DISTRICT OF COLUMBIA et al v. U.S. DEPARTMENT OF AGRICULTURE et al

Filing 3

MOTION for Preliminary Injunction by DISTRICT OF COLUMBIA (Attachments: #1 Declaration STEVEN BANKS, #2 Declaration EDWARD BOLEN, #3 Declaration TIKKI BROWN, #4 Declaration CATHERINE BUHRIG, #5 Declaration ALEXIS CARMEN FERNANDEZ, #6 Declaration STEVE H. FISHER, #7 Declaration HOLLY FREISHTAT, #8 Declaration JEFFREY GASKELL, #9 Declaration DEIDRE S. GIFFORD, #10 Declaration HEATHER HARTLINE-GRAFTON, #11 Declaration DANIEL R. HAUN, #12 Declaration KATHLEEN KONOPKA, #13 Declaration ED LAZERE, #14 Declaration BRITTANY MANGINI, #15 Declaration VICTORIA NEGUS, #16 Declaration ELISA NEIRA, #17 Declaration S. DUKE STOREN, #18 Declaration DAWN M. SWEENEY, #19 Declaration LAURA ZEILINGER, #20 Text of Proposed Order)(Konopka, Kathleen). Added MOTION to Stay on 1/24/2020 (znmw).

UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLUMBIA
DISTRICT OF COLUMBIA, STATE
OF NEW YORK, STATE OF
CALIFORNIA, STATE OF
CONNECTICUT, STATE OF
MARYLAND, COMMONWEALTH
OF MASSACHUSETTS, STATE OF
MICHIGAN, STATE OF
MINNESOTA, STATE OF NEV ADA,
STATE OF NEW JERSEY, STATE
OF OREGON, COMMONWEALTH
OF PENNSYLVANIA, STATE OF
RHODE ISLAND, STATE OF
VERMONT, COMMONWEALTH OF
VIRGINIA, and CITY OF NEW
YORK,

Case No. 1:20-cv-00119

Plaintiffs,
V.

U.S. DEPARTMENT OF
AGRICULTURE; GEORGE ERVIN
PERDUE III, in his official capacity as
Secretary of the U.S. Department of
Agriculture, and UNITED STATES
OF AMERICA,
Defendants.
DECLARATION OF EDWARD BOLEN IN SUPPORT OF
PLAINTIFFS' MOTION FOR PRELIMINARY INJUNCTION
I.
I am over the age of eighteen (18) years, competent to testify to the matters
contained herein, and testify based on my personal knowledge and information.
2.
I am a Senior Policy Analyst with the Center on Budget and Policy Priorities
(CBPP). In this position, I focus on state and federal issues in the Supplemental Nutrition
Assistance Program (SNAP), including SNAP Employment & Training (E&T) and waivers for
able-bodied adults without dependents (ABAWDs). I have provided trainings to multiple states
regarding their loss of waivers and have worked with advocates and state officials concerning
issues in implementing ABAWD time limits. CBPP is a nonpartisan research and policy
institute. We work to protect and strengthen programs that reduce poverty and inequality and
increase opportunity for people trying to gain a foothold on the economic ladder. Federal
1

UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLUMBIA
DISTRICT OF COLUMBIA, STATE
OF NEW YORK, STATE OF
CALIFORNIA, STATE OF
CONNECTICUT, STATE OF
MARYLAND, COMMONWEALTH
OF MASSACHUSETTS, ATTORNEY
GENERAL DANA NESSEL ON
BEHALF OF THE PEOPLE OF
MICHIGAN, STATE OF
MINNESOTA, STATE OF NEVADA,
STATE OF NEW JERSEY, STATE
OF OREGON, COMMONWEALTH
OF PENNSYLVANIA, STATE OF
RHODE ISLAND, STATE OF
VERMONT, COMMONWEALTH OF
VIRGINIA, and CITY OF NEW
YORK,

Civ. Action No.

Plaintiffs,
v.
U.S. DEPARTMENT OF
AGRICULTURE; GEORGE ERVIN
PERDUE III, in his official capacity as
Secretary of the U.S. Department of
Agriculture, and UNITED STATES
OF AMERICA,
Defendants.
EXHIBIT A TO DECLARATION OF EDWARD BOLEN IN SUPPORT OF
PLAINTIFFS’ MOTION FOR PRELIMINARY INJUNCTION

1275 First Street NE, Suite 1200
Washington, DC 20002
Tel: [phone]
Fax: [phone]
[e-mail]
www.cbpp.org

April 1, 2019
Ms. Sasha Gersten-Paal, Chief
Certification Policy Branch
Program Development Division
Food and Nutrition Service
3101 Park Center Drive
Alexandria, VA 22302
Re: Proposed Rule: Supplemental Nutrition Assistance Program: Requirements and
Services for Able-Bodied Adults Without Dependents RIN 0584–AE57
Dear Ms. Gersten-Paal:
We are writing to provide comments on USDA’s Notice of Proposed Rule Making (NPRM)
regarding the Supplemental Nutrition Assistance Program’s (SNAP) Requirements and Services for
Able-Bodied Adults Without Dependents. The proposed rule would restrict longstanding state
flexibility to waive areas from SNAP’s three-month time limit as well as limit states’ ability to exempt
certain individuals from the time limit. As a result, USDA estimates that when fully implemented in
a typical month some 755,000 individuals would lose food assistance benefits because they could not
document an average of 80 hours per month of employment or that they qualify for an exemption.
USDA does not provide any evidence to support its assertion that the policy would result in greater
employment or earnings. This is likely because such evidence does not exist. Instead, there is an
extensive body of research that suggests the very likely outcome of the proposed policy is that more
individuals will experience hardship and poverty, including a risk of hunger. Moreover, given
available research on work requirements and the labor market, the proposed policy is very likely to
have even worse outcomes for African Americans, Native Americans, Latinos, and individuals with
disabilities.
The Center on Budget and Policy Priorities is a nonpartisan research and policy institute. We
pursue federal and state policies designed both to reduce poverty and inequality and to restore fiscal
responsibility in equitable and effective ways. We apply our deep expertise in programs and policies
that help low-income people in order to help inform debates and achieve better policy outcomes.
We work to protect and strengthen programs that reduce poverty and inequality and increase
opportunity for people trying to gain a foothold on the economic ladder. Our work on federal
nutrition programs, including SNAP, is a core component of our organization’s work. Our food
assistance analyst team includes nine people, including eight analysts and researchers who work on
SNAP policy and operations. We have deep expertise on SNAP time limit policy including waivers
and individual exemptions. Three members of our team, as well as our organization’s President,
1

have worked on SNAP for more than two decades, including during the time period when the law
governing the time limit was enacted and the current regulations were proposed and codified.
We have deep concerns with the proposed policy and offer extensive comments to support our
strong recommendation that USDA withdraw the NPRM and maintain current policy. In addition
to causing harm to vulnerable individuals who are in between jobs or underemployed, the proposed
policy runs counter to congressional intent. When legislating the time limit policy, Congress
established a waiver authority that allows for states to waive the rule for areas with insufficient jobs
for individuals subject to the rule. Given that individuals who fall into the group subject to the time
limit face extreme difficulty in the labor market, a fact validated by extensive research, the proposed
rule would undercut Congressional intent by setting arbitrary limits unrelated to the purpose of the
waiver.
The proposed rule is also poorly argued, internally inconsistent, and wildly out of sync with
extensive research findings. It offers little, and in some cases, no reasoning or evidence to support
such a dramatic change in a longstanding federal policy that would have significant consequences on
participants, states and other key stakeholders such as retailers and small business. The Department
also provided flawed and contradictory analysis in the NPRM and did not include information
available to the agency that would have informed the rulemaking process. USDA’s rationale for
such a sweeping and harmful change was cursory at best making it almost impossible to comment in
a way that is responsive to its thinking. Because USDA did not make its reasoning transparent or
provide evidence to support its position, we feel obligated to review and provide years of wellknown research and data (some of which USDA funded) that provides evidence counter to USDA’s
proposed policy. We strongly encourage USDA to review these materials as we are concerned the
Department is unaware of the overwhelming evidence that undermines their assertions and poorly
formed conclusions in the proposed rule. This has resulted in lengthy comments in which we
conclude that the best course of action for the proposed policy and under the rulemaking process
would be for USDA to withdraw the NPRM. We strongly urge that course of action.
In this proposed rule, USDA proposed many damaging and ill-advised changes to waivers and
individual exemptions from the three-month time limit. The major changes include:
• Mandating that

areas must have a minimum of a 7 percent average unemployment rate over a
two-year period in order to qualify for a waiver from the time limit;

• Restricting states’

flexibility to define the area they wish to waive;

• Eliminating several

waiver criteria that have been part of program rules for over 20 years,
including a low and declining employment-to-population ratio;

• No longer

allowing states to implement waivers that meet USDA’s criteria while not requiring
that USDA approve waivers in a timely manner;

• Requiring states

to seek their governor’s written consent; and

• Restricting states’

ability to accumulate unused individual exemptions.

Our comments on the proposed regulation fall into several major categories:

2

Proposed Changes to Waiver Criteria
• Chapter 1: Overview of Waivers from the Three-Month Time Limit —Their Purpose and History
• Chapter 2: FNS Waiver Policy Has Been Consistent for the Last 22 Years
• Chapter 3: Setting a Floor for Waivers for Areas With 20% Above National Unemployment

Is Inconsistent with Congressional Intent and Would Be Harmful to Vulnerable Individuals
• Chapter 4: Dropping Several Key Criteria from the Insufficient Jobs Criteria Is Inconsistent with the

Statute
• Chapter 5: Restricting State Flexibility on Grouping Areas Is Counter to Evidence
• Chapter 6: Taking Away Food Benefits from Individuals Who Cannot Document 20 Hours a Week

of Work Will Not Increase Labor Force Participation for This Population
• Chapter 7: Proposed Rule’s Requirement That State Waiver Requests Have the Governor’s

“Endorsement” Violates Congressional Intent
• Chapter 8: Proposed Rule Would Make Implementing Time Limit Harder by Removing

Provisions That Give States Certainty Around Approval

Proposed Changes to Individual Exemptions
• Chapter 9: Eliminating the Carryover of Unused Individual Exemptions Would Cause Hardship and

Exceeds Agency Authority

Problems with the Proposed Rule Process
• Chapter 10: The Proposed Rule Fails to Provide Sufficient Rationale or Supporting Evidence for the

Proposed Policy
• Chapter 11: The Proposed Rule’s “Regulatory Impact Analysis” Highlights FNS’ Faulty

Justification and Includes Numerous Unclear or Flawed Assumptions
• Chapter 12: The Proposed Rule Would Disproportionately Impact Individuals Protected by Civil

Rights Laws, Violating the Food and Nutrition Act’s Civil Rights Protections
• Chapter 13: The Proposed Rule Fails to Adequately Estimate the Impact on Small Entities

Appendix that includes all cited studies and references
Appendix A: CBPP Bios
Appendix B: Materials Cited in Comments

We strongly urge USDA to withdraw the rule and maintain current policy. If you have any
questions regarding our comments, please do not hesitate to contact us.
Sincerely,

Stacy Dean, Vice President
Food Assistance Policy
3

Center on Budget and Policy
Priorities Comments on Rin
0584–AE57:
Supplemental Nutrition Assistance Program:
Requirements and Services for Able-Bodied
Adults Without Dependents

4

Table of Contents
Proposed Changes to Waiver Criteria
• Chapter

1: Overview of Waivers from the Three-Month Time Limit — Their Purpose and

• Chapter

2: FNS Waiver Policy Has Been Consistent for the Last 22 Years

History
• Chapter

3: Setting a Floor for Waivers for Areas With Percent Above National
Unemployment Is Inconsistent with Congressional Intent and Would Be Harmful
to Vulnerable Individuals

• Chapter

4: Dropping Several Key Criteria From Waiver Criteria Is Inconsistent With the

Statute
• Chapter

5: Restricting State Flexibility on Grouping Areas Is Counter to Evidence

• Chapter

6: Taking Away Food Benefits from Individuals Who Cannot Document 20 Hours
a Week of Work Will Not Increase Labor Force Participation for This Population

• Chapter

7: Proposed Rule’s Requirement That State Waiver Requests Have the Governor’s
“Endorsement” Violates Congressional Intent

• Chapter

8: Proposed Rule Would Make Implementing The Time Limit Harder by Removing
Provisions That Give States Certainty Around Approval

Proposed Changes to Individual Exemptions
• Chapter

9: Eliminating the Carryover of Unused Individual Exemptions Would Cause
Hardship and Exceeds Agency Authority

Problems with the Proposed Rule Process
• Chapter

10: The Proposed Rule Fails to Provide Sufficient Rationale or Supporting Evidence
for the Proposed Policy Change

• Chapter

11: The Proposed Rule’s “Regulatory Impact Analysis” Highlights FNS’ Faulty
Justification and Includes Numerous Unclear or Flawed Assumptions

• Chapter

12: The Proposed Rule Would Disproportionately Impact Individuals Protected by
Civil Rights Laws, Violating the Food and Nutrition Act’s Civil Rights Protections

• Chapter

13: The Proposed Rule Fails to Adequately Estimate the Impact on Small Entities

Appendix that includes all cited studies and references
Appendix A: CBPP Bios
Appendix B: Materials Cited in Comments
Note, throughout these comments, we use the terms: Food and Nutrition Service (FNS), U.S.
Department of Agriculture (USDA), and “the Department” somewhat interchangeably. We are not
aware of a particular convention and it is not our intent to suggest difference when we use one term
vs. the other. In addition, when we refer to “state” or “states” we intend to include counties in
their role administering the program in county-administered states.

