in Extra Matter

Crime and nourishment Some costs and consequences of the Supplemental Nutrition Assistance Program

Talking to interesting people about interesting food-related things creates its own kind of treadmill. I often continue to read around a topic, which means that the amount of reading around I’m doing just keeps going up. But when I come across something new and interesting, I often think, “I’ve done that. Nobody wants to know more.” Lately, a few people have said I’m wrong, they would like to know more, and that I should write more. So, as I’m travelling, gathering material for future episodes, I’m going to try that. In future, if all goes well, maybe it’ll be in addition to, rather than instead of, an episode.

The Supplemental Nutrition Assistance Program (SNAP) is where the US government spends most of the money it dedicates to food for people who need help. SNAP has two, linked, goals; to reduce food insecurity and to get people to make healthier choices for their diets. It does both, to some extent, but there are also niggling doubts.

For a start, most people who get SNAP benefits have spent their allowance before the month is up. Half of all recipients have used up their SNAP benefits within two weeks of getting them. Furthermore, SNAP benefits represent a huge, regular injection of resources into poor communities. Researchers know that government transfers of cash are associated with crime spikes, probably because some people spend their grants on drugs and liquor or else rob people who haven’t yet spent their benefits. What about SNAP?


While SNAP benefits cannot be used officially to buy drink or drugs, they might free up other income for what have been called “complements to crime” — leisure, illicit drugs and alcohol. Other factors could also be in play. If people know that SNAP benefits arrive early in the month, recipients might be easy targets for crimes. Lean times at the end of the month might prompt people to steal food. Of course, lean times at the end of the month could also potentially reduce crime, if people can’t afford complements to crime. If people getting SNAP might be getting other benefits at the start of the month, maybe shifting SNAP benefits later would help them to smooth their consumption patterns so they don’t need to steal food at the end of the month. It’s clearly complex.

Luckily, although SNAP is a federal program individual states differ in how they distribute SNAP benefits. In Indiana, for example, people get SNAP benefits depending on the initial letter of their last name, so disbursements are more spread out. And on 16 February 2010 Illinois introduced a new policy; instead of two-thirds of SNAP benefits going out on the first of the month, they were shifted to the 4th, 7th and 10th day of the month. These two natural experiments allowed Analisa Packham at Miami University and Jillian Carr at Purdue University to begin to unpick the complexity. (SNAP Benefits and Crime: Evidence from Changing Disbursement Schedules)

Crime statistics in Chicago, Il, have been dropping anyway over time. However, the SNAP policy change has a clear additional effect. Crime and theft at grocery stores are 20-30% lower after the change in policy, equivalent to about 3 grocery store crimes a day across the city, but there is no change in other kinds of crime. There is also evidence that crimes around the 15th of the month — that is two weeks after disbursement, when half of all recipients have exhausted their SNAP payments — also fell after the change in distribution dates. Adding weight to the idea that it is changes in SNAP distribution policy that drive changes in crime rates, the effects are strongest in places that have above average numbers of SNAP recipients and SNAP retailers.

In Indiana, the researchers could match crime and arrest data to the perpetrator’s name and so to the day of the month on which they would have received SNAP benefits. Alas, they had no information on the detailed nature of the crime — grocery store vs other stores, for example — nor on whether the criminal was actually receiving SNAP benefits. Despite that, the analysis shows that crime falls by 4.3% in the 3rd week after SNAP disbursements, but increases in the final week of the benefit cycle. Given the limits of the data, this is probably an underestimate. There is also no evidence of an influence on drug crimes, so this is not just reckless behaviour after possibly receiving SNAP benefits.

What’s going on? One possibility is that recipients have exhausted their SNAP benefits by the 3rd week of the cycle and so don’t have the resources to spend on complements to crime. But they’re not robbing grocery stores either, it seems. There is, however, a tantalising suggestion that women do show a tendency to commit more crimes in the 3rd and 4th week after they might have received SNAP benefits. Perhaps they are providing for children.

So changing the timing of SNAP benefits can actually reduce the level of crime in grocery stories. Another niggling doubt, though, is whether the desire to encourage a healthier diet is thwarted by the higher cost of healthier choices? That was the focus of Eat This Podcast episodes earlier this year, asking Parke Wilde How much does a nutritious diet cost? and talking to Amanda Lee about Australia: where healthier diets are cheaper … but people spend more to eat badly.

After talking to Parke Wilde I concluded, more or less, that you can use SNAP benefits to buy a healthy diet, but you might not want to. Amanda Lee’s research reinforces that view. Healthy choices are less expensive, and also less desirable.


Now along comes the first study that looks directly at the cost of following the USDA’s MyPlate dietary guidelines and relates that to SNAP benefits. Kranti Mulik of the Union of Concerned Scientists and Lindsey Haynes-Maslow at North Carolina State University, looked at the cost of eating the MyPlate guidelines[1] with three different scenarios for satisfying the fruit and vegetable requirements: all fresh, half fresh and half frozen, and one-third each fresh, frozen and canned.[2] That difference, in amount of frozen or canned food, makes no difference to the overall cost of following the MyPlate guidelines for any of the groups they looked at.[3] With a few exceptions, though, most groups had to pay more to eat the MyPlate diet than they were getting in SNAP benefits.

In some cases, the differences were quite large. A family of four with older children would have to spend $626.95 more than they get in SNAP benefits to follow the MyPlate guidelines with all fresh fruit and vegetables. Vegetarian families get off more lightly. They need only an extra $497.07 a month.

Who wins? Senior women, over the age of 51, get $17.27 more in SNAP benefits than they need to follow the all-fresh MyPlate guidelines. Senior vegetarian men get $1.28 more than they need if they eat half of their fruit and vegetables frozen. Children under the age of 7 also generally get more than they need, unless they are being fed a vegetarian diet. That costs their parents about $60 a month more for children aged 2-4 and $26 a month more for children aged 5-7.

Mulik and Haynes-Maslow say they did their study to provide evidence to inform policy decisions. And the Union of Concerned Scientists, for which Mulik works, published a report that says “increased consumption of fruits and vegetables according to MyPlate guidelines could prevent over 127,000 deaths annually from heart disease, and the value of this increased longevity would exceed $11 trillion.”

The US government is currently thinking about cutting both the levels of SNAP support and eligibility. Will they pay any attention to either the affordability of a healthier diet or the impact of SNAP disbursements on crime?



  1. They also factor in the cost of labour in preparation, as 40% of the total cost of food. []
  2. As an aside, the reason for looking separately at fresh, frozen and canned was because “it is likely that processed forms of produce are accompanied by changes in flavor quality, which can influence intake.” No mention of nutrition quality. []
  3. The differences were all in the direction you might expect, but they were not statistically significant. []