Authors

Abraham Mathew, Mengyang Yu, Yuwei Han

Research Question

Where can the middle class actually afford to live across diverse US metropolitan areas, and how can metros be grouped by the characteristics of their middle neighborhoods? The project scales a tract-level classification framework nationally and develops a metropolitan typology to help target neighborhood stabilization where it is needed most.

Background

Middle neighborhoods, the safe, affordable communities that shelter working- and middle-class households, are a critical but overlooked part of the US housing system. They support local tax bases, foster social stability and buffer against concentrated poverty, yet they are in steady decline almost everywhere. Because they do not meet the distress thresholds for most public housing assistance, they are rarely prioritized until they reach a visible crisis, when reversing disinvestment or mitigating displacement becomes far more costly. Preserving them is also tied to closing the racial wealth gap, since middle-class homeownership remains the main vehicle for household wealth in the US.

The 2025–26 capstone established a dual Plan A (data-driven) and Plan B (rule-based) tract-level classification across Cleveland, Baltimore and San Francisco. It also showed that rigid national affordability filters fail in high-cost regions, finding zero core middle neighborhoods in San Francisco.

Methodology

The project involves:

• Expanding the prior framework to a wider, more diverse set of US metropolitan areas.
• Applying advanced statistical modeling or machine learning to group metros by middle neighborhood attributes such as housing pressure, residential stability and diversity.
• Possible areas of focus include:
  • Automated regional recalibration using dynamic, percentile-based thresholds.
  • Data-driven clustering for the metro taxonomy.
  • Integrating more recent housing market data to map spatial "spillover" pressures between adjacent neighborhoods.

Deliverables

• Data visualizations
• GitHub repository with documentation
• Policy recommendations
• Technical report
• Machine learning model
• Interactive data dashboard

Data Sources

• Tract-level housing and demographic data used in the 2025–26 CUSP capstone framework
• More recent housing market data