Reducing Waste and Fraud in SNAP

Abstract: The Supplemental Nutrition Assistance Program (SNAP), the largest federal nutrition program in the United States, has experienced a troubling rise in improper payments. Overpayment rates climbed from just over 2 percent in 2012 to over 10 percent in 2023—costing taxpayers approximately $10 billion annually. Despite a 350 percent increase in spending on efforts to improve program integrity over the past decade, waste and fraud—particularly from benefit trafficking and eligibility misreporting—remain pervasive. This paper examines the structural causes of these failures, critiques the effectiveness of current fraud-prevention strategies, and proposes a series of reforms to improve accountability and efficiency. By highlighting gaps in current oversight and offering actionable policy recommendations, this paper aims to lay a path toward a more effective and fiscally responsible SNAP program. 

The levels of waste, fraud, and abuse in federal programs have never been higher. Although these types of avoidable inefficiencies have always been too high, they have recently surged with the unusual degree of federal spending brought on by the global pandemic. For over a hundred years, the Government Accountability Office (GAO) has overseen government operations and focused on measuring waste, fraud, and abuse by calculating the number and value of improper payments made to program beneficiaries. Improper payments are those that should not have been made or were made in the incorrect amount. According to GAO, improper payments have consistently been a significant government-wide problem for decades, totaling $2.7 trillion over the past 20 years alone. One trillion dollars of this amount was generated since the recent surge in spending that began in 2020, and many efforts are now underway to understand and remediate the causes. The most common measure of improper payments is the payment error rate, which is generally defined as underpayments plus overpayments as a share of total benefits paid. One federal program with an extremely high, surging payment error rate that has been a concern for many years is the Supplemental Nutrition Assistance Program (SNAP), the federal government’s largest nutritional program.[1] 

By 2023, the pandemic had pushed the national SNAP program error rate to nearly 12 percent, with overpayments amounting to about 10 percent of that figure. At the state level, overpayment rates were as high as 57 percent for Alaska. Even before the increase in the pandemic-era rates, the national overpayment rate had been rising rapidly: While it was just a little over 2 percent in 2012, the national overpayment rate hit 6 percent in 2019, before the pandemic. In dollar terms, as of December 2024, SNAP overpayments are up to $10 billion per year, and much of this amount is from benefit trafficking, which now costs $1.3 billion annually. While efforts are underway to improve the performance of the program, simply spending more on program integrity activities designed to lower the error rate is not enough. For example, spending on improving retailer integrity and reducing trafficking in SNAP benefits rose by 350 percent between 2012 and 2023, yet the rate of overpayment errors continued to surge.

This report offers an overview of SNAP (section 1) and its levels of improper overpayments, including benefits trafficking and other forms of fraud (section 2). High levels of improper payments have existed since data collection on payment errors began in the early 1980s. After falling to a low in 2013, the payment error rate has risen every single year that the data was collected. Federal attempts to control improper payments and promote program integrity have been insufficient (section 3). Unfortunately, the program has not appeared to keep track of all its efforts to improve program integrity. At the end of this report (section 4), I present proposals that could save taxpayers money and improve the delivery of benefits to program recipients. Ultimately, to allow for a data-driven effort to improve the program, the federal government should enhance its ability to evaluate returns on investment by developing measurements to assess the impact of program integrity activities.

1. The Basics of the Supplemental Nutrition Assistance Program

The federal government oversees SNAP, its largest nutrition assistance program, through the Food and Nutrition Service (FNS) of the US Department of Agriculture (USDA). SNAP served 41.7 million household participants per month in fiscal year (FY) 2024 and paid total cash benefits of $93.7 billion (down from $115 billion in FY 2023). See table 1 for a summary of SNAP benefits by household size and figure 1 for the history of program spending and levels of participation. The federal government is responsible for paying 100 percent of the benefits, and states are responsible for administering and monitoring the program. Before the pandemic the federal government paid for about half of the program’s administrative costs incurred by states, and the federal government's share increased significantly during the pandemic. The American Rescue Plan Act of 2021 added an additional $1.1 billion of federal funds to cover state administrative expenses from FY 2021 to FY 2022.[2]

The SNAP budget is determined annually through the congressional appropriations process, but it is authorized every five years in the Farm Bill. The Farm Bill is omnibus, multiyear legislation that governs an array of agricultural and food programs. The most recent bill expired at the end of FY 2023, but it has since been extended, most recently by the extension of the continuing resolution until September 30, 2025 (the American Relief Act of 2025).

The Thrifty Food Plan (TFP) sets the dollar basis for the maximum SNAP benefit allotments. The TFP is maintained by USDA and is based on dietary guidelines, food consumption data, and the cost of food items at retailers. Until the 2018 Farm Bill, the TFP was adjusted only for inflation. The bill mandated a more comprehensive update every five years to reflect broader changes in dietary guidelines, food consumption patterns, and the actual cost of a healthy diet. The first comprehensive update occurred in 2021, increasing SNAP benefits and, as a result, raising the cost of the program by 21 percent.

