The idea of the “food desert” is fading

Poor neighborhoods, Dr. Lee found, had nearly twice as many fast food restaurants and convenience stores as wealthier ones, and they had more than three times as many corner stores per square mile. But they also had nearly twice as many supermarkets and large-scale grocers per square mile. Her study, financed by the institute, was published in the March issue of Social Science and Medicine.

From another paper:

Dr. Sturm found no relationship between what type of food students said they ate, what they weighed, and the type of food within a mile and a half of their homes.

Here is much more.


I actually work on a team that is investigating this in NYC. We are finding similar impacts as well unfortunately.

Do these studies control for the size of supermarkets? 10 downtown corner stores might still be smaller than the total floor area of one non-urban supermarket.

Shouldn't matter right now. Currently we are just looking for a clean estimate on the average effect on the introduction of the treatment (supermarket). That is probably the most important portion. You can parse out other results later (i.e. the size of the floor area) in subsequent analysis. But the introduction of the treatment would be the primary concern first. No point on rehashing of the initial model is poorly understood.

So no we are focused on one large supermarket in an area where none existed (but Bodegas did).

I'm interested in the results, because a food desert makes so much intuitive sense: recently a big grocery store opened near my neighborhood and I've certainly found myself shopping more often.


+1, what do you mean unfortunately?

It is unfortunate that easy access to supermarkets doesn't reduce obesity because we could use some things to reduce obesity.

Similarly, it is good to know that a drug candidate doesn't work, but don't expect the researchers to be happy about it.

While I do find these results intellectually appealing, they should probably be taken with a few grains of salt. The Sturm paper appears to just run some basic regressions (there's no mention of instruments or endogeneity in the abstract, at least). Do also note that just as correlation does not imply causation, an absence of correlation does not imply an absence of causation.

Project that I work on uses a difference in difference estimation with survey data before the opening of a supermarket in an area with none. People don't change how they eat or behave.

I think the impact of having more fast food and corner stores that sell primarily junk food and soda in poor neighborhoods is negative, not something that can be used to disprove the food desert hypothesis.

Second, it is not very surprising that poorer neighborhoods might have more supermarkets per square mile compared to wealthy areas. Cities and poorer urban areas are more densely populated than the suburbs. In wealthy, primarily suburban, areas the population is more spread out. Sure, the grocery store may be a couple of miles away, but most families have multiple cars and the resources enabling them to buy fresh and healthy food.

Also, urban areas are not homogenous. For example, I think these findings are plausible in the cities I've lived in most recently: Boston and DC. I think they are not plausible in Detroit, where my mother is teacher.

Agreed. Nevertheless I always found price more credible as a contributing factor than unavailability. A meal worth of healthy food is on the average substantially more expensive than unhealthy food. There *are* plenty of foods that are both cheap and healthy, but nevertheless adding more fruit and vegetables and lean meat, and subtracting fat sugar and poorer meat tends to push the price upwards. Which would matter more to a poor person than to a rich person.

The other point worth considering is just how much more, say lettuce, costs in a downtown corner store than a large store in a non-urban setting. Driving barely 30 minutes from rural Illinois to downtown Chicago the price increases on food-items is huge.

What about frozen vegetables? I'd bet the change there is negligible.

We are looking at this too with geocoded data for the area. Again the results are fairly robust to that new supermarkets won't make a real difference, but more analysis must be done. We measure distance in multiple ways. Hopefully we can answer this in the future!

Is Junk Food Really Cheaper?. Mark Bittman seems to calculate the cost of a fairly typical sounding fast food meal at McDonalds as something like twice the cost of a meal with leaner meat and more vegetables.

This gets at a larger point, which is that poverty and obesity are not closely correlated. For black and hispanic men, increases in income are positively correlated with increases in obesity. For white men, there is not much correlation between income and obesity. For women generally, increases in income correlate with decreases in obesity.