5

Chapter 1: Overview of Waivers from the Three-Month
Time Limit — Their Purpose and History
The time limit is one of the harshest rules in the Supplemental Nutrition Assistance Program
(SNAP, formerly known as the Food Stamp Program). Childless adults on SNAP are extremely
poor. Like adults with children, childless adults often turn to SNAP for assistance when they are no
longer able to make ends meet, especially as they lose jobs, their hours are cut, or their wages hover
at the federal minimum. While participating in SNAP, their income averages 29 percent of the
poverty line, the equivalent of about $3,400 per year for a single person in 2016.1 The U.S.
Department of Agriculture (USDA), which administers SNAP, has established standards that have
remained consistent over the last 20 years under which states can request a waiver of the time limit
for areas with consistently high unemployment. States request waivers for multiple reasons,
including to ease administrative burden, implement more effective work programs, and exempt
vulnerable individuals who likely will struggle to find work. The proposed rule would severely
weaken this flexibility, increasing administrative burden for states and hardship for SNAP
participants who struggle to find work. This chapter describes the history of these waivers,
Congressional intent and early implementation of waiver rules, and the reasons why states choose to
waive areas in their state.
One of SNAP’s harshest rules limits unemployed individuals aged 18 to 50 not living with
children to three months of SNAP benefits in any 36-month period when they aren’t employed or in
a work or training program for at least 20 hours a week.2 Under the rule, implemented as part of the
1996 welfare law, states are not obligated to offer affected individuals a work or training program
slot, and most do not. SNAP recipients’ benefits are generally cut off after three months
irrespective of whether they are searching diligently for a job or willing to participate in a qualifying
work or job training program. As a result, this rule is, in reality, a time limit on benefits and not a
work requirement, as it is sometimes described.
In addition to being harsh policy that punishes individuals who are willing to work but can’t find a
job, the rule is one of the most administratively complex and error-prone aspects of SNAP law.
Many states also believe that the rule undermines their efforts to design meaningful work
requirements, as the time limit imposes unrealistic dictates on the types of job training that can
qualify. For these reasons, many states and organizations that represent SNAP participants have
long sought the rule’s repeal.
The time limit law does provide states with the ability to seek waivers from USDA to temporarily
suspend the three-month limit for individuals in areas with insufficient jobs. These waivers are the
Steven Carlson, et al., “Who Are the Low-Income Childless Adults Facing the Loss of SNAP in 2016?” Center on
Budget and Policy Priorities, February 8, 2016, http://www.cbpp.org/research/food-assistance/who-are-the-lowincome-childless-adults-facing-the-loss-of-snap-in-2016.
1

For a more comprehensive discussion of the time limit rule, see: Ed Bolen et al., “More Than 500,000 Adults Will Lose
SNAP Benefits in 2016 as Waivers Expire,” Center on Budget and Policy Priorities, updated March 18, 2016,
http://www.cbpp.org/research/food-assistance/more-than-500000-adults-will-lose-snap-benefits-in-2016-as-waiversexpire.
2

6

primary subject of the proposed rulemaking along with states’ authority to use flexible individual
exemptions to exempt individuals of their choosing from the time limit. Since passage of the
welfare law, many states have sought waivers for counties, cities, or reservations with relatively high
and sustained unemployment. Every state except Delaware has sought a waiver at some point since
the time limit’s enactment.3
States can choose (or choose not) to request a waiver. In some cases, states with areas that have a
persistently struggling labor market, such as the Central Valley in California or rural West Virginia,
have sought waivers to avoid penalizing those who cannot find a 20-hour-per-week job within three
months. In other cases, governors have sought waivers because extraordinary events have hurt their
local labor markets, such as the 2010 Gulf of Mexico oil spill, Hurricane Katrina, or layoffs from a
major local employer.
Many states also seek waivers from the time limit because they would prefer to devote the
resources needed to implement the administratively complex time limit to implementing a more
rational and appropriate work requirement tailored to their local economy and to available job
training programs.
FIGURE 1

“FNS Controls Over SNAP Benefits for Able-Bodied Adults Without Dependents,” USDA Office of Inspector
General, Audit Report 27601-0002-31, September 2016, https://www.usda.gov/oig/webdocs/27601-0002-31.pdf.
3

7

USDA’s guidelines regarding waiver criteria, articulated in guidance and regulations, have set clear,
consistent standards for waivers since soon after the statute adopted the time limit and waiver
provisions in 1996. A review of waivers over the last 20 years shows that just over a third of the
country (as measured by the share of the total population living in waived counties) is waived in a
typical year.4 (See Figure 1.)
In the NPRM, USDA states that the current rate of waivers was unforeseen, which is inconsistent
with the historical record that demonstrates that USDA’s original estimate of the extent of waiver
coverage under its rules was in line with current actual coverage. In the NPRM preamble, the
Department states: “The proposed rule addresses these areas of concern and places safeguards to
avoid approving waivers that were not foreseen by Congress and the Department, and to restrict
States from receiving waivers in areas that do not clearly demonstrate a lack of sufficient jobs.”5
This statement stands in contrast to USDA’s own documents. USDA was fully cognizant that its
original proposed waiver policy, which it later codified into final regulations, could result in more
than a third of the country being waived. In an internal summary of waivers from April 23, 1997
entitled, “Time Limit Waivers for Able-bodied Food Stamp Participants,” FNS staff wrote to Office
of Management and Budget staff that “Thirty percent to 45 percent of the able-bodied caseload may
be waived. However, USDA’s best estimate is that the areas that have been waived represent
approximately 35 percent of the able-bodied caseload in the nation as a whole.”6 This was written at
a time of relatively low unemployment and early in the implementation of waivers when take up of
waivers was relatively low. This would suggest that current policy, which has resulted in 36 percent
of the general population living in waived areas except during the Great Recession and its aftermath,
is consistent with what USDA originally intended rather than something that has exceeded its vision.
Moreover, the memo does not suggest any concern with the share of the country waived. And,
these criteria were nearly exact to those codified in final rules.
Under USDA’s proposed rule, however, areas eligible for waivers would be dramatically reduced.
Our organization applied the proposed rule to the areas waived in 2018 and determined that:
• Of the

985 counties (or county equivalents) waived in 2018, 639 counties (65 percent of all
waived counties) in 28 states would have lost their waivers.

• Of the

309 towns located outside of waived counties in 2018, 285 towns (92 percent of all
waived towns) would have lost their waivers, including 259 New England towns.

4

During the recession and its aftermath, Congress made a large portion of the country temporarily eligible for a waiver
in recognition of widespread elevated unemployment. Some have misinterpreted this temporary expansion of waivers as
a permanent expansion of the policy or an Obama Administration-led effort to eliminate the time limit.
5

Supplemental Nutrition Assistance Program: Requirements and Services for Able-Bodied Adults without Dependents,
84 Fed. Reg. § 980 (proposed rule February 1, 2019) found at
https://www.federalregister.gov/documents/2019/02/01/2018-28059/supplemental-nutrition-assistance-programrequirements-for-able-bodied-adults-without-dependents#p-45, hereafter we will refer to this as the “NPRM.”
6

FNS White Paper, “Time Limit Waivers for Able-Bodied Food Stamp Recipients,” April 23, 1997. Faxed from FNS to
OMB analyst Lester Cash on April 25, 1997.

8

• 170

out of the 273 reservations (62 percent of all waived reservations) waived in 2018 would
have lost their waivers.7

Under the proposed policy, we estimate that the share of the U.S. population living in waived
areas would have declined by over 80 percent in 2018, from 36 percent to 6.1 percent of the U.S.
population. The proposed rule would therefore result in a dramatic reduction in states’ ability to
waive areas from the time limit. Unfortunately, that appears to be USDA’s goal rather than
designing and implementing a policy consistent with the statute, i.e., setting waiver criteria and policy
that would allow states to waive areas with insufficient job for individuals subject to the time limit.

A. Current Rules Governing Waivers for Areas With Insufficient Jobs for
Individuals Subject to the Time Limit
The SNAP time limit provision is based in substantial part on an amendment successfully offered
on the House floor on July 18, 1996, by Reps. Robert Ney and John Kasich. When considering the
appropriateness of some of the proposals in the proposed rule, it is illuminating to example the floor
debate to see what Congress did – and did not – think it was requiring.
The floor debate indicates that the amendment’s co-sponsors believed that then food stamp
workfare (participation in which would have exempted an individual from benefit termination) to be
widespread and assumed that large numbers of those who cannot find a private-sector job would be
offered a workfare slot. For example, Rep. Kasich stated on the floor: “….let me be clear what the
amendment does so that there is no confusion. If you are abled-bodied, single, between the ages of
18 and 5-, and you get food stamps, we are saying you have to work …. If you cannot get a job, you
go to a workfare program; 45 out of 50 states have a workfare program.” 8
The sponsors heatedly disputed the statements by opponents of the amendment that the
amendment would cause substantial hardship by denying assistance to people who want to work but
cannot find a job or a workfare slot. And, they emphasized that the amendment contains waivers
and other means to avert such situations. For example:
•

Rep. Ney stated: “… if we read the text, there are hardship exemptions. It can be waived.
There are safeguards in this.”9 Mr. Ney also noted: “… it is an amendment that provides
some safety, it provides a course of a safety net [sic], it has the ability to have waivers from
the State department of human services.”10

7

Based on CBPP internal analysis of unemployment data from the U.S. Bureau of Labor Statistics and the U.S. Census
Bureau. The list of areas is included in Appendix B as “CBPP Summary of Areas That Would Have Lost Their Waivers
form the SNAP Three-Month Time Limit in 2018 if the Proposed Rule Were Implemented in 2018.”
8

142 Cong. Rec. H7905 (daily ed. July 18, 1996). In fact, only about ten states had food stamp workfare programs at
that time, and most such programs were very small. Many of them operated in only a few counties in these states, an
some were only open to families with children. Even today, SNAP workfare is unavailable to a great many people
subject to the time limits.
9

142 Cong. Rec. H7905(daily ed. July 18, 1996).

10

142 Cong. Rec. H7905(daily ed. July 18, 1996).

9

•

Rep. Kasich also addressed this issue. “It is only if you are able bodied, if you are childless,
and you live in an area where you are getting food stamps and there are jobs available, then it
applies. So, if you are able-bodied, you go and you have to work 20 hours to get your
food stamps. The of course if you cannot find a job then you do workfare. That is what
it is. But there are a number of exemptions in here for people who find themselves in
particularly difficult circumstances…”11

As their statements indicate, the amendment’s sponsors visualized the amendment largely as one
under which people were prodded to look for work, were generally provided a workfare slot if a
private sector jobs was not available and would be protected by a waiver if there were insufficient
jobs and workfare slots for them. The sponsors did not see their amendment as one under which
large numbers of individuals who want to work but cannot fund a job end up with neither work nor
food stamps. It should be noted that the sponsors were not cognizant of the extremely limited
number of food stamp workfare slots throughout the country.
In the final legislation Congress established that states could waive areas lacking jobs. USDA has
established criteria to implement that authority that have been consistent for two decades. The rule
was designed to permit states to seek waivers in areas where jobs aren’t available. To qualify for a
waiver, states must provide detailed evidence of high unemployment in local areas, in accordance
with rigorous requirements set by USDA. USDA has consistently used the same criteria to define
high unemployment since the late 1990s.
The federal law gives states the option to request a waiver of the time limit if they can document
that a given geographic area has an insufficient number of jobs (or has an unemployment rate over
10 percent). The standards that define how a state may document “insufficient jobs” were first
outlined in FNS Guidance issued in December 1996.12 In the guidance, USDA offered several
reflections on its understanding of Congressional intent at the time. First, USDA shared its belief
that Congress understood that this group of individuals could find it especially challenging to find
permanent employment and that waivers are intended recognize this problem. “USDA believes that
the law provided authority to waive these provisions in recognition of the challenges that low-skilled
workers may face in finding and keeping permanent employment. In some areas, including parts of
rural America, the number of employed persons and the number of job seekers may be far larger
than the number of vacant jobs. This may be especially so for person with limited skills and
minimal work history.”
In addition, the guidance provided key background on some of the policy that USDA seeks to
restrict in the NPRM. With respect to how states can set or define the area within the state that it
seeks to waive, USDA said, “USDA will give States broad discretion in defining areas that best
reflect the labor market prospects of program participants and administrative needs.”13 The
guidance also recognized that the statute seeks to identify whether or not there are sufficient jobs for
11

142 Cong. Rec. H7905(daily ed. July 18, 1996) (emphasis added).

12

U.S. Department of Agriculture, Food and Nutrition Service (FNS) “Guidance for States Seeking Waivers for Food
Stamp Limits,” FNS guidance to states, December 3, 1996.
13

U.S. Department of Agriculture, Food and Nutrition Service (FNS) “Guidance for States Seeking Waivers for Food
Stamp Limits,” FNS guidance to states, December 3, 1996. Copy included in the appendix.

10

individuals subject to the time limit. “The guidance that follows offers some examples of the types
and sources of data available to States as the consider waiver requests for areas with insufficient
jobs. Because there are not standard data or methods to make the determination of the sufficiency
of jobs, the list that follows is not exhaustive. States may use these data sources as appropriate, or
other data as available, to provide evidence that the necessary conditions exist in the area for which
they intend the waiver to apply. The absence of a particular data source or approach (for example,
data or statistics compiled by a university is not meant to imply that it would not be considered by
USDA if requested by a State.”14
In its original NPRM that covered how USDA would regulate the waiver authority, FNS included
the conceptual framework of the criteria detailed guidance but did not include all of the specifics in
the actual regulation language.15 Commenters, including the Center on Budget and Policy Priorities
comments that USDA should include and codify the details of the guidance into rule in order to
prevent changes in how waiver policy was interpreted and applied, allowing for consistency16. Other
commenters expressed appreciation for the substance of the waiver criteria as articulated in the
guidance and provided for in the NPRM. 17 USDA adopted the suggestion and included the
guidance almost verbatim in the final rule. These criteria were modified only slightly in USDA’s
final regulation waivers based on the experience learned during the waiver application and approval
process (for example, states were allowed to apply to more recent time periods the criteria the Labor
Department uses to identify Labor Surplus Areas in order to determine if an area qualifies for a
waiver). The regulations were proposed by the Clinton Administration and fully codified in
regulations under the Bush Administration in 2001. In setting the waiver criteria, USDA adhered to
longtime Labor Department standards to identify areas with labor-market weakness. To qualify for
the insufficient jobs standard, a state must demonstrate that a geographic area (as defined by the
state) meets specified criteria.
Federal regulations deem waiver requests that are based on certain criteria as “readily approvable”
— meaning USDA approves them once it confirms that the data are correct — because the data
clearly establish high unemployment in the area. (In other words, USDA cannot arbitrarily deny a
state that provides adequate documentation showing that the area’s unemployment rate would
qualify it for a waiver.) These criteria are:
• Designation as

a Labor Surplus Area — a criterion that several federal agencies use to prioritize
government contracts or assistance.18

14

Ibid.

15

64 Fed. Reg. No. 242, page 70920, RIN: 0584-AC39 (proposed rule December 17, 1999.)

16

CBPP Comments on 64 Fed. Reg. No. 242, page 70920, RIN: 0584-AC39, February 17, 2000.

17

Comments submitted by the Greater Upstate Law Project, Center for Civil Justice and the American Public Human
Services Administration on 64 Fed. Reg. No. 242, page 70920, RIN: 0584-AC39, February 17, 2000.
U.S. Department of Labor, “Labor Surplus Area: Frequently Asked Questions,” updated August 21, 2015,
https://www.doleta.gov/programs/lsa_faq.cfm.
18

11

• An average unemployment

rate at least 20 percent above the national average over a recent
24-month time period. This standard tracks the Labor Department’s definition of a Labor
Surplus Area but can use more recent data.

• An average 12-month unemployment

rate over 10 percent.

