SNAP Caseworker Effectiveness
Introduction
Cook and East (2025) evaluate the effect of the Supplemental Nutrition Assistance Program (SNAP) on labor supply using quasi-random assignment of applicants to caseworkers. The authors create an instrument that captures the likelihood that a certain caseworker will accept a random application. This allows them to measure the effect of receiving SNAP on labor supply for those whose application is accepted due to receiving a higher likelihood caseworker.
The authors use data on SNAP applicants from a single mountain-plains state over the period 2011-2016 which includes basic demographic information, when they apply, the caseworker they are assigned to, and SNAP benefit receipt or denial. In addition, the paper links all applicants to the state’s Unemployment Insurance database which allows them to observe earnings and employment. From their instrumental variables model the authors find small and insignificant effects of SNAP on labor supply.
Due to limited caseworker discretion in the decision to approve or deny SNAP applicants, the paper hypothesizes that variation in approval likelihood due to caseworker assignment stems from how much guidance caseworkers are giving applicants in navigating the complex application process. The paper creates an MVPF for a hypothetical policy which led to a one standard deviation increase in the likelihood of caseworkers approving a random application. Given the papers’ hypothesis it is reasonable to think of this as a policy which leads caseworkers to provide more guidance to applicants.
MVPF = 3.3
Net Cost
The paper estimates the net cost of increasing caseworker approval rates as the cost to provide benefits to the additional applicants who are approved, the administrative costs for additional applicants approved, and additional government revenue due to increased labor supply from SNAP receipt. An important assumption the paper makes is that there would be no direct cost to achieve the effect of increasing caseworkers’ approval rate.
The papers’ main estimates indicate the effect of a one unit increase in the caseworker approval rate, so the authors scale the effect by the standard deviation (.03) to get the effect of a one standard deviation change.
A one standard deviation increase in the approval rate will lead to an estimated .03*$183= $20.49 additional of benefits received.
Statistics from USDA indicate that each case has $261 in administrative costs per year or $261/4 = $65.25 per quarter. The paper estimates that a one standard deviation increase will lead to an increase of .03*2.657 = .08 quarters of SNAP benefit receipt. Thus, we can attribute around $65.25*.08 = $5.22 in administrative costs to a one standard deviation increase in the caseworker approval rate.
Finally, the paper incorporates additional tax revenue due to a change in earnings from SNAP receipt. They estimate an effect of a .03*$1,596= $47.88 increase in earnings over three years for those who receive SNAP benefits due to being assigned a caseworker with a higher approval rate.
The authors calculate the average tax rate on this population to be 40% and find the increase in government revenue due to additional earnings is $19.15. The net cost is $20.49 + $5.22 – $19.15 = $6.56.
Willingness to Pay
The paper estimates the willingness to pay as the increase in benefit receipt due to a one standard deviation increase in the approval rate. As shown above, scaling the main estimate by the standard deviation of .03 gets us an estimated $20 of additional benefits received.
MVPF Summary
Thus, increasing the conditional caseworker acceptance rate by one standard deviation has a willingness to pay of 20 and a net cost of 6 for an MVPF of 3.3
References
Cook, Jason and Chloe East (2025). “The Effect of Means-Tested Transfers on Work: Evidence From Quasi-Randomly Assigned Snap Caseworkers” NBER Working Paper 31307. http://www.nber.org/papers/31307.
Policy Features
- Category
- Social Insurance
- Sub-Category
- Nutrition
- Beneficiary Type(s)
- Adults
- Average Age
- 34
- Average Income
- 1583
- Country of Implementation
- United States
- Empirical Method
- IV
- Research Type
- Primary
- Peer Reviewed
- No
- MVPF Publication Link
- www.nber.org/papers/w31307