, ghost dwellings are strongly associated with particular building formats such as high-rises and condominiums.
Acknowledgments
The author thanks Amruta Ponkshe for her assistance in researching Census Bureau definitions of vacant housing. This research was supported by a Summer Research Assignment grant from the School of Architecture at the University of Texas at Austin.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1. Fully 70% of low-income whites in the largest metros in the U.S. live in suburbs, as compared to only 41% of low-income African Americans. Measured another way, whites make up only 24% of the low-income population in big cities, versus 44% in the suburbs (Kneebone, Citation2017).
2. Data from the year 2000 comes from the long form Census, also a sample-based means of data collection, which preceded the ACS and was discontinued upon its introduction. Census 2000 long-form data and ACS data can generally be safely compared.
3. I classified vacant housing units in AHS microdata as ghost dwellings if the VACANCY variable was equal to 6 (“held for occasional use throughout the year”), 8 (“seasonal-summer only”), 9 (“seasonal-winter only”), 10 (“other seasonal (SPECIFY)”), or 11 (“migratory”). The other possible categories for VACANCY are 1 (“for rent only”), 2 (“for rent or for sale”), 3 (“for sale only”), 4 (“rented, but not yet occupied”), 5 (“sold, but not yet occupied”), 7 (“other (specify)”), or B or −6 (“not applicable”).
4. The Retail Access Index is reported at the block group level, as part of the Location Affordability Index, Version 2.0, published in 2014. It quantifies the density of retail jobs per square mile in the vicinity of the block group, ranging from 75 to almost 94,000 in the seven cities studied. Here I have aggregated it to the census tract level using a weighted average by block group population. Technical documentation can be found here: https://www.hudexchange.info/resources/documents/Location-Affordability-Index-Data-and-Methodology-V2.0.pdf.
5. Refer to the technical appendix for an explanation of Coefficient of Variation. In a nutshell, under 12 indicates reliable results, 12 to 40 implies moderately reliable results, and over 40 indicates unreliable results.
6. The excluded tracts have 482 or fewer housing units. This is the fifth percentile of the housing unit count for all 3,969 tracts from all 11 counties (the primary county containing most or all of each of six of the seven cities, plus the five boroughs of New York City) analyzed together.
7. Miami Beach contains South Beach, a neighborhood with a namesake beach and a famous collection of Art Deco buildings. Via countless films and television programs it has come to be the avatar for the popular conception of all things “Miami”—even though it is a separate municipality from the City of Miami. Miami Beach in 2017 had over 17,000 ghost dwellings, exceeding even the much larger City of Miami, which had just over 10,000. Miami Beach’s ghost dwellings represented fully 25% of the total housing stock.
8. Tract-level median household incomes in the ACS are “top-coded” at $250,000, meaning that such incomes are reported as $250,000 in any tract where the median exceeds this amount, no matter by how much. Thus, it is possible that the transition from elite to ultra-elite status in certain neighborhoods is not detectable in certain areas using this indicator.
Additional information
Funding
Notes on contributors
Jake Wegmann
Jake Wegmann joined the faculty of the Community and Regional Planning program at the University of Texas in Austin in 2014. He received his doctorate from the University of California at Berkeley. His research lies at the nexus of housing affordability, urban economics, and land use regulation, with an emphasis on housing supply in hot markets. Prior to academia, he worked in the affordable housing development industry with several for- and nonprofit entities in Denver and San Francisco.