In addition, waivers based on unemployment rates that meet the criteria to qualify for additional
weeks of Extended Benefits (EB) under the Unemployment Insurance (UI) system may also be
approved by USDA.19 States may also make the case for a waiver for a given area based on certain
other criteria; approval of these waivers is left to the discretion of the Secretary of Agriculture. One
example is a low and declining employment-to-population ratio,20 a measure that labor economists
use to capture weak labor markets in areas where there is a notable lack of jobs relative to the size of
the working-age population. States have used this criterion sparingly, and USDA requires states to
demonstrate additional evidence of weak labor markets for approval, such as a spike in
unemployment or a significant company layoff that affects local labor markets.21 Typically, only a
handful of rural counties and Indian reservations receive waivers under this criterion.
USDA has not issued major policy changes since the criteria were initially published via guidance
in 1996, and state waiver requests have consistently been evaluated according to these criteria. The
agency has provided guidance to states on the specifics of how to do the required calculations and
what information to attach.22

B. Congressional Action to Expand Waivers During the Great Recession
Waiver criteria have been consistent since 1996, with the exception of temporary expansions in
response to the Great Recession. In response to the 2007 recession, Congress took action that had
the effect of temporarily expanding the circumstances under which an area could qualify for a
waiver. Some have mistakenly portrayed these temporary expansions as a permanent expansion of
The EB program has criteria in law under which unemployed workers in a state are eligible to receive extended
unemployment benefits, and states can opt to offer EB benefits under certain additional criteria. (For more information,
see “Conformity Requirements for State UI Laws,” Department of Labor,
https://workforcesecurity.doleta.gov/unemploy/pdf/uilaws_extended.pdf.) Because these unemployment criteria
(known as “triggers”) establish high unemployment, a state is eligible for a waiver if it meets the criteria under the
triggers, even if the state does not elect to provide EB benefits under that trigger.
19

The employment-to-population ratio is the share of the non-institutional, civilian adult population (over age 16) that is
employed. The employment-to-population ratio provides useful information in assessing labor market conditions over
the business cycle because it takes into account changes in labor market “slack” (insufficient jobs) due to changes in
both unemployment and labor-force participation. For more information, see Sarah Donovan, “An Overview of the
Employment-Population Ratio,” Congressional Research Service, May 27, 2015,
https://fas.org/sgp/crs/misc/R44055.pdf.
20

U.S. Department of Agriculture, Food and Nutrition Service (FNS), “Supplemental Nutrition Assistance Program —
Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents (ABAWD),”
December 2, 2016, https://www.fns.usda.gov/sites/default/files/snap/SNAP-Guide-to-Supporting-Requests-toWaive-the-Time-Limit-for-ABAWDs.pdf.
21

22

For example, see: “SNAP - Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without
Dependents (ABAWD)”:
https://fns-prod.azureedge.net/sites/default/files/snap/SNAP-Guide-to-Supporting-Requests-to-Waive-the-TimeLimit-for-ABAWDs.pdf

12

waiver authority. These temporary policies were the only two expansions in waiver criteria since the
time limit took effect in 1996 — and both have ended.
• In recognition of the

Great Recession’s impact on job loss and increased hardship for
unemployed workers, Congress enacted the federal Emergency Unemployment Benefits
(EUC) program in 2008. EUC, like the federal emergency unemployment insurance programs
enacted in every major recession since 1958, was a temporary program that provided
additional weeks of UI to qualifying jobless workers during periods when jobs were hard to
find.23 EUC established several “tiers,” with each tier making a specified number of additional
weeks of UI benefits available to jobless workers in the state, depending on the state’s
unemployment rate. Workers in states with higher unemployment rates would be in higher
tiers and hence could receive more weeks of UI benefits. Because qualifying for higher tiers
of benefits under EUC signified higher unemployment and a lack of jobs, the Bush
Administration allowed states to qualify for a waiver based on qualifying for at least the
second tier of EUC. 24
Congress extended and modified the EUC program several times, allowing it to operate
through January 1, 2014.25 Many states qualified for at least the second tier of EUC through
December 2013. As a result, they qualified for waivers from the time limit into 2015 (since
USDA approved waivers for up to one year from the date a state qualified for EUC).

• Meanwhile,

the 2009 Recovery Act suspended the time limit nationwide for part of 2009 and
all of fiscal year 2010. States had the option to retain the time limit if they offered work
opportunities, such as job training and workfare, to all individuals subject to the rule. During
this time, states didn’t have to request a waiver (though almost every state qualified for a
statewide waiver due to the exceptionally high levels of unemployment across the country).
The suspension of the time limit ended in September 2010. After that, most states continued
to qualify for statewide waivers for a few years under EUC-related and other, longstanding
USDA waiver criteria.

The requirement that states demonstrate to USDA that an area exceeds a high threshold of
persistent unemployment in order to qualify for a waiver has limited the waivers’ scope. A review of
waivers over the last 20 years shows that just over a third of the country (as measured by share of
the total population living in waived counties) has been waived in a typical year.26 Only during the
Chad Stone, “Congress Should Renew Emergency Unemployment Compensation Before the End of the Year,”
Center on Budget and Policy Priorities, November 21, 2013, http://www.cbpp.org/research/congress-should-renewemergency-unemployment-compensation-before-the-end-of-the-year.
23

USDA Memo, “SNAP-ABAWD Statewide Waivers – New Criteria for Unemployment Insurance Extended Benefits
Trigger,” January 8, 2009, https://fnsprod.azureedge.net/sites/default/files/snap/ABAWD%20Statewide%20Waivers.pdf. When all states were eligible for
both the first and second tiers of EUC, USDA required states to be eligible for at least the third tier to qualify for a
waiver.
24

U.S. Department of Labor, “Emergency Unemployment Compensation Expired on January 1, 2014,” updated July 1,
2015, http://ows.doleta.gov/unemploy/supp_act.asp.
25

“SNAP Time Limits: Waivers from the Time Limit Are Back to Historic Norms,” Center on Budget and Policy
Priorities, March 24, 2017, http://www.cbpp.org/sites/default/files/atoms/files/3-24-17fa1.pdf.
26

13

recession and its aftermath was more than half the county temporarily waived from the time limit,
and that was due to widespread elevated unemployment. Some have mistakenly interpreted the
temporary suspension of the time limit in 2009-2010, or the temporary expansion of waivers during
the aftermath of the recession when job growth remained sluggish for some time, as a permanent
expansion of the policy or an Obama Administration-led effort to eliminate the time limit.

C. Why Do States Seek Waivers?
Individual state decisions to seek a time-limit waiver have varied over time depending on states’
leadership and the economic circumstances at the time of their request. Nevertheless, the reasons
remain consistent with those put forward by USDA in their early guidance. USDA’s Office of
Inspector General documented states’ motivation in a recent audit of this policy. 27 Because states
waive the time limit to exempt individuals in areas lacking jobs and to ease administrative burden,
the proposed rule would significantly increase the burden on states and make the time limit less
reflective of areas lacking jobs, as we explain in greater depth later.
• The

time limit provision is very complicated and difficult to administer. State
administrators have expressed strong concern with the complexity of the time-limit provision
since its passage in 1996. The rule requires them to track individuals with a level of specificity
that is inconsistent with how they otherwise operate SNAP and other low-income assistance
programs. States find the rule to be error-prone and believe that it can increase their payment
error rate. Some states seek waivers, in part, to ease the administrative burden associated with
the rule.

• Waiving

the time limit allows states to set a genuine work requirement. Under the time
limit, states are not required to offer a job or training program to every individual (or, for that
matter, to any affected individuals), and they do not receive sufficient funds through the
SNAP Employment and Training (E&T) program to do so. In addition, the law limits the
types of slots a state can provide, making them expensive and out of sync with the needs of
much of this population. As a result, very few states commit to offering work opportunities
to all individuals subject to the time limit.
Waivers, by contrast, can make meaningful work requirements a reality. A state requesting a
waiver of the three-month time limit can still require individuals to engage in work-related
activities as a condition of receiving benefits through the SNAP E&T program. Every state
operates a SNAP E&T program, through which the state can provide a wide range of
employment-related activities to a broad range of individuals who are able to work. While
there is little evidence that SNAP E&T requirements lead to long-term sustainable jobs, they
do allow a state to require a SNAP participant to engage in work activities in order to remain
eligible.
Some states require SNAP participants to participate in a job search program, as a way of
testing an individual’s willingness to work, to remain eligible. These job search programs are
relatively inexpensive to operate. But stand-alone job search is explicitly prohibited from
being a qualifying E&T activity for childless adults subject to the time limit. The only
activities states are allowed to offer to individuals subject to the time limit are job training,

27

“FNS Controls Over SNAP Benefits for Able-Bodied Adults Without Dependents.”

14

education, and workfare programs, which typically are too expensive to offer to all such
individuals.28 Moreover, this population often isn’t a state’s priority for such investments.
In short, if a childless adult searches diligently for work but is unable to find a job or a slot in
a work or training program, he or she loses benefits after three months, despite showing effort
and willingness to work. Waivers, by contrast, allow states to ensure that they are denying
benefits based only on bad conduct, not bad luck.
• States

wish to protect individuals living in relatively high unemployment areas. Even
in states with relatively low statewide unemployment rates, parts of the state may have
significantly weaker labor markets, with few jobs available. The flexibility that allows states to
apply for area waivers recognizes that parts of a state may have insufficient jobs for lowincome workers. For example, some states may seek waivers for areas where a dominant
industry is struggling.
States frequently use waiver authority for rural areas, where about three-quarters of adults say
good jobs are hard to come by where they live.29 Urban areas as a whole have fully recovered
the jobs lost in the recession, while the number of jobs in rural areas continued to remain
below pre-recession levels in 2017.30

D. Current Waiver Authority Is Insufficient to Address Needs of
Unemployed Workers
While a waiver offers a necessary, temporary reprieve from the time limit for individuals living in
areas with high unemployment, both the waiver authority and the underlying time limit are not
responsive to the immediate employment challenges that many people subject to the rule face, even
in areas of more modest unemployment. That, in part, is why USDA’s proposed rule to restrict
states’ ability to seek waivers is so surprising and ill-informed. Geographic waivers provide needed
but inadequate protection for individuals subject to the time limit. While the underlying rule
exempts some individuals from the time limit (such as people with physical or mental conditions and
those caring for incapacitated individuals) and states can exempt a limited number of additional
individuals in unique circumstances, the waiver flexibility allows states the option to fully exempt all
individuals who face insufficient job opportunities for reasons other than area unemployment. 31 As
noted above, USDA indicated in their early guidance on waivers that the unemployment rate can
mask the labor market realities for individuals subject to this rule.
Many of the individuals subject to the time limit struggle to find employment even in normal
economic times. States utilize waivers in recognition of this fact, which also demonstrates why the
Hours spent in job search can count toward an individual’s required 20 hours per week, so long as they constitute less
than half of the total number of hours spent in E&T activities.
28

“The State of American Jobs,” Pew Research Center, October 6, 2016,
http://www.ledevoir.com/documents/pdf/etude_travail_pewresearch.pdf.
29

U.S. Department of Agriculture, “Rural America at a Glance: 2017 Edition,” November 2017,
https://www.ers.usda.gov/webdocs/publications/85740/eib-182.pdf?v=43054.
30

Federal regulations identify certain individuals as exempt (see 7 C.F.R. § 273.24(c)) and states receive a limited number
of individual exemptions they can use to exempt any individual subject to the rule, though these are underutilized in
most states (see 7 C.F.R. § 273.24(g)).
31

15

proposed rule is so harsh. Those subject to this rule are extremely poor, tend to have limited
education, and sometimes face barriers to work such as a criminal justice history or racial
discrimination. While participating in SNAP, childless adults have average incomes of 33 percent of
the poverty line — the equivalent of about $4,000 per year for a single person in 2019. About a
quarter have less than a high school education, and half have at most a high school diploma or
GED.32 SNAP participants subject to the three-month cutoff are more likely than other SNAP
participants to lack basic job skills like reading, writing, and basic mathematics, according to the
Government Accountability Office. 33 As we will discuss in much greater depth, an extensive body
of research shows why these adults likely face much higher unemployment rates than their area’s
unemployment rate and why the proposed rule would severely curtail waivers in areas where these
individuals do not have access to adequate job opportunities.
A much preferable alternative to the USDA’s proposed rule would have been an effort to make it
more possible for states to waive the time limit for more individuals who live in areas with
insufficient jobs for those subject to its eligibility restriction. Restricting this flexibility would be
counter to the intent of the law, inconsistent with more than two decades of practice, and would not
produce the stated outcomes USDA claims its proposal would achieve.

Steven Carlson, Dorothy Rosenbaum, and Brynne Keith-Jennings, “Who Are the Low-Income Childless Adults
Facing the Loss of SNAP in 2016?” Center on Budget and Policy Priorities, February 8, 2016,
http://www.cbpp.org/research/food-assistance/who-are-the-low-income-childless-adults-facing-the-loss-of-snap-in2016.
32

“Food Stamp Employment and Training Program,” United States General Accounting Office (GAO–3-388), March
2003, p. 17.
33

16

Chapter 2: FNS Waiver Policy Has Been Consistent for
the Last 22 Years
A. Current Rules Governing Waivers for Areas with Insufficient Jobs for
Individuals Subject to the Time Limit
Congress established that states could waive areas lacking jobs, and U.S. Department of
Agriculture (USDA) has established criteria that have been consistent for two decades. When the
time limit was being debated in Congress as part of the 1996 welfare law, its proponents claimed that
the proposed rule was not intended to take effect in areas where jobs weren’t available. ThenCongressman and co-author of the provision John Kasich said, “It is only if you are able-bodied, if
you are childless, and if you live in an area where you are getting food stamps and there are jobs
available, then it applies.”34 The rule was designed to permit states to seek waivers in areas where
jobs aren’t available. To qualify for a waiver, states must provide detailed evidence of high
unemployment in local areas, in accordance with rigorous requirements set by USDA. USDA has
consistently used the same criteria to define high unemployment since the late 1990s.
The federal law gives states the option to request a waiver of the time limit if they can document
that a given geographic area has an insufficient number of jobs (or has an unemployment rate over
10 percent). The standards that define how a state may document “insufficient jobs” for individuals
subject to the time limit were first outlined in FNS guidance issued in December 1996.35 In the
guidance, USDA offered several reflections on its understanding of Congressional intent at the time.
First, USDA shared its belief that Congress understood that this group of individuals could find it
especially challenging to find permanent employment and that waivers are intended to recognize this
problem:
USDA believes that the law provided authority to waive these provisions in recognition of
the challenges that low-skilled workers may face in finding and keeping permanent
employment. In some areas, including parts of rural America, the number of employed
persons and the number of job seekers may be far larger than the number of vacant jobs.
This may be especially so for persons with limited skills and minimal work history.
In addition, the guidance provided key background on some of the policy that USDA seeks to
restrict in the NPRM. With respect to how states can set or define the area within the state that it
seeks to waive, USDA said, “USDA will give States broad discretion in defining areas that best
reflect the labor market prospects of program participants and administrative needs.”36 The guidance
also recognized that the statute seeks to identify whether or not there are sufficient jobs for
individuals subject to the time limit:

Congressional Record, 104th Congress, Welfare and Medicaid Reform Act of 1996 (House of Representatives – July
18, 1996), page H7905, https://www.congress.gov/crec/1996/07/18/CREC-1996-07-18.pdf.
34

35

U.S. Department of Agriculture, Food and Nutrition Service (FNS) “Guidance for States Seeking Waivers for Food
Stamp Limits,” FNS guidance to states, December 3, 1996.
36

Ibid.

