Economics webloggers such as Tyler Cowen and Mark Thoma are covering Nathan Berg's argument that low-income neighborhoods lack good grocery stores because grocery executives don't happen to know those neighborhoods well.
But, before we move on to diagnosis, can we ask first just how ill the patient is? Some of the best research I know on this topic comes from various USDA surveys of Food Stamp Program participants and low-income nonparticipants. For example, Ohls and colleagues (.pdf) reported in 1999 that one third of low-income respondents shop within a mile of their home and another third within one to four miles. Ninety percent use supermarkets as their major source for food shopping. Almost two thirds reported that their round trip to a grocery store required less than 30 minutes of travel time, and another quarter required between 30 minutes and an hour. In low-income neighborhoods of the East Coast cities I know best, food retail has improved since this report came out.
How can we square the USDA data on the moderately agreeable retail experience of a typical food stamp participant or low-income family with the widespread concern about food deserts in low-income urban neighborhoods? Consider first the low-income urban food desert you know best (I'll hold Anacostia in DC, or eastern Baltimore, or South Bronx in my thoughts). Then, ask yourself what fraction of low income people in the corresponding metro area live there.
For example, here's a map of DC from USDA's Economic Research Service around 2000. Consider not just Anacostia, but also the low-income neighborhoods with good retail. Dark green is highest poverty, and dots are grocery stores. Likewise, fiddle around with this wonderful interactive GIS map from the New York City Coalition Against Hunger last year.
Think both about what particular neighborhoods have the worst retail (it's pretty bad), and also about what fraction of low-income New Yorkers probably have pretty good retail access. I don't recommend writing off the most desolate neighborhoods, but fixing retail in those neighborhoods should be part of broader anti-poverty strategies. Certainly, a low-income New Yorker has other things to worry about besides this!
Good empirical work on the scale of the problem is important for clear thinking about policy remedies. I worry, for example, about tax giveaways to supermarket chains, at least without some hard consideration of other policy options -- including those that give a greater role to healthy food sales from smaller store formats.