Nonbenefit costs to the federal government

In addition to contributing to state administrative expenses, the federal government covers other, nonbenefit costs associated with SNAP. First, the FNS reviews the states’ administration of the program in accordance with federal requirements. The federal government also supports state expenses for nutrition education programs, employment and training programs, benefit and retailer redemption and monitoring, payment accuracy, electronic benefits transfer (EBT) systems, program evaluation and modernization, program access, and health and nutrition pilot projects. The share of total federal spending on nonbenefit costs was about $6 billion in FY 2023, or 5 percent of the total costs (see figure 2).[3] SNAP administrative expenses vary quite widely by state.[4] For example, in FY 2016 average administrative expenses varied from $89 per case in Florida to $848 per case in Wyoming. However, there is little understanding of these differences, as this variation does not seem to be explained by economic conditions or caseload levels.

Eligibility

SNAP is broadly available to low-income households, which are generally made up of people who live together and purchase and prepare food together. Eligibility rules and benefit levels are uniformly set at the federal level, though states may tailor some aspects of the program (e.g., the maximum value of a vehicle owned by a qualifying household). Because benefits are means tested, the SNAP participation rate is countercyclical—that is, it rises after recessions—and tends to follow the unemployment and poverty rates (see figure 3).[5] It is considered one of the automatic stabilizers for the economy during a business cycle.[6] SNAP is perhaps second only to the unemployment insurance program in providing assistance during economic downturns. In 2023, 12.6 percent of US residents received SNAP benefits.

In 2022, 88 percent of the 38.3 million eligible individuals each month participated in the program.[7] One way an individual may be eligible is by being a member of a household that meets the following program-specific requirements (though the limits may be adjusted by individual states):

1. A gross monthly income at or below 130 percent of the poverty line. Households with an elderly (60 years old or more) member or member with a disability are exempt from this requirement.
2. A monthly income, net of deductions for items such as housing costs and childcare, less than or equal to the poverty line.
3. Assets below certain limits. In 2022, the general asset limit was $2,500 but higher ($3,750) for households with an elderly (60 years old or more) member or member with a disability.

While recipients can be eligible by meeting program-specific requirements, 93 percent of SNAP-eligible households qualify by meeting the criteria of some other low-income assistance program.[8] This type of eligibility is called categorical eligibility. In this case, all household members must either receive or be eligible to receive cash benefits from Temporary Assistance for Needy Families (TANF), Supplemental Security Income, or a state-financed general assistance program. Established in 1996, categorical eligibility expanded SNAP access to households with a higher income than those eligible for TANF cash assistance under the criteria listed above.

Many jurisdictions took advantage of this expansion by creating broad-based categorical eligibility (BBCE) for households. They began offering a low-cost TANF-funded benefit or service, such as a brochure or access to a telephone hotline, to “convey” SNAP eligibility. This, of course, further increased the number of households eligible for SNAP and raised the federal cost of the program. There are still varying income eligibility thresholds within states that constrain BBCE—namely, households must still meet the second SNAP criteria regarding net monthly income—but there is no asset test, as would be required under the SNAP-specific criteria in most of these jurisdictions.

BBCE has been controversial and costly for SNAP. Currently, nearly 70 percent of SNAP households are eligible because they receive noncash benefits.[9] In 2019, the FNS proposed a rule that would tighten BBCE criteria by requiring that individuals receive more than just the minimal TANF-funded benefits in order to automatically be eligible for SNAP benefits.[10] The FNS argued that making this change would close loopholes that allowed some individuals with higher incomes or assets to qualify for SNAP under lenient state-level eligibility rules and would therefore standardize eligibility criteria across different states. The FNS also estimated that this change would save about $2.5 billion annually. It projected that while 3.1 million people would no longer be eligible for SNAP under the new rule and would lose benefits, states would need to spend more money to process more individual applications and verify income and asset data. Although the rule was finalized at the end of 2020, the Biden administration halted its implementation.

Some households are ineligible for benefits regardless of their income or assets. This includes households with workers on strike, certain students in college, non–US citizens with certain immigration statuses, and, in some states, individuals with drug felony convictions. States also have separate, broad authority to impose work requirements on some adults in SNAP households. Most eligible adults ages 18–49 who do not have minor children are limited to three months of benefits unless they are working at least 20 hours per week or are in a qualifying workfare or job-training program. A state may seek a temporary waiver for the three-month time limit if detailed unemployment data show a sustained level of high unemployment for that state.

In 2020, in response to the pandemic, Congress provided additional SNAP benefits and suspended work and training requirements. Also, states were allowed the flexibility to adjust operations to deal with staffing shortages and the larger number of program applicants. The number of recipients jumped by an average of 4 million (12 percent) per month in FY 2020, rising to an average of 42.2 million per month in FY 2023 (18 percent above FY 2019). The end of some pandemic-related benefit programs has already begun to reduce SNAP enrollment to an average of 41.7 million households and to decrease program outlays by $23 billion over FY 2024. Late in the fiscal year, the Pandemic Electronic Benefits Transfer Program, which provided food benefits to households with children, ended and will likely lower program outlays further.

