Questions on "Why We Should Rethink the SNAP Interview Requirement: A Former Caseworker’s Perspective"

Context

Student and supporting organizations wrote a letter to the USDA recommending that the SNAP application interview requirement be removed, with the accompanying press release here. A news article that summarizes the entire P.R. push for this is accessible here through Fortune magazine.

Why We Should Rethink the SNAP Interview Requirement: A Former Caseworker’s Perspective is an op-ed from The Center for Law and Social Policy (CLASP) which argues SNAP interviews should be made optional. Based on the author’s experience as a caseworker, the mandatory interviews are largely a pretense for detecting fraud, and the time could be better spent providing holistic assistance to SNAP participants. These “unnecessary interviews that are ineffective at finding what infinitesimal fraud exists” only serve to increase the stigma associated with accessing food stamps by treating applicants as if they “are thieves that must be watched”.¹ Unfortunately, the lack of any evidence for or against the effectiveness of SNAP interviews (whatever their purpose may be), the confusing use of statistics, and the wider focus on stigmas associated with SNAP participation resulted in an unconvincing argument. While I agree with the author’s wider points that we should take steps to minimize stigmas of government benefits and that a primary purpose of mandatory SNAP interviews is to verify information, further study of the consequences is warranted before SNAP interviews are made optional.

First, the author confuses low fraud rates as evidence for removing fraud prevention efforts. A key quote from the blog post is: “despite the low incidence of fraud, caseworkers are required to dedicate significant time and resources to fraud prevention".² The author does not consider that the dedication of time and resources to fraud protection is exactly the reason for the low incidence of fraud.

A better argument would use data to demonstrate that SNAP interviews rarely, if ever, detect potential fraud, or explain why interviews are not an effective mechanism for catching fraud. This point is notably excluded from the post. Another point is that the author does not consider that the interviews are also meant to prevent household and agency errors. For example, the applicant may have poor English language skills and misunderstood a question, or a staff member mistyped a value. Validating this in an interview may cut down on these types of errors. But instead the author only considers fraud, as it fits the overall narrative of the essay.

What we can use is the SNAP quality control data to demonstrate that even with the interview, many mistakes still happen. For example, in FY19, 68,228 fair hearings (process to challenge decisions by a State agency which deny, reduce, suspend, or cancel benefits) were held with the State’s decision only upheld 58% of times.³ 29,436 eligibility fraud investigations were referred to hearing officials by State agency investigators.⁴ Unfortunately, I’m not able to tell how many of these resulted in any disqualification, as the data is not broken out by type of fraud. If we just look at administrative disqualification hearings, out of 34,930 hearings, only 1,865 did not find an Intentional Program Violation. However, the ADH data includes both eligibility and trafficking fraud, and it’s very possible that whether or not a violation is found is an inequitable process.

So without direct data, I do think that we need to explore more about how interviews are not ineffective in catching reporting errors (whether or not it is intentional fraud), and identify ways in which we can adjust interviews to be more effective.

Second, the author minimizes the monetary impact of fraud, . From the article:

In fiscal year 2019, only 0.1 percent of SNAP issuances were overpayments based on Intentional Program Violations—which worked out to just a dime for every $100 of SNAP benefits. Moreover, only 0.9 percent were overpayments of any sort, including household and agency errors.

Paragraph 2, Why We Should Rethink the SNAP Interview Requirement: A Former Caseworker’s Perspective; numbers sourced from SNAP “Program Integrity:” How Racialized Fraud Provisions Criminalize Hunger

I am a bit skeptical of percentages presented out of context of real numbers rather than monetary values. If I understand correctly, SNAP issuances reflects the total monetary values of SNAP benefits, e.g., if a household gets $100 of benefits from SNAP, the SNAP issuance is $100. So if we accept the author’s estimate of .1% (.001) in FY19, and we get from the FY19 USDA State Activity Report that total issuances were $55,622,284,536, then we can conclude that $55,622,284.536 (fifty-five million, six-hundred and twenty-two thousand, and two hundred eighty-four dollars and fifty-four cents (rounded-up)) was the amount of overpayments based on Intentional Program Violations. Using the .9% (.009) number, overpayments of any sort would come to just over 500 million dollars. That, to me, is quite a lot of money that was not disbursed correctly.

But I don’t totally understand how we get these estimates. The inline citation links to another report by the author, where the claim is repeated on page 4, and the footnote links references the Supplemental Nutrition Assistance Program State Activity Report: Fiscal Year 2019 (I believe I located the correct report). However, I can’t find any mention of overpayments in that report anywhere.

My best guess is that “overpayments” is synonymous with “fraud”, and as such, the .1% and .9% are arrived at by taking the Fraud Claims Established (in millions) ($59.8), and Total Claims Established (in millions) ($482.4) from Table 40: SNAP Newly Established Recipient Claims - FY 2013 - FY 2019 on page 51 with the denominator as the total issuances mentioned above. This results in .0010 and .00866, which are reasonably close to the author’s estimates.

