The geography of financial exclusion.

Overview.

Banking deserts — areas with minimal access to traditional bank branches — disproportionately affect low-income communities across the United States. The question isn't whether this pattern exists; it's where it's sharpest, and what that geography tells us.

This project maps the intersection of two county-level variables: FDIC bank branch density per 10,000 residents, and median household income from the Census Bureau's American Community Survey. The output is a bivariate choropleth covering 3,200+ U.S. counties, revealing geographic patterns of financial exclusion that single-variable maps miss.

Counties shown in dark purple/navy represent the most underserved areas: low income and low branch access. Deep green counties are well-served on both dimensions. The interesting geography is the off-diagonals — places where branch density and income diverge — and the persistent regional clusters that emerge when you look at both at once.

Methodology & pipeline.

Data sources

• Bank branch locations: FDIC Summary of Deposits, aggregated to county and normalized per 10,000 residents.
• Income data: U.S. Census Bureau American Community Survey (ACS) 2023, median household income by county.
• Geographic boundaries: Census TIGER/Line county shapefiles, simplified for web performance.

Classification approach

Each county is classified along two axes using tercile breaks (low / medium / high), producing a 3×3 bivariate grid with nine possible categories. This balances interpretability with analytical nuance — more granular than a simple above/below median split, but not so complex that the visual becomes unreadable.

Branch density is calculated as branches per 10,000 residents rather than raw branch counts, which controls for population differences between rural and urban counties. A rural county with 3 branches serving 5,000 people has better per-capita access than an urban county with 50 branches serving 500,000; raw counts would invert that ordering.

Key findings.

Several patterns emerge from the spatial analysis — not all of which would show up in either single-variable map.

Geographic concentration

Banking deserts cluster in specific regions rather than distributing uniformly. The Deep South, Appalachia, and parts of the rural West show persistent overlap between low income and low branch density. These aren't random — they reflect decades of demographic and economic patterns, including population decline, industry loss, and branch consolidation.

The rural consolidation problem

Rural counties face a compounding dynamic: as populations decline, banks close branches due to low profitability; reduced local financial access can accelerate economic decline; which drives further closures. That feedback loop is a classic complexity pattern — individually rational decisions (closing unprofitable branches) producing collectively harmful outcomes.

Urban exceptions

Not all low-income counties are banking deserts. Urban areas tend to maintain higher branch density even in lower-income neighborhoods — partly because population density makes branches viable, partly because of Community Reinvestment Act requirements. Branch density alone, however, doesn't capture quality of access: fees, minimum balance requirements, and service hours matter too, and aren't visible in this map.

A complexity lens.

Banking deserts aren't a simple resource-allocation problem. They're an emergent property of interconnected economic, demographic, and regulatory systems. A few complexity concepts help explain the patterns visible in the map.

Feedback loops

Branch closures and economic decline reinforce each other. When a branch closes in a small town, residents travel further for basic financial services. Some shift to costly alternatives — check-cashing, payday lenders. The resulting drain on household finances suppresses local economic activity, making the area less attractive for financial institutions. Each turn of the loop makes the next closure more likely.

Path dependence

Today's banking-desert map reflects historical patterns — redlining, industrial shifts, highway-construction decisions — that shaped where capital flows and where it doesn't. These aren't easily reversed, because the current state of the system constrains future possibilities. The map is a snapshot of a long process, not a present-tense anomaly.

Fintech as partial solution

Digital banking is often proposed as the answer, but it faces its own complexity challenges. Broadband access is uneven — often worst in the same areas with the fewest branches. Digital literacy varies. Many financial services still require in-person interaction. Technology can complement physical infrastructure; it doesn't fully substitute for it.

Technical notes.

The map is built as a static web application with no server-side processing. County boundaries and merged data are served as GeoJSON files, rendered with Leaflet.js on a CartoDB dark basemap. The bivariate color scheme uses a palette that stays distinguishable across all nine categories.

Interactive features include click-to-inspect popups with exact branch density and income figures for each county, plus a bivariate legend that doubles as a visual guide to the color scheme. The entire application loads client-side, suitable for static hosting on Cloudflare Pages — which is where it runs.