EBT cards

SNAP benefits are provided through an electronic benefits transfer, or EBT, card. EBT cards function much like other debit cards, and they have been the only way to receive SNAP benefits in all 50 states since mid-2004. SNAP recipients can use their card to purchase most types of foods intended for preparation and consumption at home. Benefits cannot be spent on hot or prepared foods, except under certain conditions, and cannot be spent on tobacco, alcohol, or nonfood items. SNAP accounts are debited to reimburse participating stores for food that was purchased, but the cards lack the built-in protections generally provided by bank-issued debit or credit cards. EBT cards have been credited for the dramatic rise in SNAP enrollment, as average monthly participation rose from about 17 million in 2000 to a high of more than 47 million in 2013 (see figure 1).

2. Improper Payments in SNAP

Federal oversight of SNAP includes estimating improper payments and taking actions to reduce them. Generally, waste and fraud in government programs can be at least crudely measured by what the Government Accountability Office refers to as improper payments. Improper payments are those that were made in an incorrect amount, those that should not have been made because recipients were not eligible, and those that could not be verified as proper because of insufficient documentation. According to the GAO, states made improper SNAP payments mainly because they did not fully verify the recipient’s program eligibility.[11] SNAP measures improper payments by calculating a monthly payment error rate. This rate is generally defined as the underpayments plus overpayments as a share of total benefits paid. SNAP collects error rates monthly for each state by requiring states to have independent reviewers check a sample of cases for the accuracy of eligibility and benefit decisions. Federal officials then recheck a subsample of the cases. These requirements were suspended in 2020 and 2021 because of the pandemic. A payment must exceed a threshold to be counted as improper. The threshold for 2025 is $57, so any payment that is less than or equal to $57 is not counted in the data. The GAO finds that this practice of using a threshold significantly lowers the improper payment rate. For example, the GAO analyzed data from FY 2013 and found that errors below the threshold (and that therefore were not counted) composed 38 percent of SNAP dollars actually paid in error.[12]

Overpayments

High monthly payment error rates in SNAP have been a concern for years and have seen a surge in recent years, beginning with the pandemic in 2020. Overpayments have always dominated the program error rate. In 2023, for example, they outpaced underpayments six to one. Because of this and because overpayments make the program more expensive than it needs to be, this report focuses only on overpayments. States are responsible for recovering overpayments, and that is typically done through monthly deductions from future SNAP benefits. Deductions are higher for overpayments resulting from fraud, and states must assess penalties when this happens. The specific recovery rate will depend on the state’s policies and enforcement efforts, but the rates are very low. For example, in FY2023, SNAP overpayments totaled $10.7 billion, but recovery of overpayments was just $389 million, or less than 4 percent.[13]

The overpayment rate averaged around 3 percent in the early 2010s but jumped to over 5 percent in 2017. Thereafter, it steadily rose until jumping again to around 10 percent during the pandemic. Individual state overpayment rates are now as high 57 percent (Alaska). Despite program efforts to reduce the high error rates, there has been a steady rise in overpayment rates since 2013 (see figure 4). In an attempt to improve the measurement of improper payments, USDA stopped producing the data for two years. It then released numbers using a new methodology in FY 2017. Hence, the FNS attributed the large jump in overpayment rates—from 3.0 percent in FY 2013 to 5.2 percent in FY 2017—to an improved measurement process that more accurately estimates payment errors. The FNS suggested that the loss to waste and abuse was actually much higher than was previously recorded.[14] However, it did not re-estimate the data for any prior years, and a steady rise in error rates occurred until FY 2023 (with the exception of FY 2020 and FY 2021, when the data again were not published). In any event, the realization that payment errors were so much higher than previously thought led to increased interest from policymakers in reducing the rate.[15]

Despite increased efforts to curtail waste, there has been no evidence of anything but increases in overpayments since FY 2013. Furthermore, there is no comprehensive measure of fraud in the program, but taxpayer dollars are lost whether overpayments are fraudulent or just errors. Because of the high cost of the program, these overpayment rates have certainly cost the government tens of billions of dollars (see figure 5).

In another report in 2018,[16]the GAO pointed out that USDA had not addressed all the issues the GAO had raised in its 2016 report. First, estimates of improper payments were not made in 2015 or 2016, and a better methodology had not yet been implemented. Second, USDA had not fully implemented recommendations on combating SNAP recipient fraud by state. And third, USDA had not followed guidance to reduce retailer trafficking in SNAP benefits by better identifying high-risk retailers and by targeting its resources. The GAO also noted the lack of usefulness of USDA’s trafficking estimates.

Fraud in SNAP

There are five main types of inaccuracy and misconduct in SNAP, each has a different effect on the federal budget.[17] SNAP tries to measure different types of fraud differently.

The first type of fraud, trafficking of SNAP benefits, involves their illegal sale. Traffickers may include retailers as well as recipients. Trafficking generally occurs when a recipient sells SNAP benefits, often at a discount, to an owner or employee of a participating store. The Food and Nutrition Act of 2008 mandates trafficking penalties that include fines and imprisonment. According to the most recent USDA estimate, about 2 percent of total SNAP benefits were trafficked in the 2014–17 period, and such trafficking involved about 14 percent of all authorized SNAP stores.[18] This estimate was calculated independently from estimates of SNAP improper payments. In other words, trafficking losses may be in addition to overpayments, or they may account for some of the overpayments. Either way, trafficking amounts were about 40 percent of SNAP overpayments in 2017.[19]

Second, there are problems specifically related to the use of EBT cards. These problems range from system outages that prevent transactions from being processed (an error) to fraudulent activities related to unauthorized access to EBT accounts leading to theft of benefits. EBT card problems may or may not involve benefit trafficking.

