The Conventional Wisdom on Food Deserts May Be All Wrong

The researchers worked with data, from the Nielsen Datasets at the Kilts Center for Marketing, that include grocery purchases by some 60,000 households and grocery sales at about 35,000 stores nationwide from 2004 to 2015. Allcott, Diamond, and Dubé also included the locations of 1,914 supermarkets to study the impact that new supermarkets have on healthy eating in food deserts.

People living in food deserts are widely thought to end up buying unhealthy food at drugstores or convenience stores, which sell little to no fresh items. “Food deserts are often short on whole food providers, especially fresh fruits and vegetables, instead, they are heavy on local quickie marts that provide a wealth of processed, sugar, and fat laden foods that are known contributors to our nation’s obesity epidemic,” reads the website of the American Nutrition Association.

But the data tell a different story. When a supermarket entered a food desert, consumers stopped traveling to farther supermarkets, according to the research. About 80 percent of the increase in spending share consisted of sales diverted from other large grocery stores. Meanwhile, food spending at drugstores and convenience stores dropped by only a fraction of a percentage point.

Americans are willing to travel long distances to shop, the researchers say, arguing that the notion of a food desert may therefore be misleading. They cite data from the 2009 National Household Travel Survey suggesting that even if the closest supermarket isn’t nearby, people travel to shop at it.

Households that earn less than $25,000 a year travel an average of 5 miles each way to buy groceries, clothing, and household hardware, according to the data, while those that live in a food desert travel 7 miles. Even people who are poor, live in a food desert, and don’t have a car still travel 2 miles. The data from the travel survey support another of the researchers’ findings, that households in food-desert zip codes buy almost 90 percent of their groceries from supermarkets.

Having a supermarket close by would reduce travel costs for households in food deserts but not necessarily change what shoppers choose to eat, the research suggests. For a variety of reasons that researchers and policy makers need to better understand, people’s shopping habits are driven by their food and taste preferences.

The researchers tested whether location drives preferences, to understand whether living in a food desert could cause residents to prefer certain types of foods, perhaps those that are most shelf stable. They ruled this out, finding instead a striking relationship between food preference and income. Then they developed an empirical method to estimate the preferences that the households in the panel had for fruits and vegetables, as well as various nutrients such as saturated fat, sugar, and salt. “Since that panel is collected to be nationally representative, we interpret our findings as representative of the US population of households,” says Dubé.

They find that richer households tend to place a higher value on healthy foods and nutrients, while poorer households tend to value unhealthy ones. High-income households (making more than $70,000 a year) are willing to pay almost double for vegetables and nearly three times more for fruit, the researchers estimate. By contrast, low-income households (making less than $25,000 a year) are willing to pay more for sugar and saturated fats.

This evidence suggests that preferences may be driving the nutrition gap, which has widened in recent years. The health index—a measure of the nutrition content of households’ grocery purchases—improved five times more for high-income households than it did for low-income households between 2012 and 2015, compared to 2004 through 2007.

What if everyone had the same access to healthy food, with the same prices? In that case, the nutrition gap would have shrunk by 9 percent, the researchers determine. But the researchers used their preference estimates to measure another counterfactual scenario, that everyone shared the same food and nutrient preferences. That would have shrunk the gap by 91 percent, they say.

The researchers find that education level and nutrition knowledge explain a quarter of the nutrition gap. These findings suggest that policies aimed at nutrition education could be more effective than subsidies and grants meant to encourage building more supermarkets in food deserts.