17

The guidance that follows offers some examples of the types and sources of data available to
States as they consider waiver requests for areas with insufficient jobs. Because there are not
standard data or methods to make the determination of the sufficiency of jobs, the list that
follows is not exhaustive. States may use these data sources as appropriate, or other data as
available, to provide evidence that the necessary conditions exist in the area for which they
intend the waiver to apply. The absence of a particular data source or approach (for example,
data or statistics compiled by a university) is not meant to imply that it would not be
considered by USDA if requested by a State.”37
These criteria were modified only slightly in USDA’s final regulation on waivers based on the
experience learned during the waiver application and approval process (for example states were
allowed to apply to more recent time periods the criteria the Labor Department uses to identify
Labor Surplus Areas in order to determine if an area qualifies for a waiver.) The regulations were
proposed by the Clinton Administration and fully codified in regulations under the Bush
Administration in 2001. In setting the waiver criteria, USDA adhered to long-time Labor
Department standards to identify areas with labor-market weakness. To qualify for the insufficient
jobs standard, a state must demonstrate that a geographic area (as defined by the state) meets
specified criteria.
Federal regulations deem waiver requests that are based on certain criteria as “readily approvable”
— meaning USDA approves them once it confirms that the data are correct — because the data
clearly establish high unemployment in the area. (In other words, USDA cannot arbitrarily deny a
state that provides adequate documentation showing that the area’s unemployment rate would
qualify it for a waiver.) These criteria are:
• Designation as

a Labor Surplus Area — a criterion that several federal agencies use to prioritize
government contracts or assistance.38

• An

average unemployment rate at least 20 percent above the national average over a recent 24-month time
period. This standard tracks the Labor Department’s definition of a Labor Surplus Area but can
use more recent data.

• An

average 12-month unemployment rate over 10 percent.

In addition, waivers based on unemployment rates that meet the criteria to qualify for additional
weeks of Extended Benefits (EB) under the Unemployment Insurance (UI) system may also be
approved by USDA.39 States may also make the case for a waiver for a given area based on certain
37

Ibid.

38

U.S. Department of Labor, “Labor Surplus Area: Frequently Asked Questions,” updated August 21, 2015,
https://www.doleta.gov/programs/lsa_faq.cfm.
39

The EB program has criteria in law under which unemployed workers in a state are eligible to receive extended
unemployment benefits, and states can opt to offer EB benefits under certain additional criteria. (For more information,
see “Conformity Requirements for State UI Laws,” Department of Labor,
https://workforcesecurity.doleta.gov/unemploy/pdf/uilaws_extended.pdf.) Because these unemployment criteria
(known as “triggers”) establish high unemployment, a state is eligible for a waiver if it meets the criteria under the
triggers, even if the state does not elect to provide EB benefits under that trigger.

18

other criteria; approval of these waivers is left to the discretion of the Secretary of Agriculture. One
example is a low and declining employment-to-population ratio,40 a measure that labor economists
use to capture weak labor markets in areas where there is a notable lack of jobs relative to the size of
the working-age population. States have used this criterion sparingly, and USDA requires states to
demonstrate additional evidence of weak labor markets for approval, such as a spike in
unemployment or a significant company layoff that affects local labor markets.41 Typically, only a
handful of rural counties and Indian reservations receive waivers under this criterion.
USDA has not issued major policy changes since the criteria were initially published via guidance
in 1996, and state waiver requests have consistently been evaluated according to these criteria. The
agency has provided guidance to states on the specifics of how to do the required calculations and
what information to attach.42

B. Department Claims to Return to Original Policy Intent and That Current
Waiver Standards are Inconsistent
In the 2019 NPRM, the Department declared its commitment to “implement SNAP as Congress
intended,” 43 implying that waiver policy has diverged significantly from the original policy set in the
1996 welfare reform law. It also claims that the rule will “improve consistency across states,” 44 but
fails to define what the current inconsistency is, why the current standards are causing such
inconsistency, does not provide any evidence to support its claim of inconsistency, or explain why
and how it is a problem. Two possible interpretations of the “inconsistency” claim are that current
waiver standards do not apply consistently to all states, or that the current standards produce
inconsistent waived areas across states. Neither of these claims holds up to scrutiny.
FNS Waiver Criteria Have Not Changed Significantly Since 1996
The Department’s suggestion that waiver policy has deviated from Congressional intent suggests
that either the Department now knows something that it did not 22 years ago when it put forward
guidance to implement the law or that the final regulations deviated from the original guidance set in
December 1996.
The employment-to-population ratio is the share of the non-institutional, civilian adult population (over age 16) that is
employed. The employment-to-population ratio provides useful information in assessing labor market conditions over
the business cycle because it takes into account changes in labor market “slack” (insufficient jobs) due to changes in
both unemployment and labor-force participation. For more information, see Sarah Donovan, “An Overview of the
Employment-Population Ratio,” Congressional Research Service, May 27, 2015,
https://fas.org/sgp/crs/misc/R44055.pdf.
40

41

U.S. Department of Agriculture, Food and Nutrition Service (FNS), “Supplemental Nutrition Assistance Program —
Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents (ABAWD),”
December 2, 2016, https://www.fns.usda.gov/sites/default/files/snap/SNAP-Guide-to-Supporting-Requests-toWaive-the-Time-Limit-for-ABAWDs.pdf.
42

For example, see: U.S. Department of Agriculture, Food and Nutrition Service (FNS), “Supplemental Nutrition
Assistance Program — Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without
Dependents (ABAWD),” December 2, 2016, https://www.fns.usda.gov/sites/default/files/snap/SNAP-Guide-toSupporting-Requests-to-Waive-the-Time-Limit-for-ABAWDs.pdf.
43

2019 NPRM, p. 8.

44

2019 NPRM, p. 16.

19

On the first count, the Department provided no information or evidence from legislative history
that would suggest that its knowledge or understanding of Congressional intent has improved since
it issued its first guidance on waiver policy just a few short months after the welfare law passed. In
fact, the NPRM does not provide any reference to legislative history to help reviewers understand
why current policy is out of sync with the goal of the statute. It is impossible to respond to the
Department’s reasoning other than to provide the available legislative history as we have in Chapter
1 (Overview of Waivers From the Three-Month Time Limit — Their Purpose and History) which
explains how legislative history runs counter to the Department’s assertions.
Similarly, we observe no significant policy shift in the waiver policy that the Department originally
set forth in its December 1996 guidance from current policy. In fact, comparing waiver standards
from 1996 to the current standards can provide insight into how much waiver policy has
significantly changed over the past two decades. The best evidence for this comes from FNS’ 1996
guidance, which describes in detail the waiver criteria that were available to states at the time. Table
2.1 below compares the key waiver criteria included in FNS’ December 1996 guidance to the current
criteria described in FNS’ December 2016 guidance (which is the most recent articulation of the
rules set forth in the 2001 federal regulations).
TABLE 2.1

FNS Waiver Policy Has Been Consistent Since 1996
December 1996
FNS Guidance45

January 2001
Final Rule46

December 2016
FNS Guidance47

Waiver Eligibility Criteria
Labor Surplus Area
Designation (LSA)
LSA- Like: 24-month
average unemployment
rate 20 percent above the
national average using
more current data than LSA
Qualification for Extended
Unemployment Benefits
12-month average
unemployment rate over 10
percent

Yes

Yes

Yes

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

45

U.S. Department of Agriculture, Food and Nutrition Service (FNS) “Guidance for States Seeking Waivers for Food
Stamp Limits,” FNS guidance to states, December 3, 1996.
46

66 Fed. Reg., No. 11, 4438 (January 17, 2001).

47

“SNAP - Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents
(ABAWD),” FNS guidance issued December 2, 2016, https://fns-prod.azureedge.net/sites/default/files/snap/SNAPGuide-to-Supporting-Requests-to-Waive-the-Time-Limit-for-ABAWDs.pdf.

20

TABLE 2.1

FNS Waiver Policy Has Been Consistent Since 1996
December 1996
FNS Guidance45

January 2001
Final Rule46

December 2016
FNS Guidance47

Waiver Eligibility Criteria
3-month average
unemployment rate over 10
percent
Historical seasonal
unemployment rate over 10
percent
Employment-to-Population
Ratios
Demonstration of lack of
jobs in declining
occupations or industries
Demonstration of lack of
jobs in an area

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Other Waiver-Eligibility Policy
Combining data for
geographic and economic
regions
Estimating unemployment
rates for tribal lands
Requesting two-year
waivers

Yes

Yes

Yes

Yes

Yes

Yes

No

No

Yes

Table 2.1 demonstrates that the waiver criteria set in the 1996 welfare reform law have remained
remarkably consistent over the past two decades. For example, FNS’ 1996 guidance indicated that
high unemployment areas can be waived by being designated as Labor Surplus Areas (LSA),
qualifying for extended unemployment benefits, or having average unemployment rates of over 10
percent. These are the same criteria described in current FNS guidance. Moreover, criteria that are
seldom used by states, such as demonstrating historical seasonal unemployment or a lack of jobs in
declining occupations are described in the 1996 guidance and remain the same today. The
meaningful change was to allow states to use more recent unemployment data when considering
whether an area met the LSA criteria of having average unemployment rates at least 20 percent
above the national average for a recent 24-month period. This variation of the LSA criteria also
permits areas to qualify with 24-month average unemployment rates below six percent. This
criterion is informally known as “LSA-like.” Using more recent unemployment data allows for a
more current assessment of the unemployment situation of an area and is an enhancement of the
LSA criteria, not a significant change. This was added in the early 2000s and is codified in current
federal regulations.48
48

66 Fed. Reg., No. 11, 4438 (January 17, 2001).

21

Similarly, the 1996 guidance included other waiver policies such as the ability to combine data and
estimating unemployment rates for tribal lands, urging states to “consider areas within, or
combinations of counties, cities and towns” and to “consider the particular needs of rural areas and
Indian reservations.”49 These policies remain in place in current guidance, with small changes made
over the years.
The small changes that have occurred are largely refinements of the original criteria, not major
additions to waiver policy. For example, FNS guidance issued in December 2004 revised the method
for calculating average unemployment rates over 24-month periods.50 Current FNS guidance also
provides specific instructions on how to round 24-month average unemployment rates, and a
standard methodology for estimating unemployment rates for Native American reservations.51 FNS
also offered states “the option of two-year waiver approvals” in a February 2006 memorandum;
while this was an addition to waive policy at the time, it was not a major revision of waiver standards
— the criteria for two-year waivers are more restrictive than those for shorter waivers.52 (See
Chapter 8 for more.)
The final rule published in January 2001 offers clear evidence that the Department at the time
intended to codify the waiver policies from its 1996 guidance, so that they would become a
consistent set of rules that states use to determine their waiver eligibility in the future. In the final
rule, the Department discussed the comments issued in response to its NPRM on the waiver policy,
and why this influenced its codification of waiver criteria. It acknowledged that it did not include the
1996 guidance in its initial regulations, not because it deviated from the Department’s intent, but
because “[the guidance] was extensive and detailed.”53 The Department also explained that it
“received several comments suggesting [the Department] include all or some of the guidance in the
regulations. Commenters argued that unless the guidance is incorporated into the regulations, a
subsequent administration could abolish it without public comment. Based on these comments, [it]
decided to incorporate some of the more pertinent aspects of the guidance into the regulation. More
specifically, [it] modified the regulations at 7 C.F.R. 273.24(f) to include a non-exhaustive list of the
kinds of information a State agency may submit to support a claim of 10 percent unemployment or
‘lack of sufficient jobs.’”54 The final rule goes on to list the same waiver eligibility criteria described
in Table 2.1 as part of the December 2016 guidance, and shows that Department recognized at the
49

U.S. Department of Agriculture, Food and Nutrition Service (FNS) “Guidance for States Seeking Waivers for Food
Stamp Limits,” FNS guidance to states, December 3, 1996.
50

“ABAWD Waivers – New Method for Calculating Average Unemployment Rates,” FNS Northeastern Region Food
Stamp Regional Letter, April 28, 2004.
51

U.S. Department of Agriculture, Food and Nutrition Service (FNS), “Supplemental Nutrition Assistance Program —
Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents (ABAWD),”
December 2, 2016, https://www.fns.usda.gov/sites/default/files/snap/SNAP-Guide-to-Supporting-Requests-toWaive-the-Time-Limit-for-ABAWDs.pdf.
52

“FSP – 2-Year Approval of Waivers of the Work Requirements for ABAWDs under 7 CFR 273.24,” FNS
memorandum issued February 3, 2006.
53

66 Fed. Reg., No. 11, 4438 (January 17, 2001).

54

Ibid.

22

time that a consistent and predictable waiver policy would be an essential asset to states in the
future.
This evidence demonstrates that current waiver criteria are not wildly out of step with the original
intent of waiver policy at its inception. The original guidance set the flexibility that states currently
have in waiving areas, contrary to the Department’s claim in the proposed rule that they use their
flexibility “in a way that was not likely foreseen.”55
Furthermore, the consistency in waiver policy over the decades has been important for states,
which have relied on it for 20 years. The Department’s claim that its proposed rule will allow “States
to reasonably anticipate whether it would receive approval” ignores the reality that current waiver
policy already accomplishes this goal. In reality, the rule would make it harder for states to obtain
waivers and would disrupt their long-standing waiver implementation procedures.
The Proposed Rule Does Not Provide Evidence of Inconsistency
in Current Waiver Standards
As noted earlier, the Department does not explain or justify in the rule its implication that current
waiver standards are inconsistent, and reasonable interpretations of what it meant do not hold up to
scrutiny. For example, the Department may have meant that there is not a consistent set of waiver
standards that apply to all states. This is not the case, as waiver standards apply uniformly to all
states. States might use different criteria to show their eligibility for waivers; for example, a state with
unemployment well above 10 percent might request a waiver based on the 10 percent threshold,
whereas another state with rapidly rising unemployment might request a waiver based on qualifying
for extended unemployment benefits. The fact that states use different criteria reflects differences in
their demographic composition and economies, among other factors. It does not mean, as the
Department might be implying, that states do not have the option of using any of the criteria to
show their waiver eligibility, particularly as their local economic conditions change over time.
Over the past two decades, FNS has regularly updated its guidance to states to inform them of
their options as the economy changed. One of the strengths of the current rules and USDA’s
application of them is the extraordinary consistency with which USDA applied the rules across the
years and states. Until 2017, states could predict with extreme accuracy whether the Department
would approve a waiver based on the listed criteria and guidance. It was only after the current
Administration took office that USDA began denying waivers that it had long approved – such as
no longer approving two-year waivers for areas that met the standards set in guidance or for areas
eligible under the Employment-to-Population ratio.56 While there were not a lot of these types of
requests historically, it was the new Administration that introduced uncertainty into the process.
Similarly, waiver requests that would typically be approved in one to three months can now take
upwards of six months to approve. This has resulted in USDA sometimes not approving waivers
until after the requested start date. FNS regional and national office staff have not known what
would and would not be approved or when. The political leadership at USDA has introduced
55

NPRM, p. 7.

56

U.S. Department of Agriculture, Food and Nutrition Service (FNS), “Supplemental Nutrition Assistance Program —
Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents (ABAWD),”
December 2, 2016, https://www.fns.usda.gov/sites/default/files/snap/SNAP-Guide-to-Supporting-Requests-toWaive-the-Time-Limit-for-ABAWDs.pdf.

23

uncertainty and inconsistency in the review and approval process. Moreover, they have been
inconsiderate of states’ need for certainty and predictability in order to implement waivers after
approval.
If the Department meant instead that the current waiver standards do not produce consistent
waived areas across states, then it is making an unreasonable argument. The only inconsistency
across states is the Department’s own application of the flexibility afforded to it, not in USDA’s
application of the rules (until recently). It is incumbent upon USDA to define and demonstrate the
inconsistency it observes given that this argument is a core element of its reason to re-regulate these
long-standing rules.
The evidence shows how little of the Department’s proposed rule is based on a clear knowledge
of the waiver policy’s history and an intimate understanding of the waiver standards’ application to
states. This clearly demonstrates the brittle nature of the Department’s justifications of the changes
to current waiver policy contained in the proposed rule.

