Abstract: Preserving Affordable Housing through Predicting and Mapping Neighborhood Distress (Society for Social Work and Research 24th Annual Conference - Reducing Racial and Economic Inequality)

Preserving Affordable Housing through Predicting and Mapping Neighborhood Distress

Saturday, January 18, 2020
Marquis BR Salon 14, ML 2 (Marriott Marquis Washington DC)
* noted as presenting author
Sonja Payne, MSW, Community Health Mobilizer, Shippensburg Community Resource Coalition, Shippensburg, PA
Dorlisa Minnick, PhD, Associate Professor, Shippensburg University of Pennsylvania, Shippensburg, PA
Michael Lyman, PhD, Associate Professor, Shippensburg University of Pennsylvania, Shippensburg, PA

Housing is a major source of racial and economic inequality. Lower income residents are often on waiting lists for years for government programs such as public housing, Housing Choice Vouchers, 811 vouchers, and Low-Income Housing Tax Credit homes. Municipal officials and housing authorities struggle in maintaining and expanding affordable housing stock as they battle blight and market availability. One critical issue of this problem is the ability to identify neighborhoods at the start of the housing decline process when a small amount of financial resources can make the most impact in revitalizing the neighborhood to maintain affordable housing.

Large cities have used predictive analytics to identify neighborhoods in various stages of housing decline to stop the neighborhoods from declining further. No models have been created for small cities and rural towns, like so much of the geographic makeup in the United States. This study reports the creation of a predictive model of housing decline called the Neighborhood Early Warning System (NEWS). A pilot application of the model, in combination with GIS mapping, was developed for a small city in Pennsylvania to assist affordable housing authorities and non-profit agencies to maintain and expand affordable housing.


Eleven indicators of housing decline grounded in the literature (resident tenure, property tax arrears, housing code violations, vacancies, renter-occupied, poverty rate, water arrears, % of Black resident composition, crime rate, household income, and high school diploma/GED) were identified.  Local housing experts then created percentile rankings for each indicator to create an overall risk index score which was used to categorize neighborhoods into four, ordinal-scale stages: stable, bubble, declining, and distressed. Using four US Census Tract Block Groups, ArcGIS was then used to map the household data plus the Census indicators.


A preliminary analysis of the 11 indicators led to a refining of the model to nine indicators. The final model using the remaining nine indicators demonstrated a good fit when overlayed onto the Census Neighborhood Block data. The model accurately predicted stable, bubble, declining, and distressed neighborhoods. A GIS map of the community with the NEWS model data projection was presented to local housing professionals who confirmed that the predictions of the model match their own expert assessment of neighborhood status.


This study demonstrates how the NEWS model in combination with GIS mapping is helpful to County and non-profit organizations that are making difficult decisions about maintaining and expanding affordable housing. The NEWS predicted when and where to target neighborhood revitalization providing important data to support community organizations and key stakeholders’ funding applications for affordable housing as part of community development initiatives. Having neighborhood-level data to demonstrate funding need before the neighborhood becomes distressed is an asset essential to evaluating funding requests. Moreover, the model provides support to policy changes regarding sustainable funding and programming in helping neighborhoods remain stable after revitalization. Such changes are critical in the ongoing push to ameliorate the inequities that have been inherent in the US housing market.