Authors

Shengyang Du, Raoyichen Zou, Haoyu Li

Research Question

Middle neighborhoods as a concept have been discussed for roughly a decade and have had a fairly loose narrative definition during that time. What robust statistical definition can be used to accurately identify middle neighborhoods in seemingly disparate regions of the country to facilitate comparison across regions as well as best practice and policy sharing?

Background

Middle neighborhoods are traditionally home to the middle and working class; they have high rates of homeownership and are largely made up of single-family (1-4 units) housing stock. The housing market is less competitive than other areas and can be a good source of NOAH (naturally-occurring affordable housing) as well as municipal tax revenue. Middle neighborhoods are typically more diverse, both economically and racially/ethnically. They are an extremely important part of NCST's mission to close the racial wealth gap, given that homeownership is both the best way to build household wealth in this country and is increasingly out of reach for many of the types of folks that traditionally form the population base of middles.

Middle neighborhoods are also in decline virtually everywhere we have looked. Because most residents do not meet income thresholds required for much public assistance, middles are in danger of both decline, which raises the cost of public or non-profit intervention, and gentrification and displacement, which disproportionately affects minority residents and low-income renters.

Up to this point, the NCST has been using median income as a proxy for middle neighborhoods, but this does not paint a complete picture and is not even applicable in all housing markets. NCST wants to develop a framework for defining middle neighborhoods that is general enough to apply to places as different as NYC and Detroit, but can still be used confidently in statistical analysis.

Methodology

Up to this point, the analysis of middle neighborhoods has focused on a single metro or city at a time, both as a result of market dynamics shifting from place to place and the size of the currently utilized dataset. In order to develop a national model that is suitable everywhere in the country, the capstone team may choose to utilize AI/machine learning techniques to scale the solution, or advanced statistical analysis, such as K-means clustering, to create a universal framework. Spatial analysis is also important and the use of GIS software to map neighborhoods are valuable to the team.

Deliverables

• Development of a metro typology of middle neighborhoods to facilitate comparison across regions as well as best practice and policy sharing.
• Middle neighborhoods dashboard to supplement the existing online map or a predictive model to identify future middle neighborhoods (2030).

Data Sources

• National Community Stabilization Trust (NCST)

• U.S. Census Bureau