Applied economics research examining food access, affordability, and availability across California. We challenge conventional wisdom about food deserts using rigorous data analysis of grocery store locations, SNAP enrollment, and transit accessibility.
The term "food desert" has dominated policy discussions about hunger in America for two decades. The concept is intuitive: if people live far from grocery stores, they cannot access healthy food. Build more stores, solve the problem. Yet when we analyzed 6,613 grocery stores across California and matched them to 9,033 census tracts, we found something unexpected.
Zero food deserts exist in Santa Clara County by the federal definition. Every resident lives within one mile of a supermarket. Yet SNAP (food stamp) participation varies 4.7-fold across demographically similar communities. Some neighborhoods with identical income levels and grocery access show enrollment rates of 3%, while others exceed 14%.
This finding challenges the core assumption of food desert policy: that geographic proximity determines food security. Our research suggests the relationship is more complex.
When we mapped grocery store density against food insecurity indicators, we discovered what we call the vulnerability paradox: the most economically vulnerable communities often have the best geographic food access. Dense urban neighborhoods with high poverty rates typically have multiple grocery options within walking distance. Wealthier suburban areas, by contrast, may require a car to reach the nearest store.
This pattern makes sense when we consider how grocery stores choose locations. They follow population density and purchasing power. Low-income urban neighborhoods offer both: concentrated populations and consistent demand for staple goods. The result is adequate geographic access but persistent food insecurity.
If proximity were sufficient, these neighborhoods would show the lowest rates of food insecurity. They do not.
Our research identifies four factors that better predict food insecurity than distance to the nearest grocery store:
These findings suggest that building more grocery stores—the primary policy response to food deserts—addresses only part of the problem. Stores already exist in most high-need areas. The barriers are economic and logistical, not geographic.
More promising interventions might include:
The food desert framework is not wrong—geographic access matters. But it has become a convenient shorthand that obscures the economic and structural factors that actually determine whether families can put food on the table.
Our food security research uses multiple data sources and methods:
All analysis code and processed data are available in our public replication repository.
Summary of our food security research findings
Santa Clara County has no federal food deserts. Every resident lives within 1 mile of a supermarket, yet food insecurity persists.
Enrollment varies 4.7-fold across similar communities. Geography alone cannot explain this disparity.
Transit accessibility predicts food insecurity better than store proximity. 2.7 million routes analyzed.
Most vulnerable communities often have the best geographic access. The problem is economic, not geographic.
Our food security research organized by focus area
Analyzing the relationship between grocery store locations, transit routes, and food security outcomes.
How income, employment, and housing costs affect food security beyond geographic access.
Data validation, index construction, and methodological innovations in food security research.
The USDA's food desert framework assumes distance to grocery stores drives food insecurity. Across 9,033 California tracts, the tracts with the best geographic access often have the highest vulnerability, an inversion that reframes what food-access policy should target.
Analysis of 6,613 grocery stores reveals zero food deserts in Santa Clara County by federal definition, yet SNAP participation varies 4.7x across similar communities.
County-wide SNAP rates rose 2 points over four years. But what happens when we look at neighborhoods instead of the whole county? A more complex picture emerges from longitudinal analysis of census tract-level food security.
County-wide SNAP rates dropped -0.09% but high-vulnerability tracts increased +0.76%. Difference-in-differences analysis reveals the differential pandemic impact across census tracts.
Even in America's wealthiest county, thousands of working families qualify for food assistance but don't enroll. We examine why.
Combining food access, transit accessibility, and economic indicators into a single index that predicts food insecurity better than any single measure.
A food desert is a geographic area where residents have limited access to affordable, nutritious food. The USDA defines it as an area where at least 500 people or 33% of the population live more than 1 mile from a supermarket (10 miles in rural areas). However, our research shows this definition misses critical factors like transit access and economic barriers.
Not necessarily. Our analysis of 6,613 grocery stores in Santa Clara County found zero federal food deserts, yet SNAP participation varies 4.7x across similar communities. Proximity is necessary but not sufficient for food security.
Our research identifies transit accessibility, housing cost burden, employment stability, and enrollment friction as stronger predictors than geographic proximity alone. The "mobility desert" concept captures locations where stores are nearby but unreachable by transit.
New food security analysis delivered to your inbox. No spam, unsubscribe anytime.
Subscribe for Free