But the report states that “sum of the fraud associated with disqualifications is a better measure of the ultimate amount of fraud claims than the newly established amount.”⁵ It is a bit hard to be sure, but I believe this is referring to table 41: SNAP Collected Recipient Claims; if this is the case, Fraud Claims Collected (in millions) is $96.6, and Total Claims Collected (in millions) is $376.6, which would suggest .0017 and .0067, respectively.

In the SNAP Quality Control 2019 report, Table 7: Issuance and Issuance In Error, the dollar amount over issued is reported by state. Once again, this is a bit confusing. A U.S. Average is reported that is the same as the sum of all the state-level numbers: $3, 424,700,104. Why don’t we use this number, which compared to total of issuances, would come to 6.16%?

Additionally, the USDA reports a national weighted (by state caseload) average overpayment rate, which for FY2019, was 6.18% (which is relative to the total dollars issue). Based on the SNAP Quality Control Annual Report FY 2019, the U.S. Average estimate for total proportion of active cases with a dollar error was 8.86%.⁶ This USDA Payment Integrity Scorecard shows that overpayment as a percentage of total outlays is $3,306M, or 5.59% in FY19.

Even if I better understood the proper calculations, it is still clear to see that these statistics don’t really give us any evidence for or against the author’s main point that we should end required SNAP interviews. Despite the focus on Intentional Program Violations, at least one legislator questions if the distinction in how the error is made matters. Program integrity doesn’t only mean intentional fraud, but decreasing errors anywhere in the process. Additionally, we can see that in most cases, the agency is likely at fault for overpayments.⁷

Can the interview serve as an opportunity to reduce these errors? Where in the process do these errors occur? Reporting here makes it difficult for me to tell: the only discovery options reported are discovered from case file, household, or collateral contacts.⁸ Most reports will only offer vague points, but don’t address if interviews cut these types of errors down.

Third, the author fails to demonstrate that mandatory SNAP interviews meaningfully limit access or is a major administrative burden.

In this section:

• State agencies should implement better communication and interview processes with applicants, which will streamline the process without removing the interview entirely. The USDA and state agencies recognize this, and we are seeing examples of funding for process modernization projects.

• If the point is to reduce administrative burden, adding holistic support and assistance in place of mandatory interviews will not achieve this.

To get it out of the way, the author incorrectly states that the SNAP program reaches less than a quarter (22%) of eligible people.⁹ The first bullet point of the linked summary that the author uses as a source contradicts the author’s quote; the SNAP program reaches 78% of eligible people (at least in the report year). We should certainly explore why some individuals remain not enrolled, but SNAP is reaching a majority of eligible applicants.

The anecdotes about how interviews may be difficult to attend, or incorrectly scheduled, are unpersuasive: both of these will still occur if the SNAP interview is optional, and these errors result from the states’ implementations. If these scheduling issues are pervasive, then the state needs to investigate, fix, and modernize their own systems. For example, the original Code For America study (outside of the blog post) found that one in three applications in LA were denied for a missed interview. This type of research should be replicated across more states.

While not cited in the blog post, the anecdotes are similar to the concerns raised in the CalFresh study that the original letter to the USDA relies on. Neither the blog post nor the original letter bother to discuss the actual intervention made by Code For America to implement flexible interviews. Here are just two of the intervention’s successes:

Applicants who received information about flexible interviews and who called the flexible interview line received a benefit determination roughly seven days sooner than those in the control group and were 6% more likely to be enrolled in CalFresh (an 11% increase in the approval rate).

The positive effect on approval rates was largest for people whose county office approval rates were the lowest, which means that the flexible interview line makes outcomes more equitable across LA County zip codes.

Think Big, Start Small: How Implementing Flexible Interviews Improves Benefit Delivery

So clearly, there are ways that state agencies can streamline their interview process. As Code For America summarizes:

The success of the flexible interview pilot shows that large county departments can successfully plan, implement, and evaluate major changes to their business processes. They can do this in a way that centers the client experience and optimizes client outcomes while actually improving efficiency for caseworkers and streamlining their business processes.

Think Big, Start Small: How Implementing Flexible Interviews Improves Benefit Delivery

A Center on Budget and Policy Prorities (CBPP) and the author’s own organization CLASP released some excellent recommendations in 2018 to improve flexibility of interviews and completion rates - maybe the USDA can work with states to implement these first.

Outside of the author’s blog post, we can find comments that point blame on interviews as an administrative burden; the Director of Alaska’s Division of Public Assistance, is quoted here in Stateline in Feb. 2024 stating “[Interviews] were taking an extreme amount of time… We were growing the backlog.”¹⁰ That said, the news article initially provides context, pointing out that a backlog was created by a Alaska DHSS policy misinterpretation, a cyberattack, outdated computer systems, budget cuts and low pay. It’s also mentioned that $60 million was approved in funding, “with the bulk of that dedicated to computer upgrades”, and that the department is actively hiring eligibility technicians, which to me suggests that interviews aren’t the only cause or blocker of this backlog.¹¹

This example hints at a problem that the author doesn’t address in their blog post: how the COVID-19 pandemic affected states’ ability to administer SNAP and the backlogs created by the expiration of these temporary benefits. Undoubtedly, proponents of making SNAP interviews at optional will point to all of these waivers and argue that they are proof that the program works without required interviews. But we can’t say that, as data for 2020 or 2021 doesn’t exist while reporting requirements were suspended. And based on 2022 data, 44 states were sent letters pointing out declines in key benchmarks (note that there is some potential that the measured error rates could be less accurate than most years due to emergency allotments, see the State Certification Flexibilities Prioritized SNAP Access section in this CBPP report).