Although specific data on the proportion of SNAP payment errors directly attributable to EBT-related issues is not available, it is clear that technical glitches and fraudulent activities involving EBT can contribute to the overall error rate. As previously mentioned, SNAP EBT cards lack the same built-in protections as bank-issued cards and credit cards, and they have proved to be susceptible to at least two types of scams: skimming, where a device is used to steal a card number and PIN, and phishing, where recipients receive fraudulent email messages that appear to be from SNAP employees requesting EBT card information.

EBT scams have been a problem for a long time, and USDA is finally encouraging states to issue safer SNAP EBT cards that include chips, as is the current industry standard for bank-issued cards and credit cards. Disappointingly, no states have yet implemented chips, and only California and Oklahoma have begun transitioning to them. Moreover, SNAP retailers are not yet required to accept chip cards. From 2023 until the end of 2024, states have been allowed to use federal funds to reimburse recipients whose benefits were stolen via card skimming, card cloning, and similar methods. Reimbursements have totaled $150 million and included about 300,000 households.

Scams were also a problem with the P-EBT cards issued in response to the COVID-19 pandemic. Like SNAP EBT cards, P-EBT cards were provided to households with children who would have received free or reduced-price school meals if not for COVID-related school closures or COVID-related reductions in school hours or attendance. The P-EBT card program issued $70.9 billion in benefits from 2020 to 2023.[20]

There is also a multistate problem with EBT cards. Each state can issue its own EBT cards, which can be used in any state. However, individuals may not simultaneously use EBT cards issued by two different states. This rule prohibits “duplicate participation,” which allows the receipt of SNAP benefits in more than one state simultaneously and is considered fraudulent under federal regulations. States may perform background checks of individuals applying for SNAP to check if the applicant is already receiving benefits elsewhere. States regularly exchange information to identify individuals receiving benefits in multiple locations.

A third type of problem is retailer application fraud, which is illegal SNAP participation by an ineligible store. There are no records of how much retailer application fraud occurs, but USDA takes action on dozens of retailers for business integrity violations of one kind or another. If implicated in fraud, stores can be disqualified as a SNAP retailer, face financial penalties, or even face criminal charges. 

Fourth are errors and fraud by households during their application for SNAP benefits. These problems are considered errors if they happen unintentionally but are regarded as fraud if they involve intent to violate program rules. Fraud can vary from falsifying one’s income or identity to selling one’s benefits for cash (trafficking). Penalties range from disqualification, to repayment of illegal benefits, to criminal charges. An individual may fraudulently claim (1) to be unemployed or have a minimum income, (2) to purchase and prepare meals separately, and (3) to have a false identity or false eligibility.

Fifth are errors and fraud by state agencies. Again, if such problems are unintentional, they are considered errors, but some of these problems may result from misconduct in the quality control of a state. There is no current measure that counts recipient errors, recipient fraud, or agency errors.

3. Federal Control over Payment Errors

Federal controls over SNAP payment errors are very weak. While the federal government oversees the program, the states control the specifics of ensuring that accurate benefits are provided to eligible households. States, therefore, have autonomy in addressing and rectifying errors. A portion of the SNAP budget is dedicated to program integrity activities, but there is no summary of how much in total is spent. A range of activities aim to improve payment accuracy, establish quality control, and prevent fraud. Measures include providing training for caseworkers, improving data systems, and implementing new procedures. Fraud, for example, is targeted by grants that pay for state media campaigns to educate recipients about their EBT cards, staff training to detect fraud, and the implementation of new features in EBT technology. This past year SNAP awarded just $5 million in grants but provided no estimate of how much fraud occurs within the program besides the approximately $2 billion in trafficked benefits.

Some federal leverage exists because SNAP measures state-level payment error rates. SNAP is required to calculate a penalty amount when a state error rate exceeds the national payment error rate and meets additional statutory criteria. The liabilities are reduced because state error rates only count errors that exceed a threshold level, as mentioned earlier.[21] The state can pay back the entire penalty to USDA or invest half the penalty in activities focused on improving the administration of the program. The remaining 50 percent is not paid at all unless the state is assessed a liability for three straight years. USDA does not make a penalty rate public, but in FY 2023 Oregon was assessed a $15.7 million penalty for an error rate of 16.7 percent (or $305 million in improper payments) that was above the national average of 11.7 percent. This error rate followed an error rate of 23.0 percent in FY 2022. The state allocated $7.8 million of the penalty toward initiatives aimed at reducing error rates and enhancing SNAP administration.