24

Chapter 3: Setting a Floor for Waivers For Areas With
20 Percent Above National Unemployment Is
Inconsistent With Congressional Intent and Would Be
Harmful to Vulnerable Individuals
The most significant change of the proposed rule would drastically roll back waivers of the time
limit by requiring states to show that areas meet an unemployment rate threshold of 20 percent
above the national average (which the Department of Agriculture, or the Department, and we will
refer to as the “20 percent standard”) and, if the 20 percent standard is below a specific threshold,
meet this specific threshold, referred to as the “unemployment rate floor” to qualify for a waiver.
We believe this proposal is out of sync with the goal and purpose of the underlying legislation.
Furthermore, the Department did not discuss whether it considered a substantial body of relevant
research that contradicts the claims it made in support of this change and provided little to no
evidence to back up its proposal, making it difficult for us to comment on the process the Food and
Nutrition Service (FNS) used to develop this regulation. Below we discuss each of the following
reasons in detail that explain the flaws in this process:
• This

proposal is contrary to Congressional intent, which clearly was to allow states flexibility
to use a variety of metrics to demonstrate that the population subject to the time limit does
not have access to enough jobs. Congress has rejected past proposals to impose an
unemployment rate floor and otherwise restrict the current waiver criteria.

• Considerable

evidence shows that the adults without dependent children potentially subject to
the rule face overlapping labor market disadvantages, and therefore experience significantly
higher unemployment rates than the general unemployment rate for their area. Because an
area’s overall unemployment rate overstates job availability for the individuals subject to the
time limit, imposing an unemployment rate floor would disqualify many areas from eligibility
for a waiver where childless SNAP participants have very few job opportunities. The statute
clearly gives states that want to the ability to waive the time limit for some or all individuals in
areas where there aren’t enough jobs to employ these individuals.

• The

Department misleadingly cites the unemployment rate floor used by the Department of
Labor in establishing Labor Surplus Areas (LSAs) to support the proposal, without
recognizing that LSAs are meant for different purposes, and that LSAs also include an
unemployment rate ceiling.

• The

Department uses the concept of a “natural rate of unemployment” to support the
proposed unemployment rate floor of 7 percent, which is a misinterpretation of a
macroeconomic concept that is not a fixed or precisely identifiable unemployment rate.
Furthermore, the Department then suggests a significantly higher unemployment rate floor
than what it states the natural rate is without explaining how the natural rate relates to the
proposed unemployment rate floor of 7 percent. This lack of explanation for choosing the
substantially higher rate of 7 percent demonstrates how this specific unemployment rate floor
was chosen arbitrarily.

• While

no specific rate of unemployment would properly reflect these individuals’
circumstances, evidence shows that 7 percent unemployment specifically is too high, given
25

that many of these individuals are often in groups that experience unemployment rates
significantly above that level and they often face barriers to employment.
• The

proposal would fail to adequately provide states with waiver coverage during times of
rising unemployment, as the combination of the high unemployment rate with the lengthy 24month lookback would preclude many states with rising unemployment from eligibility. The
Department lacked transparency in not referencing whether they examined the potential
impact of this proposal at other times in the business cycle besides the current moment.

• The

Department attempts to support its proposed unemployment rate floor by explaining that
such a floor would decrease the share of who it refers to as “ABAWDs” living in a waived
area. This justification ignores Congressional intent and lacks transparency in the underlying
assumptions and methodology used to estimate this metric.

• The

Department also sought feedback on alternative unemployment rate floors of 6 and 10
percent, which are both unworkable and an inappropriate reading of the statute. Its proposal
of these alternate floors demonstrates the arbitrariness of the proposed 7 percent floor, but
also shows that it is impossible to designate a specific unemployment rate floor that would
adequately interpret the law by accurately reflecting jobs available to childless adults.

In proposing an unemployment rate floor for waivers based on the 20 percent standard, the
Department ignores the intent of Congress and uses misleading justifications with no transparent
evidence to support its claims. While the current 20 percent standard may not perfectly represent
areas that lack jobs for childless adults because the overall unemployment rate masks divergent labor
market opportunities for sub-groups such as these individuals, the proposed rule would only
exacerbate the shortcomings of current policy.

A. Unemployment Rate Floor Proposal Inconsistent With Congressional
Intent
When Congress established the three-month time limit in the Personal Responsibility and Work
Opportunity Reconciliation Act of 1996 (PRWORA), Public Law 104-193, it established that a state
may seek a waiver for a geographic area. Congress gave states this authority in recognition that
individuals may not have success in finding a job if there are limited job opportunities. When the
House Committee on Budget reported the original bill, the report stated:
The committee understands that there may be instances in which high unemployment rates in
all or part of a State or other specified circumstances may limit the jobs available for ablebodied food stamp participants between 18 and 50 years with no dependents. Therefore the
Secretary, upon request from a State, is provided with the authority to waive job requirements
in these circumstances or if unemployment rates are above 10 percent.57
Congress created waiver authority to enable states to waive areas with “high unemployment rates”
or “otherwise specific circumstances,” indicating that a range of circumstances may be indicative of
depressed labor market conditions. The welfare reform law established that a state could seek a
57

H.R. Report 104-651, Welfare and Medicaid Reform Act of 1996, https://www.congress.gov/congressionalreport/104th-congress/house-report/651.

26

waiver for an area if it: “(i) has an unemployment rate of over 10 percent; or (ii) does not have a
sufficient number of jobs to provide employment for the individuals.”58 (Herein, as with the current
regulations, we will use “area” to refer to geographic areas, which generally refers to areas for which
states generally seek waivers, such as counties, cities, towns, tribal areas, or metropolitan areas.)
Congress therefore created two distinct categories to establish the circumstances under which a
State can request a waiver:
• The

first criterion establishes that an area with an unemployment rate of 10 percent may
qualify for a waiver. The unemployment rate measures the share of the labor force that is
actively looking for work. Historically, a 10 percent unemployment rate is an indicator of
severe labor market distress, such as during an economic downturn. Since the Bureau of
Labor Statistics (BLS) began publishing monthly unemployment rates in 1948, the national
unemployment rate has equaled or exceeded 10 percent only during the 1981-1982 recession
and during the Great Recession of 2007-2009. Congress recognized that a local area with such
a high unemployment rate likely would not offer adequate job opportunities so that people
who are subject to the time limit could find work. With such high unemployment, even the
most readily employable jobseekers will likely struggle to find work, and those who are more
disadvantaged will face even more challenges. States that prefer to waive only areas with
extremely high unemployment rates can also request waivers based on this criterion.

• The

second criterion is focused on measuring employment opportunities for the specific
individuals affected by the time limit. Congress recognized that while useful for measuring the
health of a local labor market, the unemployment rate may not give a complete picture of job
availability for all workers in that area, particularly for individuals facing labor market
disadvantages. An area may not have a sufficient number of jobs because the share of
jobseekers who are out of work is relatively high, as indicated by the employment rate. Even
with a low unemployment rate, however, there can be instances where there aren’t enough
jobs to provide employment for specific individuals or groups. Even if there are enough jobs
in number to match the number of jobseekers, the individuals’ skills might not match the
requirements of the available jobs, the jobs may be inaccessible due to geographic or transit
limitations, or employers may discriminate against some jobseekers based on their race, work
history, disability, or other characteristics, for example.

In its original interpretation, the Department recognized that Congress intended for the
“insufficient jobs” criterion to include a range of metrics that are targeted towards the individuals
subject to the time limit. The Department published guidance on December 3, 1996, which stated:
The statute recognizes that the unemployment rate alone is an imperfect measure of the
employment prospects of individuals with little work history and diminished opportunities. It
provides States with the option to seek waivers for areas in which there are not enough jobs
for groups of individuals who may be affected by the new time limits in the Food Stamp
Program.

58

Food and Nutrition Act, 7 U.S.C. § 2015(o)(4). This language is identical to the language in P.L. 104-193, PRWORA.

27

To some extent, the decision to approve waivers based on an insufficient number of jobs
must be made on an area-by-area basis. Examples of such situations include areas where an
important employer has either relocated or gone out of business. In other areas there may be a
shortage of jobs that can be filled by persons with limited skills and work experience relative
to the number of persons seeking such jobs.59
The Department therefore originally (in 1996) interpreted the intent of Congress in creating the
second category for waiver authority as a recognition of the shortcomings of the unemployment rate
for measuring job opportunities for the individuals subject to the time limit, and established that it
could use flexibility in determining whether a State demonstrates a lack of jobs. In response to
comments, when preparing the original final rule, the Department balanced the need to provide
specific guidance that would be codified in regulation so that it would remain consistent across
subsequent administrations with the need to retain the flexibility that the Department recognized
that Congress had created in its original lawmaking. The final rule stated:
Based on these comments, we have decided to incorporate some of the more pertinent
aspects of the guidance into the regulation. More specifically, we have modified the
regulations at 7 CFR 273.24(f) to include a nonexhaustive list of the kinds of information a
State agency may submit to support a claim of 10 percent unemployment or ‘lack of sufficient
jobs.’60
FNS’ original (2001) interpretation therefore was clear that in providing guidance about specific
methods states can use to demonstrate a lack of jobs in an area, it was not precluding states from
using other data or metrics to demonstrate insufficient jobs, given that it is a concept not easily
shown by any one numeric quantity or metric.
By proposing an unemployment rate floor, the Department is proposing to restate the waiver
criteria in a manner that is inconsistent with the intent of the statute. Currently, states can waive
areas with insufficient jobs to employ a specific, more disadvantaged, population. The current 20
percent standard already has limitations in its ability to reflect jobs available for individuals subject to
the time limit, who likely experience much higher unemployment rates than the overall
unemployment rate in their area. As we discuss in detail below, areas with unemployment rates that
are 20 percent above the national average may still lack jobs for those with barriers to
unemployment. As we will explain, there are several reasons why current aspects of the 20 percent
standard in the context of the current regulations allow for a greater ability to demonstrate a lack of
sufficient jobs than the proposed regulation would allow. The proposed regulations would therefore
significantly worsen the problem with the current 20 percent standard as a measure of “insufficient
jobs.”
First, under the current regulation, an area with elevated unemployment compared to national
unemployment can qualify for a waiver, without meeting a specific unemployment rate standard.
Defining high unemployment at a relative level rather than a specific unemployment rate threshold
allows for greater consideration of trends such as those in labor force participation, which may
affect low unemployment rates, especially relevant for disadvantaged groups. If workers who are not
59

USDA, “Guidance for States Seeking Waivers for Food Stamp Limits,” December 3, 1996.

60

66 Fed. Reg., No. 11, 4438 (January 17, 2001), p. 4462. https://www.federalregister.gov/d/01-1025/p-205

28

employed stop looking for work and therefore exit the labor force, measures of labor force
participation will decline. Because the unemployment rate measures the share of the labor force that
is not employed but is actively seeking work, lower labor force participation may be a signal of weak
labor markets that is not reflected in the unemployment rate (for example, if discouraged workers
stop looking for work).
Overall labor force participation has fallen over the last two decades, including particularly sharply
during the Great Recession, and only began rebounding in about 2015. Labor force participation fell
sharply among prime-age workers (thus less affected by population aging and retirement) with lower
educational attainment from 2000 to 2015 and in 2018 were still below 2000 levels.61
Lower unemployment rates are thus less indicative of strong labor markets in recent years than in
the past, and particularly so for a group that tends to fare worse in the labor market, such as those
with lower levels of education. The 20 percent standard, which currently does not have a floor, relies
on unemployment rates, which are an imperfect proxy of jobs available for this population. Because
the current unemployment rate threshold needed to qualify for a waiver varies along with national
trends, however, the current standard gives more flexibility to capture those trends. Not having a
specific unemployment rate floor therefore allows for the 20 percent standard to better capture
insufficient jobs than it would with a specific floor. In addition, currently states have the ability to
group together counties to better represent local labor market opportunities, which the proposed
rule would also restrict. (We discuss these changes in more detail in Chapter 5.) This flexibility also
helps mitigate some of the shortcomings in the current 20 percent standard.
Second, the Department is also proposing to eliminate other criteria existing in current regulations
that can serve as an alternative to measuring “insufficient jobs” in cases where the 20 percent
standard does not adequately reflect job opportunities. In the context of these changes, the 20
percent standard takes on increasing importance as one of the sole methods to demonstrate a lack of
sufficient jobs. The effect of these proposed changes largely results in a requirement that states
demonstrate a specific unemployment rate threshold to qualify for a waiver under the “insufficient
jobs” criterion, when Congress expressly intended for this criterion to encompass a broader range of
metrics.
The Department proposes to eliminate most of the remaining alternatives to metrics based on the
unemployment rate that current regulations at 7 C.F.R. § 273.24(f)(2)(ii) allow, such as the
elimination of the option to demonstrate a “low and declining employment-population ratio” or to
demonstrate declining industries. The Department would also sharply reduce the ability of states to
request waivers for groups of neighboring counties, which may be useful in cases where the
unemployment rate is a particularly poor proxy for labor market opportunities for individuals
subject to the time limit. (We discuss the changes to employment-population ratio and other means
of showing a lack of sufficient jobs in Chapter 4, and changes to grouping in Chapter 5.) With these
changes, for the most part an area could only qualify for a waiver by demonstrating that it has a 12month unemployment rate average of at least 10 percent, a two-year unemployment rate of at least 7

61

Audrey Breitwieser, Ryan Nunn, and Jay Shambaugh. “The recent rebound in prime-age labor force participation,”
Brookings Institution, August 2, 2018. https://www.brookings.edu/blog/up-front/2018/08/02/the-recent-rebound-inprime-age-labor-force-participation/

29

percent, or that it qualifies for extended unemployment insurance benefits, the eligibility for which is
based on a recent three-month insured or total unemployment rate.
The proposal does allow for states to demonstrate “exceptional circumstances,”, but even then
suggests that it must support this claim with evidence, such as of a 10 percent unemployment rate:
“the request must demonstrate that the exceptional circumstance has caused a lack of sufficient
number of jobs, such as data from the BLS or a BLS-cooperating agency that shows an area has a
most recent three-month average unemployment rate over 10 percent.”62 Under the proposed rule,
states will largely be limited to demonstrating that an area meets a specific unemployment rate
threshold to qualify for a waiver under the “insufficient jobs” category of waivers, which does not
align with the intent of Congress to provide for multiple metrics under this category.
Congress regularly includes specific unemployment rate thresholds for policy purposes when that
is its intent. Congress included 10 percent unemployment as one of the criteria to qualify for a
waiver of SNAP’s three-month time limit, as stated above. Similarly, in the same legislation, Public
Law 104-193, Congress created a specific definition of a “needy state” under the TANF program,
which allows states additional weeks of job search and readiness. One of the qualifications for a
“needy state” was a three-month unemployment rate of at least 6.5 percent that exceeds 110 percent
of the unemployment rate for the same period in either of the last two years.63 Congress clearly
understood that unemployment rates may be an appropriate threshold in some instances, but chose
to include a criterion that was more loosely defined and allowed for alternative economic measures
to demonstrate a lack of jobs. Congress also chose to allow waivers based on economic
circumstances that reflect jobs available for a targeted population, the individuals subject to the time
limit. Had Congress intended to allow states to qualify for waivers only based on unemployment
rates, it would have only included waiver criteria with those unemployment rate parameters, rather
than including the second criteria targeted towards childless adult SNAP participants.
In the original final rule, published in 2001, the Department made clear that they interpreted the
“lack of sufficient jobs” as encompassing a broad range of metrics and not exclusively tied to
demonstrating a high unemployment rate. By proposing a specific unemployment rate threshold for
the 20 percent standard, reducing the ability of states to group together areas, and eliminating most
of the alternative criteria that would let states use alternative information, the Department has
substantially changed its interpretation of how states can demonstrate that an area lacks jobs for the
individuals subject to the time limit. In practice, except during times when states qualify for
extended benefits, under the proposed regulation, states would largely be limited to showing that an
area has a 7 percent unemployment rate over two years to show it lacks enough jobs to employ
people subject to the time limit. The Department did not attempt to demonstrate that a specific
unemployment rate threshold shows an area lacks jobs for these individuals, instead discussing the
unrelated fact that the proposal would have the effect of narrowing the number of waived areas,
which we explain below. The Department therefore provides no evidence that the changes in the
rulemaking are aligned with the intent of the statute to allow waivers in areas lacking sufficient jobs,
a broader concept than areas meeting specific unemployment rate thresholds.

