The Supplemental Nutrition Assistance Program (SNAP) is a US federal aid program that offers food-purchasing assistance to eligible, low-income individuals and families through monthly food stamp benefits.
Summary
The purpose of this project is to evaluate the impact food policies have on SNAP users’ spending patterns, particularly its effect on their consumption of fruits and vegetables. A simulation model was developed and implemented through the Arena Simulation software. A comparison of several food policies over 10,000 four-person households suggests that healthy advertising and price incentives on fruits & vegetables may most effectively increase the volume of produce purchased.
Background
With over 48 million Americans living in food insecure households in 2014, the accessibility of low-income households to healthy foods is an issue that must be addressed. One such initiative to address this issue has been the United States Department of Agriculture’s (USDA) Supplemental Nutrition Assistance Program (SNAP), formerly known as the Food Stamp Program.
Under SNAP, low-income households may apply and receive monthly financial assistance for food purchases in the form of SNAP benefits. The amount of SNAP benefits received by a household depends on its size, income and expenses, and may be used to pay for food at supermarkets, convenience stores and other food retailers.While SNAP currently helps about 44 million persons per month, it has not fully addressed hunger in the country as 14% of all households still face food insecurity. That is, at least one person in 14% of all households are missing meals and experiencing disruptions in food intake due to insufficient resources for food.
This simulation project aims to evaluate the different food policies under SNAP in terms of their ability to increase participants’ access to healthy foods.
Objective
The goal of this project is to evaluate the effects of food policies on SNAP users food purchases. Past food policies will be used to verify and validate the original model, and new policies will be made to try to increase SNAP users consumption of fruits and vegetables.
Methods
Description of the Model
The monthly food purchase process of a SNAP beneficiary was modeled as follows:
Through SNAP, the USDA distributes a specific amount of aid per month to an accepted food stamp household. The amount of aid given to each household depends on its size, income and expenses. Each month, the beneficiary makes a decision to go to a grocery store or a convenience store to purchase food. After choosing which store type to go to, the beneficiary then makes a decision on what food type to spend his or her aid on as well as the amount of benefits to spend on the chosen food type. This amount is then deducted from the SNAP benefits has remaining for the month. The beneficiary is then able to loop through the decision making processes and purchase more food, until the beneficiary’s monthly aid has been consumed.
Deriving Distributions
- Amount of SNAP benefits:
The amount of benefits a SNAP user receives was modeled with a piecewise function with data taken from the USDA’s annual program report.
A uniform distribution was used to assign an exact amount, in order to maximize the entropy within each benefit interval.
2. Store type: This was modeled using a piecewise function with data taken from a separate report prepared by the Economic Research Services. It was reported that supermarkets accounted for 83% of the dollar value of food stamps redeemed.
3. Food type: Again, a piecewise function was used to model the choice of food type, where the probability of choosing a specific food type is dependent on the types of food normally available at each store type and the percent of the dollar value reportedly spent on that food type.
4. Amount of money spent on a food type: This was approximated using an exponential random variable with an expected value equal to the percent of the food dollar spent on a food type. An exponential was chosen to maximize the entropy of the distribution with a known expected value but unknown bounds.
Potential Policies
Policy 1: Price Incentives on Produce
One of the most widely recognized barriers to healthy eating is the cost barrier. In the past 20 years in the United States, the real cost of fresh fruits and vegetables has risen nearly 40 percent. At the same time, there has been a decrease in the real cost of fats and oils, sugars, sweets, and soft drinks. Together, these changes in real cost make eating poorly an economical option.
In 2011, the Healthy Incentives Pilot (HIP) launched and decreased the cost healthy eating, by providing an additional 30 cents for fruits and vegetables to SNAP recipients. Under this policy, the purchase of fruits and vegetable increased by 20 %. In the simulation, a change in policy lead to a 29% increase in the amount of fruits and vegetables bought.
Policy 2: Mobile Market
Many cities across the nation are turning to Mobile Markets to help answer their nutrition problems. Mobile markets are buses that carry food staples like produce, meat and dairy, eggs, and cereal grains. The goods are sold at grocery store prices in areas that are currently food deserts. They help reduce the amount of people who were before unable to obtain affordable, healthy food options.
Within the simulation, this policy was implemented by increasing the probability of choosing a grocery store from 83% to 95%. This resulted in a 16.93% increase in the amount of produce purchased.
Policy 3: Healthy Advertising and Price Incentives on Produce
Based on the results of the current state model and the results of the above food policies, a new policy idea was created. To capture the increase in produce purchased from the Price Incentives policy, this policy was kept. In addition to this, an advertising campaign should be launched to try and encourage more SNAP users to expand their produce eating habits. This advertising campaign should appeal to users so that they are 10 percent more likely to buy fruits and vegetables than they are in the current SNAP state.
The combination of the changes implemented in this policy led to 112.71% increase in the amount of fruits and vegetables purchased. However, it also lead to a slight increase in the purchase of “other foods”, which consists mostly of junk food.
Limitations
Modeling Human Behavior
It is difficult to fully capture human behavior as certain policies can lead to consequences that can’t be determined without real world testing. For example, discounting produce to incentivize beneficiaries might actually lead them to spend money leftover from buying the same amount of produce on other food types.
Price Fluctuations
The model also does not take into account price fluctuations across stores and geographic locations as the Midwest average of price per pound was used for each food type.
Model Structure
Many policies tied to education and community action could not be tested because it would require building a model that would be difficult to validate. This limited the project to policies that could be tested without changing the structure of the base model.
Results & Conclusion
The final results are as follows:
This simulation suggests that all of the food policies have a positive impact . Based on the data from the models, implementing the Price Incentives on Produce leads to a 29.97% increase of fruits and vegetables, and the Mobile Markets Policy to a 16.93% increase.
The policy with the highest increase in produce purchased is the Healthy Advertising and Price Incentives on Produce Policy. This policy had an increase of 112.71% in fruits and vegetables purchased. Based on the results of the simulation, Policy 3 should be implemented nation-wide to improve the nutrition of SNAP beneficiaries.
The group project was done as part of the Simulation class under the University of Minnesota. Check out the actual final report here.
Nina Domingo is a junior majoring in Industrial & Systems Engineering and one cool bean. If you have any questions regarding the group’s project, feel free to shoot her an email at nggdomingo@gmail.com.