Recent reactions to the surge in SNAP payment errors

Since 2009, the Office of Management and Budget (OMB) has been required to maintain a website (http://www.paymentaccuracy.gov) that offers a centralized place where the public can report suspected incidents of fraud, waste, and abuse. For the past two years, OMB has also maintained a list of highest-risk programs for high levels of improper payments. SNAP has been on this list for both years and has faced additional reporting requirements intended to describe the root causes of improper payments, outline planned mitigation actions, and track the program’s improvement status.[22] FNS recently set up a National Accuracy Clearinghouse (NAC) designed to prevent SNAP participants from illegally receiving benefits in multiple states. However, although implementation began in early 2024, so far only seven states have joined.[23] Furthermore, in October 2024, USDA announced plans for a series of actions to streamline processing of SNAP applications. These actions included changes in the program’s quality control system, which measures how accurately SNAP state agencies determine a household’s eligibility and benefit amount. In addition, USDA’s FY 2025 budget request asks for an additional $11 billion for the Retailer Integrity and Trafficking program related to SNAP and other nutritional assistance programs—a 30 percent increase in the program’s funding.

Efforts to reduce the steadily rising error rates in SNAP from 2017 until today have had no obvious effect. Furthermore, the pandemic changes in the program were obviously related to the recent surge in overpayment errors to 10 percent. In September 2024, the GAO released yet another report on the program.[24]

In it, the GAO noted that USDA’s Office of Inspector General (OIG) reported that in FY 2023, the FNS was not in compliance with certain criteria in the Payment Integrity Information Act of 2019 for SNAP and other programs. They recommended that the FNS provide information describing the actions the agency would take to come into compliance with annual reporting to OMB. Furthermore, in that report the GAO noted that three of its five recommendations from the earlier 2018 report were still not implemented, including that the FNS improve its trafficking estimates.

Return on investment for program integrity in SNAP

The federal government can use several tools to improve the efficiency of its programs. For example, the government already makes significant use of cost-benefit analysis (CBA) to systematically evaluate proposed policies, projects, and regulations. CBA allows the comparison of different options in considering large-scale projects or regulations with significant economic implications. For example, since the 1970s, federal agencies have been required to consider the costs and benefits of certain regulations that are expected to have large economic effects. A detailed OMB analysis for a new rule is even required for economic impacts exceeding $200 million.[25]

A similar but slightly less demanding tool for government efficiency is to require the calculation of the return on investment (ROI) from certain types of spending or funding for a new project. The GAO and others have recommended its use to evaluate integrity spending in programs such as SNAP. By assessing ROI, agencies can determine the cost-effectiveness of their efforts to prevent fraud, waste, and abuse. The concept of ROI is fairly straightforward. It is usually expressed as a ratio of the gain or loss of an action divided by the cost of making that action. Any spending with an ROI greater than one pays for itself and is therefore worth the investment.

Accurately estimating the ROI of program integrity spending helps improve government efficiency in several ways. First, accurate estimating improves accountability for the agencies involved. They are producing actual evidence of cost savings from program integrity funding. This, of course, works both ways in that an ROI greater than one suggests spending is effective and perhaps even more spending would be justifiable. In contrast, an ROI of less than one would be evidence that the agency’s program integrity spending is not effective and perhaps this funding should be reduced. Second, the calculation of ROIs helps agencies shift resources between different program integrity activities. Program budgets can be better optimized, and programs have an incentive to make their antiwaste and antifraud activities more effective. Third, having available ROI assessments for agency activities provides Congress and other stakeholders the information they need to make informed decisions about the true value of federal programs. Finally, calculating ROIs encourages continuing assessment of the operation of a program and its integrity activities. A declining ROI would possibly signal the need for program changes and enable a program to improve its efficiency.

While many agencies ask for and receive authorization and funding for program integrity activities, few systematically track the effectiveness of those activities. Estimating the ROI for various program integrity activities can be quite complex for many programs, but advancements in data collection and analysis methods have facilitated estimation and made it attainable in most contexts. For this reason, agencies such as USDA have the tools to focus on combating waste and fraud and on evaluating their ROI. In fact, in 2020 GAO pushed SNAP to take a step toward what should eventually enable the program to calculate ROIs for different types of program integrity efforts. GAO made a high-priority recommendation that the FNS identify and analyze the root causes of improper SNAP payments at the state level, identify potential similarities among the states, and develop and implement SNAP corrective actions at the agency level, if appropriate, to help address those problems. The FNS has begun to implement this recommendation and provide states with extensive technical assistance that includes agency-level correctives.[26] Once the FNS has ideas on what corrective actions could be taken, the next step is to develop additional data on the effect of the actions, which will then allow the calculation of ROIs.

SNAP has identified what it considers to be factors affecting its recent issues with payment inaccuracy. However, this information may not significantly educate an ROI effort since the factors do not necessarily help explain the already high overpayment rates that preexisted the pandemic. SNAP cites the increased program flexibilities allowed for states during the pandemic that resulted in changing policies, readjusted operations, and case-processing backlogs.[27] As those temporary actions expire, SNAP will likely learn a lesson about the tradeoffs involved in having allowed those flexibilities. SNAP also cites a problem with high staffing turnover. High turnover generally represents a loss of knowledge and experience, thereby requiring resources to train new staff members. In addition, the program notes that there was a high SNAP application volume that slowed processing of applicants. Generally, however, there is no obvious relationship between the size of the program and the error rates. And lastly, SNAP cites a need to update its data systems and technology. This is always a challenge that should be anticipated for any program.