62

NPRM, p.992.

63

Personal Responsibility and Work Opportunity Reconciliation Act of 1996, P.L. 104-193, § 403(b)(6),
https://www.congress.gov/104/plaws/publ193/PLAW-104publ193.pdf.

30

Congress Recently Rejected Proposals
to Limit Current Waiver-Approval Standards
Congress also has rejected attempts to narrow waiver-approval criteria to impose an
unemployment rate floor for the “20 percent standard.” H.R. 2, the House Agriculture
Improvement Act of 2018, as passed by the House on June 21, 2018, included a restriction similar to
the Administration’s proposal, requiring an area to have an unemployment rate of at least 7 percent
to qualify based on having a two-year unemployment average greater than the national average. The
Senate did not include such a restriction on waivers. The Conference Committee adapted the
Senate’s approach, which then passed and was signed into law. As Rep. Marcia Fudge, a conferee,
noted in the Congressional Record:
The Conference Committee also rejected House provisions that would shorten SNAP's threemonth time limit to one month and expand the population subject to the rule to a broader
group of recipients. We also rejected the House’s proposal to limit states’ flexibility to waive
high-unemployment areas from the three-month limit.64
Similarly, the Conference Report noted that Congress chose not to change the underlying statute:
The Managers also acknowledge that waivers from the ABAWD time limit are necessary in
times of recession and in areas with labor surpluses or higher rates of unemployment. The
Managers intend to maintain the practice that bestows authority on the State agency
responsible for administering SNAP to determine when and how waiver requests for
ABAWDs are submitted.65
Congress therefore chose not to change the criteria by which states could request area waivers.
While the Administration cited the House-passed version of the H.R.2 to support the proposed 7
percent unemployment floor, Congress ultimately rejected this proposal in favor of the Senate
approach, demonstrating intent to keep the current interpretation of the “insufficient jobs” criterion
intact.

B. Unemployment Rates Overstate Jobs Available to Childless Adult SNAP
Participants
By proposing an unemployment rate floor for the “20 percent standard,” the Department argues
that areas with unemployment rates below this threshold offer enough jobs so that those individuals
can find work. For example, when describing its support for its proposed unemployment rate floor,
the Department states, “The Department views the proposal as more suitable for achieving a more
comprehensive application of work requirements so that ABAWDs in areas that have sufficient
number of jobs have a greater level of engagement in work and work activities, including job
training.”66 The Department therefore states that areas with unemployment rates below its proposed
64

See the floor statement by Congresswoman Fudge, 164 Cong. Rec. H10149 (daily ed. December 12, 2018),
https://www.congress.gov/congressional-record/2018/12/12/house-section/article/H10142-3.
65

H.R. Rept. 115-1072, Agriculture Improvement Act of 2018, Title IV (3), https://www.congress.gov/congressionalreport/115th-congress/house-report/1072.
66

NPRM, p. 984.

31

floor of 7 percent over two years offer a sufficient number of jobs to the individuals subject to the
time limit. This interpretation that areas with lower unemployment have enough jobs to employ
adults without dependent children ignores the reality that overall unemployment rates overstate jobs
available to disadvantaged individuals.
The Department states that the unemployment rate floor proposal would prevent areas with low
unemployment from qualifying for a waiver but ignores evidence that the individuals subject to the
time limit are in demographic groups that experience higher unemployment rates than their area’s
average. In explaining why it chose to propose an unemployment rate floor, the Department noted:
Based upon operational experience, the Department has observed that, without an
unemployment rate floor, local areas will continue to qualify for waivers under the
Department’s 20 percent standard based on high unemployment relative to the national
average even as local unemployment rates fall to levels as low as 5 to 6 percent (depending
upon the national rate).
The Department is therefore stating that the floor is necessary to prevent areas with
unemployment rates it considers “low” from qualifying for a waiver. Adult SNAP participants
without dependent children, however, are likely to face barriers to employment that result in fewer
jobs available for those individuals than for the general population. It is unrealistic to set a specific
threshold that guarantees that the labor market creates a sufficient number of jobs to provide
employment to this group, and any such threshold based on the overall unemployment rate in an
area would guarantee that many areas where childless adult SNAP participants could not find work
were ineligible. When it explained its position that it does not believe areas with low unemployment
rates should qualify for waivers, the Department did not provide any research to support its position
that areas with low unemployment rates provide enough jobs so that the individuals subject to the
time limit can find work, nor did it address the extensive research that demonstrates that these
individuals struggle to find work even when unemployment rates are low. Because the Department
did not provide this information, it is difficult for commenters to understand how they are
interpreting a specific unemployment rate as measuring job availability for this population and to
respond to this reasoning.
The unemployment rate is a broad labor market metric that masks differences in the labor market
outcomes experienced by different groups. Some groups, such as African American workers, have
historically and consistently higher unemployment rates. The recent Great Recession also
demonstrated how less-advantaged groups fared more poorly in the recession, losing more jobs and
recovering more slowly.
Evidence shows that the adults targeted by the time limit often face barriers to work. While these
low-income adults without dependents are a diverse group and there has been limited research on
this specific population, the available evidence demonstrates that many face greater struggles to find
work than the overall population. This group, while diverse, has many characteristics that, as we will
explain below, are associated with worse labor market outcomes:
• Over

three-quarters of this group have a high school diploma or less, and studies show that
many lack skills sought by employers.

32

• This

group is demographically diverse. Of adult SNAP participants aged 18 through 49 who
do not receive disability income or have children in the household, about 53 percent are male,
and 47 percent are female. About two-fifths are aged 18 through 29, one-quarter are aged 30
to 39, and one-third are aged 40 to 49. About two-fifths are white, over one-quarter are
African American, and approximately 20 percent are Latino.67 They live in a range of areas:
about two-fifths live in urban areas, two-fifths live in suburban areas, and about 15 percent
live in rural areas. 68

• Like

most SNAP participants, this group largely works, but in low-wage jobs that provide little
stability, and as a result, many move in and out of work and experience periods when they are
out of work.

• Research indicates

that many of these individuals face barriers to employment, including low
skills, inconsistent work history, health conditions that limit their ability to work, inadequate
access to transportation, criminal justice history, or unstable access to housing.

Because this population is distinct from the United States population, and faces greater
disadvantages with regards to accessing employment, an overall unemployment rate or other overall
labor force metric will largely overstate the jobs available to this group. The section below explains
the research documenting the unique barriers to employment that childless adult SNAP participants
face, and the higher unemployment rates associated with many of these characteristics.
Childless and Non-Custodial Parent Adult SNAP Participants Are Likely to Have Lower
Levels of Educational Attainment, Which Is Associated With Higher Unemployment Rates
and More Sensitivity to Labor Market Shocks
The majority of adult SNAP participants without dependents have a high school education or less.
According to 2017 USDA Household Characteristics data, about one-quarter (24 percent) of nondisabled individuals aged 18 through 49 in households without children report having less than a
high school education, and about 54 percent report a high school diploma or a GED. (Some 8
percent do not report educational attainment.)69 They are more likely than other SNAP participants
to lack basic job skills like reading, writing, and basic mathematics, according to a 2003 Government
Accountability Office (GAO) study.70 A more recent study of SNAP employment and training
(E&T) participants, which includes many childless adults ages 18 through 49, but did not separately
report results for that population, found that three-quarters of employment and training providers
67

We looked at U.S. Agriculture Department’s fiscal year 2017 SNAP Households Characteristic data (QC), the 2017
American Community Survey (ACS) 1-year estimates, and the March 2018 Community Population Survey (CPS).
Reporting of race/ethnicity is voluntary and is missing for 13 percent of ABAWDs in QC. About 12 percent of
ABAWDs self-identified or were coded by an eligibility worker as “Latino or Hispanic”, but the share increased to 17
percent in high-reporting states (missing for less than 10 of SNAP participants). CPS and ACS capture more Hispanics
than QC. Hispanics account for 22 percent of ABAWDs in CPS and 20 percent in ACS. Compared to ACS, the
disability income questions are much more detailed and comprehensive in CPS.
68

CBPP analysis of the March 2018 Current Population Survey. Some 12 percent are unknown.

69

CBPP analysis of FY 2017 USDA Household Characteristics data.

70

“Food Stamp Employment and Training Program,” United States General Accounting Office, revised March 2003,
https://www.gao.gov/assets/240/237571.pdf.

33

surveyed found that at least some of the E&T participants they serve lack basic skills when they
enter the program, over half said some participants have low literacy levels or were high school
dropouts, and over two-fifths cited that participants’ skills were mismatched to industry needs or
were out of date. Over one-quarter of E&T participants surveyed identified limited education as a
barrier to employment.71 Caseworkers in a work experience program in Ohio found signs of
functional illiteracy even among those with a high school degree. 72
Research shows that adults with lower educational attainment have higher unemployment rates
than those with more education. (Figure 3.1.) For example, in 2018, while the unemployment rate
for workers with a bachelor’s degree or more was 2.1 percent, the unemployment rate for high
school graduates was 4.1 percent, and for those with less than a high school education, 5.6 percent.
African Americans with less than a high school diploma had an unemployment rate of 10.4
percent.73
FIGURE 3.1

71

Gretchen Rowe, Elizabeth Brown, and Brian Estes, “SNAP Employment and Training (E&T) Characteristics Study:
Final Report,” United States Department of Agriculture, Food and Nutrition Services, revised October 2017,
https://fns-prod.azureedge.net/sites/default/files/ops/SNAPEandTCharacteristics.pdf.
72

“A Comprehensive Assessment of Able-Bodied Adults Without Dependents and Their Participation in the Work
Experience Program in Franklin County, Ohio,” Ohio Association of Foodbanks, revised 2014,
http://admin.ohiofoodbanks.org/uploads/news/WEP-2013-2014-report.pdf.
73

“Employment Status of the Civilian Population 25 Years and Over by Educational Attainment,” Bureau of Labor
Statistics, revised February 1, 2019, https://www.bls.gov/news.release/empsit.t04.htm.

34

Workers with less education are more likely to lose jobs during an economic downturn and will
recover more slowly in the aftermath of a recession. Researchers have found that an increase of one
percentage point in the state unemployment rate leads to almost a two-percentage-point increase in
unemployment for workers with less than a high school degree compared to less than 0.5percentage-point increase for those with a college degree.74 Workers with a high school diploma had
lower employment rates in 2007 than college graduates: 55 percent for those with only high school
education, compared to 72.5 percent for those with a bachelor’s degree. Employment rates, or the
share of the population with a job, fell more sharply for the group with lower levels of educational
attainment, and in 2018 had yet to return to pre-recession levels.75 Counties with large shares of
workers with less than a high school degree also saw greater employment losses during the Great
Recession.76
Workers with less education may be hit harder by recessions in part because when unemployment
rises, employers may raise the skill requirements for positions: one study found that a onepercentage-point increase in the local unemployment rate raises the fraction of jobs requiring a
bachelor’s degree by about 0.4 percentage points and the fraction of jobs requiring two or more
years of experience by about 0.8 percentage points. 77 Evidence also suggests that for workers
entering the labor market during a recession, the effects can be long-lasting: those workers had
reduced earnings that persisted up to ten years into workers’ careers, and the effect was most
pronounced for those with less than a high school education, driven by greater losses in
employment. 78
The majority of adult SNAP participants without dependents have a high school diploma or lower
educational attainment. Evidence shows that workers with a high school diploma or less have higher
unemployment rates, lower employment rates, experience greater employment losses during
economic downturns, and recover more slowly. The overall unemployment rate therefore will
significantly overstate the employment opportunities available to less-educated workers, particularly
during a recession and the aftermath. FNS does not appear to have considered any of this research
in developing this proposal. We urge FNS to carefully review this literature, which demonstrates that
because adults with less education, typically have higher unemployment rates than the overall

74

Hilary Hoynes, Douglas L. Miller, and Jessamyn Schaller, “Who Suffers During Recessions?” Journal of Economic
Perspectives (Summer 2012), pp. 27-48. https://pubs.aeaweb.org/doi/pdfplus/10.1257/jep.26.3.27
Lauren Bauer and Jay Shambaugh, “Workers with Low Levels of Education Still Haven’t Recovered From the
Recession,” The Hamilton Project (September 2018), pp. 1-4,
http://www.hamiltonproject.org/blog/employment_rate_gap_workers_with_low_levels_of_education_still_havent_rec
ov.
75

76

Brian Thiede and Shannon Monnat, “The Great Recession and America’s Geography of Unemployment” Demographic
Research (September 2016), pp. 891-928. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5486972/
77

Alicia Modestino, Daniel Shoag, and Joshua Balance, “Upskilling: Do Employers Demand Greater Skill When Skilled
Workers are Plentiful?” Harvard Kennedy School Taubman Center for State and Local Government (May 2015), pp. 14. https://www.hks.harvard.edu/sites/default/files/centers/taubman/files/Upskilling.pdf
78

Hannes Schwandt and Till von Wachter, “Unlucky Cohorts: Estimating the Long-Term Effects of Entering the Labor
Market in a Recession in Large Cross-Sectional Data Sets,” Journal of Labor Economics, Vol. 37, No. 51 (January 2019),
S161-S198, https://www.journals.uchicago.edu/doi/10.1086/701046

35

average in their area, the proposed unemployment rate floor would be a much higher rate for adults
with less education, the majority of childless adults.
Over Two-Fifths of Childless Adult SNAP Participants Aged 18-49
Are African American or Latino, Groups That Experience Higher Unemployment Rates and
More Employment Discrimination
Over one-quarter of childless adult SNAP participants targeted by the time limit are African
American and approximately 20 percent are Latino.79 These groups, particularly African Americans,
also have higher unemployment rates than white Americans and are more affected by recessions.
Black and Latino workers generally have higher unemployment rates than white Americans.
According to data published by the BLS, in the fourth quarter of 2018, for example, the overall
unemployment rate was 3.6 percent and 3.2 percent for white workers, but Latinos had an
unemployment rate of 4.3 percent, and the unemployment rate for African Americans was 6.1
percent.80 In fact, for about the past four decades, unemployment rates among black workers have
been about double those of white workers. 81 This relationship is true even when comparing
unemployment rates for those with similar education levels. The unemployment rate among African
American workers with less than a high school education in 2018 was 10.4 percent, more than
double the unemployment rate of whites with the same education level, which was 5.1 percent.
Black high school graduates had unemployment rates of 6.7 percent in 2018, close to double the
unemployment rate for white high school graduates in 2018, of 3.5 percent.82
These disparities are also found at the local level. Researchers have found significant racial
disparities in labor force statistics within the same city, which may be explained in part by complex
and deeply rooted factors such as industry concentration, investments in housing and infrastructure,
and demographic trends. Chicago, San Francisco, Washington, and the borough of Manhattan all
had relatively low black employment rates in 2015 (56, 53, 64, and 62 percent, respectively), and
white employment rates that were at least 20 percentage points higher (83, 84, 88, and 85 percent,
respectively).83 It is unclear if the Department considered the consistently high unemployment rates
among African American and Latino workers when proposing a minimum unemployment rate floor

79

CBPP analysis of FY 2017 USDA Household Characteristics data, the March 2018 Current Population Survey, and
2017 American Community Survey (ACS) 1-year estimates.
80

“Table E-16. Unemployment Rates by age, sex, race, and Hispanic or Latino ethnicity,” Bureau of Labor Statistics,
revised January 4, 2019, https://www.bls.gov/web/empsit/cpsee_e16.htm.
81

Valerie Wilson, “Before the State of the Union, a fact check on black unemployment,” Economic Policy Institute,
February 2019, pp. 1-4. https://www.epi.org/blog/before-the-state-of-the-union-a-fact-check-on-blackunemployment/.
82

“Labor Force Statistics from the Current Population Survey, Table 7. Employment status of the civilian
noninstitutional population 25 years and over by educational attainment, sex, race, and Hispanic or Latino ethnicity,”
Bureau of Labor Statistics, revised January 18, 2019, https://www.bls.gov/cps/cpsaat07.htm.
83

Martha Ross and Natalie Holmes, “Employment by Race and Place: Snapshots of America,” Brookings Institution,
February 2017, pp. 1-16. https://www.brookings.edu/blog/the-avenue/2017/02/27/employment-by-race-and-placesnapshots-of-america/.