I never quite understood why we started believing that poor people were all fat and middle and upper middle class people were not. Except for the super fats, who are much more likely to be poor (and, because of their large backside, very visible), there is not a strong, simple relationship between income and obesity.

His article really seems to underestimate the opportunity cost of time spent cooking. To cook at home you need to spend time traveling to the grocery, shopping, preparing, eating, and cleaning; while to eat fast food, you only need to spend time traveling and eating. For someone working and raising children, the cost of time is going to be very high, and it's got to be a lot easier to take the kids to McDonalds where they can carry on and nobody will much care than to keep them under control at the grocery store and to split your attention between the eggs and the toddler while cooking, etc... He also misses the inventory costs of cooking: dishes, pots, pans, oils, seasonings, etc. That first $13 meal is closer to $40 even if you have all the cookware ahead of time.

Sure, for some people cooking is a habit they need to learn, but for a lot of people, it truly is work that requires a time sacrifice somewhere else, and the cost of that sacrifice might legitimately not be worth it.

Carrots (and most root crops) and Bananas are pretty cheap as are green and beans.

Except that in Detroit, almost everyone (poor families included) live in single-family homes and have access to cars. You plunk yourself down in Detroit with street view and see block after block 'filled' with empty lots (to the point where it often looks like the countryside a mile or two from downtown), but there tend to be cars parked outside the remaining houses.

Are poor Detroiters able to travel outside the city to shop for groceries? Yes they are:

Ubiquitous groceries are only one piece of the larger issue of food access in Detroit. Understanding how vulnerable populations currently utilize our city’s food resources is another essential piece to promoting equitable access and developing effective policy. Using data obtained from the State of Michigan Department of Human Services, we analyzed Electronic Benefit Transfer cards (EBT), also known as Bridge Cards, in order to track where Michiganders have been spending their food assistance dollars. Over $250 million a month was spent across the state through EBT cards between February and July 2011. Nearly $34 million was spent in Detroit, representing 13.7% of statewide spending. This does not mean that Detroit residents received only $34 million in food assistance; rather it means that 13.7% of all EBT dollars were spent within Detroit city limits. In fact, 28.0% of all Detroit households receive food assistance, compared to 9.6% of out-state residents (all of Michigan except Detroit).[8] If Michigan residents spent their EBT money in the cities and townships in which they lived, we would expect to see double the amount of EBT spending in Detroit. It seems that Detroiters are spending a considerable portion of their EBT money outside of the city.

The same article is also interesting in discussing the restaurants in Detroit that have jumped through the hoops to get approved by the state to accept 'bridge cards'. They are predominantly fried chicken outlets -- and then there are the 'you buy, we fry' establishments. It seems pretty clear that the market is routing around bureaucracy to respond to local tastes.

Fried chicken? Oh, dear.

For the record, and contrary to popular belief, Detroit does have grocery stores selling fresh food:

Yes, this article exposes the food desert myth as lazy reporting and yellow journalism unfortunately repeated by Tyler on this blog.

Huh? His post is about how the 'food desert' meme is failing -- he didn't post in support of the idea.

Slocum -- Benny doesn't actually read the blog, he just explains in the comments why Tyler is wrong.

Over one fifth of Detroit households don't even have one car, which for a city that is nowhere near as dense as Boston, New York, Chicago or even Atlanta, is pretty bad: /wiki/List_of_U.S._cities_with_most_households_without_a_car Single family homes has nothing to do with this, by the way. What does have something to do with it is that Detroiters usually have three or more generations in a single home, meaning that the one car for the household doesn't go very far for the family. Most people in Detroit with jobs must drive to work, as the public transit system is totally inadequate and most jobs are in the suburbs.

And then there's this from the article you link: "It seems that Detroiters are spending a considerable portion of their EBT money outside of the city...31% of all Detroit households’ grocery bills are spent outside of the city." Would they be shopping outside the city because it's fun, convenient, just plain adventurous? No, probably because the options in the city are insufficient.