The GAO has often emphasized in its reports the need for explicit calculation of the ROI for program integrity for various programs. Here are some notable examples:

• Fraud Risk Management: 2018–2022 Data Show Federal Government Loses an Estimated $233 Billion to $521 Billion Annually to Fraud, Based on Various Risk Environments. In this report directed to congressional committees, the GAO highlighted the need for fraud estimation to improve fraud risk management:

Fraud estimation provides opportunities to improve fraud risk management, according to OIG and agency officials. For example, estimates can demonstrate the scope of the problem, improve oversight prioritization, and help determine the return on investment from fraud risk management activities. While it is not possible to eliminate fraud, with a better understanding of the costs, agencies will be better positioned to manage the risk.[28]

• Medicare: Actions Needed to Better Manage Fraud Risks. In its findings in this report, the GAO made recommendations to the Centers for Medicare and Medicaid Services (CMS) in regard to antifraud efforts that included the following:

CMS used return-on-investment and savings estimates to measure the effectiveness of its Medicare program-integrity activities. In developing an antifraud strategy, consistent with the Framework, CMS could include plans for refining and building on existing methods such as return-on-investment, to evaluate the effectiveness of all of its antifraud effort.[29]

• Medicare Integrity Program: CMS Used Increased Funding for New Activities but Could Improve Measurement of Program Effectiveness. The GAO found that the Medicare Integrity Program (MIP) should improve how it measures the ROI of its activities:

To enhance the reliability of data used to calculate the MIP ROI, we recommend that the Administrator of CMS . . . expeditiously complete the implementation of data system changes that will permit CMS to capture accurate MAC [Medicare administrative contractor] spending data, thereby helping to ensure an accurate ROI.[30]

In addition, the report recommends that the administrator “periodically update ROI calculations after contractor expenses have been audited to account for changes in expenditure data reported to CMS and publish a final ROI after data are complete.”[31]

4. Recommendations to Reduce Waste and Fraud in SNAP

Since the SNAP program began in 1939, it has gone through significant changes up through the passage of the Farm Bill of 2014. It has been plagued by significant payment error rates for decades, particular the overpayment of benefits. Attempts to control waste and fraud in the program have consistently failed, as overpayment rates have risen every year since 2014—except in years when error rates weren’t estimated. The federal government is now wasting about $10 billion a year from overpayments, and it is time to pursue significant reforms to the program to enhance its cost efficiency and reduce vulnerability to waste and fraud.

1. The proposed SNAP changes in the draft 2023 Farm bill addressing waste and fraud should be implemented. They are all sound policy proposals:[32]

   a.Create an office of program integrity within the FNS. This office would focus on reducing the outrageously high waste and fraud in SNAP. For example, the office could initially concentrate on the interesting recommendations made by the United Council on Welfare Fraud (UCOWF). The UCOWF’s recommendations include improving recipient integrity through changes in eligibility and oversight; enhancing retailer integrity through changes in the compliance and verification processes; and improving local and state administration with better operations and strengthened fraud prevention mechanisms.[33]

   b. Require the disclosure of payment errors of any size, instead of just those over $57. This requirement would significantly raise the measured error rate in the program and would incentivize states to identify and prevent all payment errors rather than focusing on the larger mistakes.

   c. Allow states to retain more of the funds that they recover when they detect fraud. This change would better incentivize states to find fraud.

   d. Permit states to disenroll retailers that are taking advantage of SNAP. This change would raise the punishment faced by retailers that commit fraud and create payment errors.

   e. Implement the National Accuracy Clearinghouse, which helps prevent duplicative enrollment, in all 50 states. This data-sharing system is designed to help states detect duplicate program participation by individuals across state lines. Implementing the system in all 50 states would increase the identification of those attempting to enroll in SNAP in multiple states at the same time. Duplicate enrollment results in fraud if it is intentional and in payment errors if it is not.

2. Although USDA is finally encouraging states to issue safer SNAP EBT cards with chips, the older-style cards have contributed to fraud in SNAP for a long time. States should be required to join California and Oklahoma by moving to safer SNAP EBT cards with chips, as is currently the industry standard for bank-issued and credit cards.
3. EBT cards issued by one state are generally allowed to be used in another state. This arrangement makes interstate travel less complicated for program recipients. However, according to the FNS, most benefit theft occurs from out-of-state purchases. In November, the FNS announced a pilot program in which SNAP transactions in states where participants are unlikely to be shopping are automatically blocked. Participants can reactivate their cards for out-of-state use when needed.[34] If this pilot does not achieve significant success, the FNS should consider requiring recipients to notify the program of a change of state within a specified period.[35]
4. The FNS can enhance its ability to evaluate ROIs by continuing to improve its data systems and metrics for assessing the impact of program integrity spending on SNAP. Part of this effort could be to identify existing program integrity activities so that the FNS can estimate the amount spent on each activity and develop the data and methodologies to track the activity’s effectiveness. Also, according to the GAO, the FNS has recently made efforts to identify the root causes of improper payments and to develop new tools, guidance, training, and technical assistance to address those problems.[36] The FNS could continue this process, along with developing data collection on the effectiveness of state corrective actions, with the ultimate goal of estimating ROIs for its programs.[37]
5. The FNS could follow the GAO’s recommendation to improve the methodology that it uses to estimate trafficking in SNAP benefits.[38] If 2 percent or more of SNAP benefits are trafficked, more serious efforts are needed to reduce this type of fraud. The FNS could also develop estimates of other types of fraud.
6. The 2020 rule that FNS proposed and finalized could be implemented. This rule tightens up categorical eligibility by requiring more than minimal TANF-funded benefits to automatically achieve eligibility for SNAP benefits.[39]The FNS argued that the proposed rule would increase the efficiency of the program by closing loopholes that allowed some individuals with higher incomes or significant assets to qualify for SNAP because of lenient eligibility rules at the state level.
7. The share of the cost of SNAP allocated between federal and state funds could be reconsidered. Currently, federal funds cover 100 percent of benefits and 50 percent of administrative costs (though much less from FY 2021 to FY 2023). If states covered some of the benefit costs, they would have an increased incentive to enact program reform aimed at reducing overpayment errors.[40]
8. A similar proposal to share the cost of benefits with states was offered by Senator John Boozman (R-AR). This proposal would allow the FNS to require states with persistently high error rates to pay a portion of the benefit costs of SNAP.

