The purpose of this study was to answer the following questions:
What is the effect of foreclosures on the levels of crime in a neighborhood and how does that relationship change over time? Do the two phenomena have a recursive relationship?
Do foreclosures in one area have a "spillover" effect, increasing crime in a neighboring area at an immediate or later time period?
How do the effects of foreclosures on crime differ in the short, medium, and long term?
What are the perceptions of key informants and residents on foreclosures and crime in their neighborhoods, on the impact of foreclosures on the crime rate, and on the best approaches to addressing the spillover effects of the foreclosure crisis?
This study simultaneously modeled the temporal and spatial effects of foreclosures on neighborhood crime levels and of crime on neighborhood foreclosure rates.
Foreclosure indicators in the District of Columbia were obtained from the D.C. Recorder of Deeds (ROD) and the D.C. Office of Tax and Revenue (OTR). NeighborhoodInfo DC, a local data intermediary operated by the Urban Institute, processed the raw data from these agencies to produce the following three indicators; number of properties with a foreclosure start, properties in the foreclosure inventory, and number of properties with completed foreclosures. Address-level incident data with geographic coordinates for District of Columbia was obtained from the Metropolitan Police Department (MPD). The offenses were classified into personal (violent) and property offenses. Personal offenses included homicide, sexual offenses, assault, and robbery. Property offenses included burglary, theft, motor vehicle theft, and theft from a motor vehicle. These data were aggregated into quarterly counts by census tract for the period January 2003 through December 2010, a total of 32 quarters of data for each of 188 census tracts
Foreclosure data were obtained from the Clerk of Courts for Miami-Dade County, Florida. The data included property case data, all civil case data, and the docket file from raw recorder files. From the case and docket files, foreclosure filing and sales information were extracted. Urban Institute purchased complete parcel data and property sales data for the period August 2003 through April 2011 for Miami-Dade County from the Miami-Dade County Property Appraiser. These data were matched to the foreclosure filing and sales data to add property information and property address to the County Clerk's data. Parcels were then matched to census tracts in order to create a count of foreclosure sales by census tract. Analysis was only carried out with foreclosure sales data as it was not possible to obtain data on foreclosures that did not end in a sale. Crime report data was obtained from the Miami-Dade County Police Department and the City of Miami Police Department. The offenses were classified into personal (violent) and property offenses. Personal offenses included homicide, assaults, and robbery. Property offenses included burglary, theft, motor vehicle theft, and theft from a motor vehicle. Both the foreclosure and crime data were aggregated into quarterly counts by census tract for the period October 2003 to March 2011, a total of 30 quarters of data for each of 329 census tracts.
To gain a local perspective on the foreclosure crisis, investigate the nature of the neighborhoods that might have been most affected by the foreclosure crisis, and to understand foreclosures' potential effects on crime, local experts were interviewed. In the District of Columbia nine individuals representing a variety of local groups, most of which provided community-based housing assistance or focused on community development efforts, and a focus group of 14 residents were interviewed in person and by phone. From Miami-Dade County, Florida five individuals representing a variety of local groups, including Florida International University, a community-based housing assistance organization, a local grass-roots organizer, and an organization doing a wide array of social service provision and policy work in South Florida, with an emphasis on housing were interviewed in person.
District of Columbia and Miami-Dade County, Florida were selected for the following reasons:
Levels of both foreclosures and crime were sufficiently high in each site to make statistical analysis of both phenomena feasible for small geographic units.
Foreclosure rates in Washington, D.C., were much lower than those in Miami, creating an opportunity to compare impacts in two cities with different foreclosure experiences and responses.
Crime rates for each city were higher relative to national levels.
Detailed data on foreclosures and crime that were required for the statistical analyses were readily available.
Large part of western Miami-Dade County, Florida was not included as census tracts were much larger than those near the coast and did not contain or had very few housing units. Also, census tracts not in the jurisdiction of Miami-Dade County Police Department or City of Miami Police Department was dropped as crime data was not available.
Time Series: Discrete
All foreclosures and crime committed in the District of Columbia and Miami-Dade county, Florida between 2003 and 2011.
City of Miami Police Department
District of Columbia Office of Tax and Revenue
Metropolitan (District of Columbia) Police Department
United States Census Bureau
District of Columbia Recorder of Deeds
Miami-Dade County Clerk of Courts
Miami-Dade Police Department
administrative records data
geographic information system (GIS) data
District of Columbia data consists two GIS shape files and two CSV text files.
DCTimeSeriesData.csv CSV text file (n=6,016) contains nine variables on unique id, census tract id, year, quarter, total crime count, property and violent crime count, foreclosure inventory, and foreclosure sales.
DC_SWM_Q_RowStd.csv CVS text file (n=1,130) contains three variables on unique id, neighborhood id, and weight.
DCTracts00Crime_Forecl.shp GIS shapefile (n=188) contains variables on counts of total, violent, and property crimes, foreclosure sales, and foreclosure inventory for years 2003 to 2010.
DCTracts10Demog.shp GIS shapefile (n=179) contains variables on percent living below poverty level, percent black population, percent white population, and median housing value.
Miami-Dade Country, Florida data consists of two GIS shape files and two CSV text files.
MiamiTimeSeriesData.csv CSV text file (n=9,870) contains eight variables on unique id, census tract id, year, quarter, total crime count, property and violent crime count, and foreclosure sales.
Miami_SWM_Q_RowStd.csv CSV text file (n=1,774) contains three variables on unique id, neighborhood id, and weight.
MiamiTracts2010_CrimeForecl.shp GIS shapefile (n=329) contains variables on foreclosure sales, total crime count, and counts of violent and property crimes for years 2003 to 2011.
MiamiTracts2010_Demog.shp GIS shapefile (n=519) contains variables on percent living below poverty level, percent black non-Hispanic population, percent foreign born population, percent Cuban born population and median housing value.