Sunday, April 22, 2012

Supermarket deserts by the numbers

Gina Kolata in the New York Times this week cast doubt on claims that supermarket deserts contribute to the obesity epidemic.  The start of her article cites recent research that finds no association between supermarket deserts and risk of obesity.  It notes that residents of low-income urban neighborhoods have as much access to supermarkets as residents of higher-income neighborhoods have. 

This fact at first seems counter-intuitive to most people concerned about supermarket deserts, but it is easy to understand with some further reflection.  Low-income urban neighborhoods commonly have high population density, and they contain many medium-income residents along with the impoverished residents, so they sometimes offer too big a market for retail chains to overlook. 

The NYT article generated some controversy.  Conservative pundits, as you might imagine, falsely claimed that food deserts are a "make-believe issue" and an "Obama lie."  The liberal website Media Matters debunked the conservative coverage with typical thoroughness.  Media Matters also found "food experts" to characterize the NYT article as "misleading," which I think was too harsh a description for reporting that seemed basically sound. 

My favorite authoritative statistics about the extent of supermarket deserts put the problem into quantitative perspective, without exaggeration.  The key thing to understand is that most Americans, rich and poor, shop in supermarkets and supercenters.  Likewise, most Americans, rich or poor, shop by automobile.  Supermarkets and supercenters are fundamentally an automobile oriented retail format, and if we pretend that most people walk to the grocery store we will misdiagnose the problem.

USDA's 2009 Report to Congress about supermarket deserts emphasizes statistics showing how many households are far from a supermarket and lack access to a vehicle:
  • 2.3% of U.S. households live more than 1 mile from a supermarket
    and lack vehicle access.
  • 5.7% of U.S. households live more than 0.5 miles from a supermarket
    and lack vehicle access. 
One gets much higher percentages by ignoring vehicle access, but that approach is misleading.  One cannot ignore vehicle access, because, even in low-income areas, most grocery trips are by automobile.  The USDA report finds (in Table 2.9):
  • In low-income areas with high access to food retail, about 65.3% of
    grocery trips are by automobile.
  • In low-income areas with poor access to food retail, about 93.3% of
    grocery trips are by automobile.
Naturally, neighborhoods with adequate retail have a higher concentration of people without cars.  And in rural areas without adequate retail, even most low-income Americans shop by car.

My best summary of the evidence is that perhaps 2 to 6% of U.S. households lack good supermarket access.  Food retail access is a serious concern for people without vehicle access.  Possible remedies to improve local food retail have some merit, but should be carefully targeted based on need, and one should not expect these remedies to carry much of the burden of solving the obesity epidemic for the population as a whole.

Some low-income neighborhoods are supermarket deserts and some are not.


Biomimicry said...

Last November I attended the Food Justice Conference in California. This was my first introduction to the terminology of Food Security, Food Sovereignty, Food Deserts etc. Who was the first to bring these terms into usage and why? So many questions to ask in understanding both sides. Thank you.

Anonymous said...

This is a great level-headed coverage. I'm curious what your take is on studies Investigating possible correlations between supermarket deserts and health outcomes, prices consumers must pay(due to shopping at non-supermarkets) or other posible dependent variables like feelings of social exclusion

usfoodpolicy said...

The federal government started measuring household food security in the 1990s. A large movement promoting community food security incorporated food retail access issues, particularly from the early 2000s onward. The "food desert" terminology is newer -- I first noticed it in UK research in the 2000s -- and covers many of the same issues.

About associations with health outcomes, I have two observations:
1. If the authors of a study say, "we are just looking at associations and did not design our research to address cause and effect," then take that warning to heart. The main results show that tough outcomes run in packs: low-income neighborhoods face all sorts of hardship at higher rates than high-income neighborhoods do.
2. On the other hand, if the authors of a study say, "we used a research design (such as instrumental variables or quasi-experimental research design) that seeks to understand cause and effect," then the reader must check the details to evaluate plausibility.

Basically, my own summary is that the research is highly interesting, but I haven't yet seen research proving that poor retail access makes us unhealthy.