Conclusion

Although some of these recommendations to improve the program integrity of SNAP have been made in the past, none have been implemented. The overarching goal of these changes should be to enable the program to estimate the return on investment in current and future integrity efforts in order to reduce both current and future issues of waste and fraud.

About the Author

Keith Hall is a distinguished visiting fellow at the Mercatus Center at George Mason University. He has more than 25 years of public service experience, notably serving as the director of the Congressional Budget Office and commissioner of the Bureau of Labor Statistics.

In addition, he served as the chief economist and director of economics at the US International Trade Commission (USITC), chief economist for the White House’s Council of Economic Advisers, chief economist for the Department of Commerce, and senior international economist for the USITC. He also previously worked as a senior research fellow at Mercatus. He has been a full-time faculty member at the Georgetown University McCourt School of Public Policy and in the economics departments at the Universities of Arkansas and Missouri.

Notes

[1] The Department of Government Efficiency (DOGE), the secretary of agriculture, and the secretary of health and human services have recently suggested reducing some of the items eligible for purchase with SNAP benefits. These suggestions have had no obvious effect on the amount of waste, fraud, and abuse in the program.

[2] See Food and Nutrition Service, “Supplemental Nutrition Assistance Program (SNAP)— Disbursement of SNAP State Administrative Expense Funding Provided by American Rescue Plan of 2021” (policy memo, April 29, 2021).

[3] See National Level Annual Summary data, which can be downloaded from the FNS website at https://www.fns.usda.gov/pd/supplemental-nutrition-assistance-program-s….

[4] Daniel Geller, Borjan Zic, Julia Isaacs, and Breno Braga, Exploring the Causes of State Variation in Supplemental Nutrition Assistance Program (SNAP) Administrative Costs (final report, submitted by Manhattan Strategy Group and Urban Institute to Food and Nutrition Service, June 2019).

[5] Kenneth Hanson and Victor Oliveira, “How Economic Conditions Affect Participation in USDA Nutrition Assistance Programs” (Economic Information Bulletin 100, Economic Research Service, US Department of Agriculture, September 2012).

[6] See, for example, Patrick Canning and Brian Stacy, The Supplemental Nutrition Assistance Program (SNAP) and the Economy: New Estimates of the SNAP Multiplier (Economic Research Report 265, Economic Research Service, US Department of Agriculture, July 2019).

[7] Alma Vigil, “Trends in USDA Supplemental Nutrition Assistance Program (SNAP) Participation Rates: Fiscal Year 2020 and Fiscal Year 2022” (Research Summary, Food and Nutrition Service, US Department of Agriculture, October 2024). 

[8] See Kathryn Cronquist, Characteristics of Supplemental Nutrition Assistance Program Households: Fiscal Year 2019 (Report SNAP-20-CHAR, submitted by Mathematica to Food and Nutrition Service, March 2021). 

[9] See Cronquist, Characteristics of Supplemental Nutrition Assistance Program Households. 

[10] See “Revision of Categorical Eligibility in the Supplemental Nutrition Assistance Program (SNAP),” Federal Register 84, no. 142 (2019): 35570–81.

[11] These improper payments came from failure to verify such basic information such as citizenship, education, employment, finances, household size, identity, and residency. See Government Accountability Office, “Improper Payments: USDA’s Oversight of the Supplemental Nutrition Assistance Program” (GAO-24-107461, September 26, 2024). 

[12] Kay E. Brown, “Supplemental Nutrition Assistance Program: Policy Changes and Calculation Methods Likely Affect Improper Payment Rates, and USDA Is Taking Steps to Help Address Recipient Fraud” (GAO-16-708T, Government Accountability Office, July 6, 2016).

[13] See SNAP State Activity Report FY2023.

[14] In 2016, the GAO reviewed SNAP efforts to measure improper payments and complained that states’ use of program flexibilities and other previous program changes made these measures problematic. In fact, these data quality issues finally led USDA to stop reporting improper payment estimates for 2015 and 2016 until the more accurate methodologies were implemented beginning in 2017. See Brown, “Supplemental Nutrition Assistance Program.” 