The key takeaway (haha) is not that "poor people have access to fast food and bodegas selling junk food so there is no such thing as a food desert," rather that the high density of places selling food shows that the key barrier to healthier food options is not a lack of disposable income to spend on food, but rather individual preferences and convenience, suggesting that encouraging (or mandating) the presence of vendors of healthier food won't necessarily have much (or any) effect.

Taking into account the opportunity cost of the time it takes to locate, purchase, transport and prepare fresh, healthy food, especially in urban areas, fast food and corner store options are always cheaper.

Ha. I knew it was implausible.

How is density per-square-mile a useful measure? Don't poor neighborhoods (mostly urban) also have more people per-square-mile?

Sure but the theory of a food desert is that the groceries are so far away it's simply impossible to get to them. In reality, in tightly spaced urban areas basically nothing is so far away that you can't get to it.

And people in high density areas don't travel as far. The study uses a 2 mile distance from home, but it seems to me that the urban poor don't venture out 2 miles from home on a regular basis (excepting work).

Maybe you're right, I don't know. But the hypothesis isn't that they aren't willing to travel relatively short distances, it's that the groceries are so far away that it's literally impossible for them to do so.

Maybe the real takeaway is how little people actually value the idea of shopping at a full-service supermarket rather than a junk food-filled corner store. The supermarket may be just an extra 5 or 6 blocks farther away, but it simply isn't worth it to them.

The Sturm study tests for distances from .1 miles to 1.5 miles.

That doesn't really identify the causal effect of whether opening a new supermarket spurs better eating habits though. The supermarket is likely opening due to demand and people in that area now just shop closer to home.

See above.

And for full disclosure, you are right this was a subsidized supermarket opened for this type of area. And having looked at the data you are fairly wrong on the assumption that individuals are shopping closer to home.

Sorry, this was in response to Jonathan above.

Very tangentially related, but interesting, and maybe of interest to readers of Tyler's food stuff:

There was a paper in the BMJ about a decade ago that knocked "food deserts" on the head. They'd established the facts by walking around. Who started the food deserts bunkum anyway? Sociologists?

Culture matters. What else would explain the poor dietary habits of those living in rural areas of the south. Southern food culture among both whites and blacks, especially among low SES overwhelmingly favors sugary foods and fried stuff. Frozen vegetables and chicken are relatively cheap, if you make the effort to buy them. No one consumes mountain dew or purple grape drinks as an alternatives to food.

If you took out whites in the inter-mountain region in the south, white health results would be overwhelmingly better than the rest of the world. Likewise, if you are talking about inner-city health problems, but really mean black health problems, don't forget that the majority of black folks live in the south and are not overwhelmingly concentrated in urban areas - that's just what you see on tv or from where you live.

fwiw, I'm not condemning any of these folks. I've been the outsider many times in my life and thus can find some affinity.

She uses Census tracts as "neighborhoods" which is arguably an insufficient method of determining proximity. Census tracts require a certain number of people and, in some areas of the country, can be (geographically) quite large.

Study that we will be using uses BOTH self reported distance from the supermarket they shop at, and their address in relation to the new supermarket (not the same study as listed above, but finding similar results).

Oh, I don't know -- I think there's actually been some pretty tremendous innovation in that respect. For instance:

Cut your intake of processed sugar and flour. I think Gary Taubes is onto something.

everybody blames sugar =(

what about alcohol? i drink 20+ beers a week. beer lovers drink even more. is there any kind of research on that? alcohol-obesity relationship?

Uh. You do realize beer is full of sugar right? It's pretty much on par or worse than soda ounce per ounce.

What? I'm not sure you understand how yeast works. There is very little sugar in beer.

Good point! Looks like they average about half the carbs of soda. But apparently it depends on the beer.

I believe a can of Coke has 28g of carbs (all sugar, I think).

OTOH I'm not sure if the beer carbs are leftover sugars, or some more complex carbohydrate that might not be as deleterious.

I believe alcohol has zero human-available calories (i.e. your body can't burn it for fuel).

Comments for this post are closed