Wednesday, December 09, 2009

Food stamp participation by race and ethnicity

The New York Times recently posted a terrific web utility for mapping SNAP (food stamp) participation across the counties of the United States.

The federal government's Food and Nutrition Service publishes only state-level SNAP data on the web, not county level. When one maps state-level data, the visual result is too blocky to communicate much information. So, the clever and hard-working New York Times data folks contacted states directly for county-level information.

As a result, one gets a clear image of the landscape of American poverty from county-level data. At a glance, one sees the layout of high rates of SNAP participation across Appalachia, the Mississippi delta and the deep south, the Texas borderlands, and remote rural parts of the West.

Mostly, the outstanding features of the map above reflect the economic landscape rather than state boundaries (although I wonder what to make of the Missouri / Arkansas border, which may reflect either data quality issues or policy differences across states).

An especially interesting feature of the New York Times utility is the ability to map changes in SNAP participation from 2007 to 2009:

Here, the outstanding features are not the overall level of poverty, but instead: (a) the impact of the current financial crisis, including terrible rates of foreclosure in places like Atlanta and Florida, and (b) state boundaries seem more pronounced. For example, notice the Wisconsin border, and the Ohio / Indiana border. Does this mean differences in state SNAP policies have contributed to differences in enrollment recently?

The New York Times utility also permits the user to break down the data by age (children versus adults) and by race and ethnicity. For some purposes, it would be even more interesting to look at the number of SNAP (food stamp) participants relative to all poor people, by race and ethnicity, rather than relative to all people. Such a map, which cannot be created using the New York Times utility, would be more useful if you are interested in discrimination issues in food stamp policy.

A few years ago, Chris Dicken at ERS and I created maps of that sort using California data in a working paper. For example, here is a map of white and black food stamp participants as a fraction of the corresponding poor population.

Similarly, here is the map showing Hispanic food stamp participants as a fraction of the corresponding poor population.

The New York Times mapping utility was produced by Matthew Bloch, Jason DeParle, Matthew Ericson and Robert Gebeloff. DeParle and Gebeloff wrote a related feature article on the SNAP program, which has experienced rapid growth recently. The newspaper also ran a lively debate about SNAP policy with leading bloggers and new media writers.


Anonymous said...

Some of those high proportion counties in the west appear to be those with high Native American populations.

Alissa said...

With regards to southern Missouri: the Missouri bootheel region has extremely high rates of poverty, and Missouri has the highest participation rate in food stamps of any state.

Interestingly, the bootheel also has high rates of obesity and chronic disease.

Unknown said...

This is tangentially related and something readers of the post would be interested in. This morning in NPR, there was a segment on Morning Edition about food insecurity and poverty in migrant farm laborers. You can check it out at:

The irony is cutting, and the story is well produced. Does anyone know, are migrant workers not included in surveys on food insecurity? I suppose logistically they are a difficult group to include in a survey.