[15] See, for example, Randy Alison Aussenberg, “Errors and Fraud in the Supplemental Nutrition Assistance Program (SNAP)” (CRS Report R45147, Congressional Research Service, September 28, 2018).

[16] Kathryn Larin, “Supplemental Nutrition Assistance Program: Observations on Employment and Training Programs and Efforts to Address Program Integrity Issues” (GAO-18-504T, Government Accountability Office, May 9, 2018). 

[17] See Aussenberg, “Errors and Fraud in the Supplemental Nutrion Assistance Program.” 

[18] Hoke Wilson, The Extent of Trafficking in the Supplemental Nutrition Assistance Program: 2015– 2017 (Nutrition Assistance Program Report, submitted by Manhattan Strategy Group to Food and Nutrition Service, September 2021). 

[19] The total SNAP overpayment rate in 2017 was about 5 percent. See Wilson, Extent of Trafficking in the Supplemental Nutrition Assistance Program.

[20] The Consolidated Appropriations Act of 2023 replaced the P-EBT program with a new, permanent program called the Summer Electronic Benefits Transfer Program for Children (Summer EBT). This program began in 2024 and provides households with school-age children eligible for free or reduced-price school meals with benefits to buy groceries during summer months, when most schools are closed. As of June 2024, 45 states, territories, and tribal nations had notified their intent to launch Summer EBT.

[21] See “SNAP Quality Control—Error Tolerance Threshold” on the FNS website at https://www.fns .usda.gov/snap/qc/ett.

[22] See “High-Priority Programs,” on the OMB website at https://www.paymentaccuracy.gov/payment -accuracy-high-priority-programs/. 

[23] See https://www.fns.usda.gov/snap/nac.

[24] Government Accountability Office, “Improper Payments: USDA’s Oversight of the Supplemental Nutrition Assistance Program,” September 26, 2024. 

[25] This analysis is primarily required by Executive Order 12866, which was issued in 1993 and remains in effect.

[26] See, for example, “SNAP Payment Accuracy: Effective Practices, Promising Initiatives, and Helpful Resources,” on the FNS website at https://www.fns.usda.gov/snap/qc/payment-accuracy-tips. 

[27] See SNAP Payment Accuracy: Effective Practices, Promising Initiatives, and Helpful Resources.

[28] Government Accountability Office, “Fraud Risk Management: 2018–2022 Data Show Federal Government Loses an Estimated $233 Billion to $521 Billion Annually to Fraud, Based on Various Risk Environments” (GAO-24-105833, April 2024).

[29] Seto J. Bagdoyan, “Medicare: Actions Needed to Better Manage Fraud Risks” (GAO-18-660T, July 17, 2018).

[30] Government Accountability Office, “Medicare Integrity Program: CMS Used Increased Funding for New Activities but Could Improve Measurement of Program Effectiveness” (GAO-11-592, July 2011), 22.

[31] Government Accountability Office, “Medicare Integrity Program,” 22.

[32] See Leslie Ford and Angela Rachidi, “Missed Opportunities in the Proposed Farm Bill,” AEIdeas, May 22, 2024.

[33] United Council on Welfare Fraud, “Improving SNAP: Program Integrity Initiatives Needed in the 2024 Farm Bill” (white paper, April 2024).

[34] “USDA Expands Safeguards for SNAP Benefits with Pilot Targeting Out-of-State Fraud,” OFA Peer TA, last accessed May 5, 2025, https://peerta.acf.hhs.gov/content/usda-expands-safeguards -snap-benefits-pilot-targeting-out-state-fraud. 

[35] Paige Terryberry, “Congress Should Restore Program Integrity to Food Stamps by Requiring Enrollees to Report Out-Of-State Moves” (white paper, Foundation for Government Accountability, September 21, 2023).

[36] Government Accountability Office, “Improper Payments.” 

[37] See Government Accountability Office, “Payment Integrity: Selected Agencies Should Improve Efforts to Evaluate Effectiveness of Corrective Actions to Reduce Improper Payments” (GAO- 20-336, April 2020). In addition, the GAO has developed resources to help agencies combat improper payments and fraud. See Government Accountability Office, “A Framework for Managing Improper Payments in Emergency Assistance Programs” (GAO-23-105876, July 2023); Government Accountability Office, “A Framework for Managing Fraud Risks in Federal Programs (GAO-15-593SP, July 2015); Government Accountability Office, “The GAO Antifraud Resource” (website), https:// antifraud.gaoinnovations.gov/.

[38] Government Accountability Office, “Improper Payments.”

[39] See “Revision of Categorical Eligibility in the Supplemental Nutrition Assistance Program.” 

[40] See Ford and Rachidi, “Missed Opportunities in the Proposed Farm Bill.”

Suggested Citation

Keith Hall. “Reducing Waste and Fraud in SNAP.” Mercatus Research, Mercatus Center at George Mason University, 2025.

Metadata

© 2025 by Keith Hall and the Mercatus Center at George Mason University
JEL codes: I3, I38
Keywords: SNAP, program integrity, fraud and abuse

Disclaimer

The views expressed in Mercatus Policy Research are the authors’ and do not represent official positions of the Mercatus Center or George Mason University.