36

of 7 percent, which would essentially be an unemployment rate that is close to 14 percent for
African Americans.
Employment outcomes for African Americans are also more affected by the business cycle than
white Americans. One study found that over the period of 1990 through 2004, as the unemployment
rate increased by one percentage point, men were 0.16 percentage points more likely to become
unemployed, but this rate rose to 0.27 percentage points for African American men. Black men were
also less likely to transition from unemployment to employment than white men, though the
researchers found that this relationship didn’t change significantly during the business cycle, the
same study found. These results control for differences in education and other characteristics.84
Another study found that black and Latino workers are more likely to work part-time for economic
reasons than white workers, even after controlling for other demographic and economic differences
between the groups. This analysis found that this involuntary part-time work rose for all groups
during the Great Recession, but recovered much more quickly for white men than for black men,
with black men much less likely to transition from part-time to full-time work in the years following
the recession than white men.85
Multiple deep-rooted factors contribute to these employment disparities. For example, decades of
discriminatory housing policies have contributed to unequal access to quality education for black
children, which may affect employment opportunities later in life.86 In addition to these complex
causes, a large body of research also demonstrates that employer discrimination contributes to
higher unemployment rates among African Americans, especially compared to white Americans.
Researchers have conducted dozens of field studies over the past three decades in which they
have compared outcomes for otherwise identical job applications that differ only by racial or ethnic
markers (such as identical resumes with distinct names). One meta-analysis of such studies found
that white applicants receive 36 percent more callbacks than African Americans with the same
qualifications, and 24 percent more callbacks than Latinos. They found there was little change in the
callback disparities between white and black Americans over the 25 years studied, from 1990 to
2015, and a slight reduction in the disparities between Latino and white applicants, though barely
statistically significant.87 We strongly urge FNS to review all of these studies, as they help explain
why an unemployment rate is an especially poor predictor of job availability for African American
workers, who may not be hired for available jobs due to discrimination. For example:

84

Kenneth Couch and Robert Fairlie, “Last Hired First Fired? Black-White Unemployment and the Business
Cycle,” Demography (February 2010), pp. 227-247. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3000014/.
85

Tomaz Cajner et al., “Racial Gaps in Labor Market Outcomes in the Last Four Decades and over the Business
Cycle,” Federal Reserve Board, June 2017, pp. 1-33, https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2996084 .
86

Richard Rothstein, “The Racial Achievement Gap, Segregated Schools, and Segregated Neighborhoods – A
Constitutional Insult,” Economic Policy Institute, November 12, 2014, https://www.epi.org/publication/the-racialachievement-gap-segregated-schools-and-segregated-neighborhoods-a-constitutional-insult/.
87

Lincoln Quillian et al., “Meta-Analysis of Field Experiments Shows No Change in Racial Discrimination in Hiring
Over Time,” Proceedings of the National Academy of Sciences of the United States of America (April 2017), pp. 1-6,
https://www.pnas.org/content/early/2017/09/11/1706255114.

37

• Two field

studies, in Milwaukee and New York City, found consistently higher callbacks for
white applicants compared to African American applicants. Both studies had young men (ages
21 to 24) play the role of job applicants. They were matched with applicants with similar
appearance and verbal and social skills, and presented with similar resumes demonstrating
similar levels of education and job experience, and they received job interview training to be
similarly prepared. In both Milwaukee and New York, white applicants received callbacks or
job offers at roughly double the rate of African American applicants.88

• Another

field study found that black applicants were about half as likely to receive a callback
as white applicants. This study also found that white applicants who were recently released
from prison had similar levels of callbacks as black and Latino applicants: whites with criminal
records obtained positive responses in 17.2 percent of job applications, compared to 15.4
percent for Latinos and 13.0 percent for blacks.89

• One

field experiment found that when comparing outcomes of identical resumes with names
that were typically associated with white or black identities, white applicants had a 50 percent
higher chance of being called back.90

While they make up a small share of childless adults subject to the time limit, Native Americans
are likely to be disproportionately affected by this proposed rule given the estimate that many tribal
reservations may lose waiver eligibility, as outlined in Chapter 1. Native Americans also traditionally
have higher unemployment rates and worse labor force outcomes than white Americans, in part due
to sparse job opportunities on or near tribal and other rural areas and the legacy of historical factors
contributing to lower educational attainment and other barriers to employment. (Figure 3.2.)91

88

Devah Pager and Bruce Western, “Identifying Discrimination at Work: The Use of Field Experiments,” Journal of Social
Issues (2012) pp. 221-237,
http://scholar.harvard.edu/files/pager/files/identifying_discrimination_pager_western.pdf?m=1462807104.
89

Devah Pager, Bruce Western, and Bart Bonikowski, “Discrimination in a Low-Wage Labor Market: A Field
Experiment,” American Sociological Review (October 2009), pp. 777-799,
http://scholar.harvard.edu/files/bonikowski/files/pager-western-bonikowski-discrimination-in-a-low-wage-labormarket.pdf.
90

Marianne Bertrand and Sendhil Mullainathan, “Are Emily and Greg More Employable than Lakisha and Jamal? A
Field Experiment on Labor Market Discrimination,” National Bureau of Economic Research (July 2003), pp. 1-27
https://www.nber.org/papers/w9873.
91

Ed Bolen and Stacy Dean, “Waivers Add Key State Flexibility to SNAP’s Three-Month Time Limit,” Center on
Budget and Policy Priorities, updated February 6, 2018, https://www.cbpp.org/research/food-assistance/waivers-addkey-state-flexibility-to-snaps-three-month-time-limit.

38

FIGURE 3.2

This evidence shows that black and Latino workers as well as Native American workers, have
historically and consistently higher unemployment rates than white workers, and that these
outcomes cannot be explained solely by differences in education or other characteristics. Significant
numbers of individuals subject to the time limit therefore are in groups that experience
unemployment rates that are significantly higher than that of their state or local area. An
unemployment rate floor would therefore disallow states from requesting waivers from areas where
black and Latino workers have few job opportunities, even if the general unemployment rate for
their area is relatively low. If implemented, this proposal would therefore disproportionately harm
black, Latino and Native American adults subject to the time limit. This disparate impact of this
policy is therefore in conflict with 7 U.S.C. § 2020(c)(2), which establishes that program
administration of SNAP must be consistent with existing civil rights law.
Childless Adult SNAP Participants Are More Likely to Work in Jobs
With High Rates of Un- And Underemployment and Instability
The individuals who are targeted by the time limit do work, but in occupations where workers
experience instability, including underemployment, gaps in employment, and higher unemployment
rates. The general unemployment rate for the area therefore does not reflect the unemployment
rates for workers such as those in service occupations, who are more likely to be unemployed at any
given time than other workers.

39

SNAP participants who work generally work in service or sales occupations, such as cashiers,
cooks, home health aides, janitors, or drivers.92 A recent study of SNAP E&T participants, which
includes many childless adult SNAP participants ages 18-49, found that sales and service
occupations, such as cashiers and food preparation workers, were among the most common
reported by participants.93
There are higher unemployment rates among workers in many of these occupations. People who
report their occupation as a service occupation had unemployment rates about 23 percent higher
than the general unemployment rate in 2018, with food preparation and serving workers reporting
unemployment rates about 56 percent higher than the overall average.94 One analysis that looked at
working-age workers who did not receive disability income and did not have young children in the
household found that unemployment rates among those individuals were especially high for
cashiers, housekeepers, and laborers in 2017.95
Low-skill and low-wage workers are also more likely to be working part time for economic
reasons, and to cycle in and out of the labor force. For example, one study found that the share of
workers in low-skill jobs (classified by the types of tasks required, which are manual and routine, as
opposed to cognitive and non-routine) who were working part time involuntarily was about three
times that of workers in the highest-skill occupations (11 percent versus 3 percent), and another
found that involuntary part-time workers were concentrated in the retail trade and hospitality
industries.96 Another study observed a broader trend of workers cycling in and out of the labor
force.97
One of the reasons why these workers might have higher rates of un- and underemployment and
non-labor force participation is higher turnover: because these occupations lack stability, workers are
likelier to move in and out of jobs, and likelier to be unemployed or out of the labor force at a given
time or take a part-time job when they would desire a full-time job. Low-paying jobs often have
92

“SNAP Helps 1 in 10 Workers in the United States Put Food on the Table,” Center on Budget and Policy Priorities,
revised November 2018, https://www.cbpp.org/sites/default/files/atoms/files/factsheets_11-27-18fa_us.pdf.
93

Gretchen Rowe, Elizabeth Brown, and Brian Estes, “SNAP Employment and Training (E&T) Characteristics Study:
Final Report,” United States Department of Agriculture, Food and Nutrition Services, revised October 2017,
https://fns-prod.azureedge.net/sites/default/files/ops/SNAPEandTCharacteristics.pdf.
94

“Labor Force Statistics from the Current Population Survey, Table 25, Annual Averages, Unemployed persons by
occupation and sex,” Bureau of Labor Statistics, revised January 18, 2019, https://www.bls.gov/cps/cpsaat25.htm.
95

Kristin Butcher and Diane Whitmore Schanzenbach, “Most Workers in Low-Wage Labor Market Work Substantial
Hours, in Volatile Jobs,” Center on Budget and Policy Priorities, revised July 24, 2018,
https://www.cbpp.org/sites/default/files/atoms/files/7-24-18pov.pdf.
96

Maria E. Canon, Marianna Kudlyak, Guannan Luo, and Marisa Reed, “Flows To and From Working Part Time for
Economic Reasons and the Labor Market Aggregates During and After the 2007-09 Recession,” Economic Quarterly, Vol.
100, No. 2, Second Quarter 2014, Pp.87-111. https://www.richmondfed.org//media/richmondfedorg/publications/research/economic_quarterly/2014/q2/kudlyak.pdf; Lonnie Golden, “Still
falling short on hours and pay,” Economic Policy Institute, December 5, 2016, https://www.epi.org/publication/stillfalling-short-on-hours-and-pay-part-time-work-becoming-new-normal/#epi-toc-8.
97

John Coglianese “The Rise of In-and-Outs: Declining Labor Force Participation of Prime Age Men,” Working Paper,
February 28, 2018, https://scholar.harvard.edu/coglianese/publications/rise-of-in-and-outs.

40

irregular schedules that change from week to week. Workers in low-wage jobs are sometimes given
little notice of schedule changes or are expected to be on call, and are more likely to work part-time
hours when they would prefer a full-time schedule.98 Low-wage jobs are also more likely to lack paid
sick leave or other paid leave. For example, only 46 percent of workers in jobs with average hourly
wages in the bottom 25 percent of the wage distribution had paid sick leave in 2016, compared to 91
percent of workers in the highest-paid jobs (and 72 percent overall).99
Workers in jobs with lower wages, more volatility, and fewer benefits are more likely to experience
turnover, research shows. For example, a study found that workers with access to paid sick leave or
paid vacation were more likely to stay in their current job. This study found these effects even when
controlling for other characteristics of workers, such as education level or income, or characteristics
of jobs (such as the size of the firm and other benefits provided) that are associated with more job
separations.100 Another study that examined data from a large chain of retailers found that workers
who earned lower wages and had more schedule volatility (which was driven by changes in
consumer demand, not by employee choice) were more likely to leave their jobs; this study found
that these effects were not due to worker ability.101
In addition to job conditions, the lack of key supports such as stable housing may also contribute
to volatility or periods of joblessness among low-income workers. For example, recent research
finds that low-income renters who experience a forced move (such as following an eviction) are
more likely to be laid off from their jobs, compared to similar renters who did not experience a
forced move.102
At least in part because of these conditions, workers in low-wage jobs are more likely to be
employed at jobs for shorter periods and to experience periods of non-work. Workers with lower
levels of education (who are more likely to work in low-wage jobs) spend more weeks unemployed
than those with more education and they experience less wage growth over the course of their
lifetimes, according to studies by the Bureau of Labor Statistics that follow workers over time. These
studies also find that younger workers with less education are more likely to have short-term jobs of

98

For more information, see Brynne Keith-Jennings and Vincent Palacios, “SNAP Helps Millions of Workers,” Center
on Budget and Policy Priorities, May 10, 2017, https://www.cbpp.org/research/food-assistance/snap-helps-millions-oflow-wage-workers.
99

Bureau of Labor Statistics, Employee Benefits
Survey, https://www.bls.gov/ncs/ebs/benefits/2017/ownership/civilian/table32a.htm. This survey constructs hourly
wage percentiles by looking at average reported hourly wages for workers by occupation. (Because these averages are for
occupations, workers may fall into a different percentile than their occupation as a whole if they earn more or less than
the average for their occupation.)
100

Heather Hill, “Paid Sick Leave and Job Stability,” Work and Occupations, Vol. 40, Issue 2,
2013, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3825168/.
101

Saravanan Kesavan and Camelia Kuhnen, “Demand fluctuations, precarious incomes, and employee turnover,”
working paper, May 2017, http://public.kenan-flagler.unc.edu/faculty/kuhnenc/research/kesavan_kuhnen.pdf.
102

Matthew Desmond and Carl Gershenson, “Housing and Employment Insecurity Among the Working Poor,” Social
Problems, Vol. 63, Issue 1, February 2016, https://academic.oup.com/socpro/article-abstract/63/1/46/1844105.