Other opportunities to learn are available. There still are waivers for streamlined interviews for select populations and states face-to-face waiver for quality control interviews; we could study how error rates change when these waivers are in place. Unfortunately, I’m not sure if these findings would be generally applicable to a wider population, but it could be an indicator. We also need to compare the decrease in administrative burden by removing SNAP interviews to the potential increase in resolving cases of fraud or agency and household errors. I don’t mean to assume that fraud will go up if we remove SNAP interviews; it could be that without SNAP interviews as a supplementary data point, the investigations take more work. To what extent are data from the interviews used, and what would change if that data was not verified in the interview? I am not sure, ultimately, but it is worth studying.

As a last note, the blog post claims that removing the interview requirement would “help minimize agency errors”.¹² The author offers no follow-up evidence or explanation to demonstrate why this would be the case, so I won’t bother with it. Instead, I’ll point to some key actions by the USDA’s Food And Nutrition Service that could replace some of the verification done in the interview with automated practices, which is definitely part of a way forward.

FNS is:

Awarding $5 million in grants per year to select state agencies to develop and implement projects that use technology to improve the quality and efficiency of SNAP application and eligibility determination systems; and

Contracting with national payroll data providers to help states improve income verification for SNAP applicants and recipients, which is expected to reduce payment errors and improve the timely processing of applications and recertifications.

Agriculture Secretary Calls on States to Take Action to Improve SNAP Administration for Families in Need

The $5 million in grants offer a competing vision of how the SNAP application process could be streamlined. Five out of the six grant winners of the FY 2023 Process and Technology Improvement Grant from the USDA FNS are funded to modernize and automate state application systems (project descriptions adjusted for brevity):

• The District of Columbia Department of Human Services will launch OCR and a bot to process paper documents and add them to SNAP case files.

• The North Carolina Department of Health and Human Services will develop Remote Identity Proofing (RIDP) to improve access to optional online SNAP services for clients by removing the need to go into offices.

• The South Carolina Department of Social Services will create a system for real-time Social Security Number validation and will provide real-time electronic SNAP case notifications for clients to enhance customer service.

• The Ohio Department of Job and Family Services will implement text messaging for reminders and secure case updates.

• Kansas Department for Children and Families will automate manual SNAP administrative tasks using a bot including case registration, data entry, and creating workflow tasks for caseworkers.

Projects like these should be pursued.

I’m not convinced that SNAP interviews the best opportunity for holistic case management. I don’t have experience as a caseworker, so I only have questions here, which are:

• is the time provided for the interview not enough to confirm the application and help search for alternative resources?

  • Are there automated verification methods outside the interview that we can incorporate into the process?

• if we “must include an optional interview for those who would benefit from it,” how do we identify the applicants that would benefit from it?

  • If those optional interviews occur, where is the harm in confirming an applicant’s information?

• What are NGO or other government opportunities (social service nonprofits, calling a 2-1-1 line) for case management, and why should an optional SNAP interview replicate their efforts?

  • If decreasing administrative burden is a goal, then limiting the interviews to only address SNAP benefits seems more effective than expanding the amount of assistance a caseworker is expected to provide.

Collecting trusted data about SNAP fraud is exactly the evidence that we can use to disprove the stereotypes about public benefits recipients. When politicians try to cut SNAP benefits and create these myths about people not working or stealing from the government, they can easily be challenged with data from a robust and verified data collection process. How will state and national agencies defend this change when many politicians are already reinstating qualifications that preclude participation entirely in many states across the U.S.?

Lastly, I support the author’s intentions to remove stigmas around government benefits and to provide more holistic care to address the challenges of impoverished families. It is important to make sure that people feel respected when going through a process like this - that is the role of excellent case managers. I agree with the author that caseworker jobs should be “secure, well-paid and high quality”.

But I am left unconvinced that streamlining the SNAP application process and ensuring dignity to participants means removing the mandatory SNAP interviews. We should instead first gather more data to determine the effects of the change and explore enhancements to state implementations.

Additional Resources

• Here is a fraud study from the Congressional Research Service last updated in 2018.

• Here is Code for America’s overview page on food assistance programs.

Notes

Relevant quotes and sources from the original blog post are listed below. Last accessed 4/8/2024. The entirety of the article as a PDF is embedded below from 4/8/2024.

why-we-should-rethink-the-snap-interview-requirement-a-former-caseworkers-perspectiveDownload