41

six months or less, compared to workers of a similar age with more education. 103 Workers in jobs
that tend to have low wages, such as in the leisure and hospitality industry and service occupations,
also tend to stay at jobs for shorter lengths than workers in other jobs.104 We strongly urge FNS to
carefully review this research, which helps explain why unemployment rates may not capture the
dynamic nature of work for low-wage workers, who comprise the vast majority of SNAP
participants.
Working SNAP participants often work in occupations and industries with low wages and more
volatility. Compared to all workers, a greater share of workers who participate in SNAP are
employed in service occupations and in industries such as retail and hospitality, where jobs are more
likely to pay low wages and have other features of low quality.105 Childless adults are also likely to
experience gaps in employment, despite being employed regularly. For example, one study that
compared a snapshot, December 2013, with the 24-month period surrounding that month (January
2013 through December 2014), found that while three-quarters of childless adult SNAP participants
were in the labor force at some point during this period, only about one-third consistently worked at
least 20 hours per week throughout the entire period. 106 Because childless adult SNAP participants
work in jobs that contribute to periods of non-employment, unemployment rates in their area likely
do not capture their labor trends. The Department did not say whether it considered the
unemployment rate floor in the context of the types of jobs that childless adults are likely to work in.
Its lack of transparency makes it difficult to assess the claim that individuals subject to the time limit
have access to a sufficient number of jobs in an area with 7 percent unemployment over a two-year
period.
Many Childless Adult SNAP Participants Have Health Conditions
That Limit Their Ability to Work
While adults are exempt from the time limit if they are “medically certified as physically or
mentally unfit for employment,”107 evidence suggests that many childless adult non-elderly SNAP
participants have health conditions that serve as a barrier to employment. These adults may not fit
the state’s definition of “unfit for work” or may struggle to understand the rules or document their
condition in order to obtain an exemption. Research shows that adults with disabilities and other
health issues tend to have higher unemployment rates and fewer employment opportunities than
individuals without such conditions.

103

Bureau of Labor Statistics, “America’s Young Adults at 29: Labor Market Activity, Education and Partner Status:
Results from a Longitudinal Survey,” April 8, 2016, https://www.bls.gov/news.release/nlsyth.nr0.htm; Bureau of Labor
Statistics, “Number of Jobs Held, Labor Market Activity, and Earnings Growth Among the Youngest Baby Boomers:
Results from a Longitudinal Survey Summary,” August 24, 2017, https://www.bls.gov/news.release/nlsoy.nr0.htm.
104

Bureau of Labor Statistics, “Employee Tenure in 2016,” https://www.bls.gov/news.release/pdf/tenure.pdf.

105

Brynne Keith-Jennings and Vincent Palacios, “SNAP Helps Millions of Workers,” Center on Budget and Policy
Priorities, May 10, 2017, https://www.cbpp.org/research/food-assistance/snap-helps-millions-of-low-wage-workers.
106

Lauren Bauer, “Workers Could Lose SNAP Benefits Under Trump’s Proposed Rule,” The Hamilton Project,
December 2018,
http://www.hamiltonproject.org/blog/workers_could_lose_snap_benefits_under_trumps_proposed_rule.
107

42

7 C.F.R. § 273.24(c)(2)

Various sources illustrate the health conditions that many childless adult SNAP participants have.
Survey data indicate that among individuals aged 18 through 49 who do not receive disability
benefits or identify as disabled, nor have children in their household, about one-fifth report a health
problem or disability that prevents them from working or limits the type of work they can do, report
leaving the job or the labor force due to disability, or report not having worked in the last year due
to disability.108 Research funded by FNS reports that state caseworkers found multiple barriers to
employment among individuals subject to the time limit as they worked to implement welfare
reform’s time limits and work requirements. The most frequently cited barriers included medical or
mental health issues, or substance use disorders.109 A detailed study of childless adults who were
referred to a work experience program in Franklin County (Columbus), Ohio found that one-third
have a mental or physical limitation, including depression, post-traumatic stress disorder, mental or
learning disabilities, or physical injuries.110 A more recent study of characteristics of employment and
training participants, which includes many childless adults ages 18-49, found that about 30 percent
of E&T participants identified health issues as a barrier to employment. 111
Research finds that people with disabilities tend to face greater barriers to employment and have
worse labor force outcomes than individuals without disabilities. The Bureau of Labor Statistics, for
example, finds that the unemployment rate among working-age individuals with a disability (ages 16
to 64), was 10 percent in 2017, more than double the unemployment rate for working-age
individuals without a disability, which was 4.2 percent.112 Earlier BLS research describes some of the
barriers to employment that individuals with disabilities identified. About half of the individuals who
were out of the labor force or unemployed in May 2012 identified experiencing these challenges,
with the most common being that their own disability limited their work ability, and with smaller
shares identifying a lack of training, lack of transportation, or the need for accommodations. 113
Other research has identified challenges jobseekers with disabilities face, such as difficulty finding
appropriate jobs (which could include difficulty finding jobs that provide appropriate
accommodations), lack of social networks to facilitate job connection, or lack of accessible

108

CBPP analysis of the March 2018 Current Population Survey

109

John L. Czajka, et al., “Imposing a Time Limit on Food Stamp Receipt: Implementation of the Provisions and Effects
on Food Stamp Program Participation, Volume I,” U.S. Department of Agriculture, Food and Nutrition Service,
September 2001, https://fns-prod.azureedge.net/sites/default/files/abawd.pdf.
110

“A Comprehensive Assessment of Able-Bodied Adults Without Dependents and Their Participation in the Work
Experience Program in Franklin County, Ohio,” Ohio Association of Foodbanks, revised 2015,
http://ohiofoodbanks.org/wep/WEP-2013-2015-report.pdf.
111

Gretchen Rowe, Elizabeth Brown, and Brian Estes, “SNAP Employment and Training (E&T) Characteristics Study:
Final Report,” United States Department of Agriculture, Food and Nutrition Services, revised October 2017,
https://fns-prod.azureedge.net/sites/default/files/ops/SNAPEandTCharacteristics.pdf.
112

“Economic News Release: Table 1, Employment status of the civilian noninstitutional population by disability status
and selected characteristics, 2018 annual averages,” Bureau of Labor Statistics, revised June 21, 2018,
https://www.bls.gov/news.release/disabl.t01.htm.
113

“Persons With a Disability: Barriers to Employment, Types of Assistance, and Other Labor-Related Issues,” Bureau
of Labor Statistics, revised April 24, 2013, https://www.bls.gov/news.release/pdf/dissup.pdf.

43

transportation for job-seeking.114 Research has also documented that employers’ attitudes may also
harm jobseekers with disabilities, as employers may inadvertently underestimate the capacity of
individuals with disabilities.115
Many childless adults report physical or mental health conditions that limit their ability to find a
job and to work, and research shows that adults with disabilities face higher unemployment rates.
While some of these individuals may be exempt from the time limit based on disability, others face
difficulty documenting their health conditions, or have conditions that fall short of the “unfit for
work” standard. Imposing an unemployment rate floor for 20 percent standard waivers would
therefore cause many areas to lose waivers where individuals with health conditions are unable to
find jobs. The Department did not address the disparate impact of the proposed rule on people with
disabilities and other health conditions.
Childless Non-Elderly Adults Seeking Work
May Face Geographic or Transportation Limitations With Respect to Available Jobs
As stated above, about two-fifths each of childless adult SNAP participants without disabilities
live in urban or suburban areas, and about 15 percent live in rural areas. Many rural areas have
stalled economic development, which may result in relatively few job opportunities available. Even
in urban or suburban areas where jobs may be more plentiful, many workers face transportation
limitations to access those jobs. Some childless adults may face obstacles to finding work based on
geographic factors.
Individuals living in rural areas are less likely to be employed than other areas. Factors such as outmigration of younger workers and the aging of the remaining workforce, and declining infrastructure
and investment have contributed to this trend. Beginning in the 1970s, the share of men with less
than a high school education who are employed has declined more in rural areas than in urban areas.
By 2016, only about 50 percent of these men with lower educational attainment were employed,
about 15 percentage points lower than men without a high school diploma in urban areas.116 While
both rural and urban counties saw steep employment losses that began recovering in 2010, the
recovery stalled for rural counties, contributing to a significant employment gap between rural and
urban counties, according to a 2014 USDA study. Counties with large share of African Americans
saw greater impacts from the recession, which can only partially be explained by factors such as
industry mix or educational composition of those counties, this study found.117

114

Pamela Loprest and Elaine Maag, “Barriers to and Supports for Work Among Adults with Disabilities: Results from
the NHIS-D,” The Urban Institute, January 2001, pp. 1-22, https://www.bls.gov/news.release/pdf/dissup.pdf.
115

Katrina Vornholt et al., “Disability and Employment – Overview and Highlights,” European Journal of Work and
Organizational Psychology (2018), pp. 40-55, https://www.tandfonline.com/doi/pdf/10.1080/1359432X.2017.1387536.
116

James Ziliak, “Restoring Economic Opportunity for ‘The People Left Behind’: Employment Strategies for Rural
America,” Aspen Institute, revised 2019. https://www.aspeninstitute.org/longform/expanding-economic-opportunityfor-more-americans/restoring-economic-opportunity-for-the-people-left-behind-employment-strategies-for-ruralamerica/.
117

Tom Hertz et al., “Rural Employment Trends in Recession and Recovery,” U.S. Department of Agriculture,
Economic Research Service, revised August 2014,
https://www.ers.usda.gov/webdocs/publications/45258/48731_err172.pdf?v=0.

44

In addition to individuals living in rural areas may facing significant challenges to finding work,
many individuals in all types of areas do not have access to available jobs, either because their skills
do not align with the job requirements, or because transportation options to those jobs are
inadequate. In these areas, the area unemployment rate is an especially poor proxy for jobs available
to those individuals, if such jobs are inaccessible. We strongly urge FNS to carefully review all of the
below studies, which are key to understanding why an unemployment rate for an area does not
accurately portray the number of jobs available to the individuals subject to the time limit.
• The

number of jobs within a typical commuting distance in major metro areas fell by 7
percent between 2000 and 2012, with steeper losses for Latino and African American
residents, which fell by 17 and 14 percent respectively, and for residents with income below
the poverty line, which fell by 17 percent, compared to 6 percent for non-poor residents. The
majority of Census tracts with high poverty rates or a majority of residents of color
experienced losses in accessible jobs.118

• In a

number of metropolitan areas, low-income workers live far from available jobs, one
recent study found by comparing the distance between the residence of low-wage jobseekers
and job postings based on data from an online marketplace for hourly jobs. This study found,
for example, that in 12 major metropolitan areas, within at least 9 percent of zip codes in each
area, job postings far exceeded jobseekers in those zip codes.119 In some of these cities such as
San Francisco, jobs may be clustered in areas of the city where housing costs are high, and
low-wage job-seekers live farther away and have limited transit options. For example, in
Boston, low-wage job postings far exceed the number of applicants in 41 percent of zip codes,
and in New York, San Francisco, Chicago, Minneapolis, and Denver, available jobs far
outnumber job seekers in about one-quarter or more of zip codes. There are also many
pockets of metropolitan areas with the opposite issue, where many job seekers are clustered,
but available jobs are far from where they live. This is the case for over half of the zip codes in
Atlanta and Miami, where job applicants far outweigh open jobs (measured as zip codes in the
bottom quintile of job seekers minus job postings within 6.3 miles of the zip code’s center). In
cities such as Columbus, Detroit, Austin, and Nashville, there are far more job applicants then
there are available jobs in over one-quarter of zip codes. In many cities in the study such as
Columbus, Nashville, Dallas, and Washington, D.C., sizable shares of zip codes have both
problems, demonstrating the mismatch between the distribution of available jobs and
workers.120

• Increasing job accessibility,

a measure of proximity to employment opportunities relative to
other nearby jobseekers, significantly increases the chance of finding a job for African
Americans and Latinos, a 2006 study that looked at jobseekers in three major metropolitan
areas found. This shows how disparities in access to jobs contribute to disparities in labor

118

Elizabeth Kneebone and Natalie Holmes, “The Growing Distance Between People and Jobs in Metropolitan
America,” Metropolitan Policy Program at Brookings, March 2015, pp. 1-24, http://kedc.com/wpcontent/uploads/2015/04/Brookings_JobCommuteDistance2015.pdf.
119

The researchers identified zip codes where job postings “far exceeded” jobseekers as zip codes that are in the top
quintile of job seekers minus job postings within 6.3 miles of the zip code’s population-weighted centroid, which is the
average distance between job seekers and applicants in their dataset.
120

Christina Stacy, Brady Meixell, and Serena Lei. “Too Far from Jobs: Spatial Mismatch and Hourly Workers,” Urban
Institute, February 21, 2019, https://www.urban.org/features/too-far-jobs-spatial-mismatch-and-hourly-workers

45

market outcomes. The authors identified several factors as contributing to this phenomenon,
including that African Americans were more likely to live in central cities farther from
suburban areas with more jobs and were less able to move to a new neighborhood due to
housing segregation and residential discrimination; jobs were located in cheaper suburban
areas due to land use policy; and that there was a lack of public transit options and lower car
ownership rates among African Americans.
This study found that increasing accessibility by one standard deviation above the mean value
would increase the probability of completing a job search within six months by 61 percent for
black non-college graduates, while not increasing this effect for white workers with similar
levels of education. Having access to a car, searching in a job-rich area, being able to accept a
longer commute, having a higher-quality social network, and having more education were all
associated with an increase in the probability of finding a job within six months, while being
black or having child care concerns were associated with a decrease in this probability. The
cumulative effects of spatial job search variables such as job accessibility or car ownership
rates accounted for about 40 percent of the gap between the time it takes black and white
jobseekers to find a job. 121
• Increased

job accessibility reduced the length of time it took recently laid-off workers in nine
metropolitan areas to find a job. This study looked specifically at jobseekers who had been
employed but laid off to ensure that these individuals were searching for reasons unrelated to
characteristics associated with higher unemployment. It found that “an increase in one unit in
job accessibility (from -0.5 to 0.5) is approximately equal to an increase from the 20 th to the
80th percentile of job accessibility. Such an increase is associated with a 5.0 percent reduction
in search duration for finding any job, and a 6.6 and 8.3 percent reduction for accessions to a
new job with 75 and 90 percent of prior job earnings, respectively.” Black and Hispanic
workers were more sensitive to job accessibility than were white workers.122

• An analysis

of job accessibility in the Chicago metropolitan area found that increasing job
accessibility is linked with lower unemployment. At the mean, an increase in job accessibility
of one standard deviation was associated with a 0.43-point reduction in the unemployment
rate overall, a 0.57-point reduction in the African American unemployment rate, and a 0.47reduction in the unemployment rate for low-income households.123

Research shows how geographic factors can influence labor market outcomes such as
employment for individuals in ways that are not readily apparent based on unemployment rates. An
individual living in an area with a relatively low unemployment rate may not have access to jobs for
which they are qualified due to transportation limitations. Some rural areas may have relatively low
unemployment rates due in part to low labor force participation, and individuals living there may
have relatively few job opportunities. We are concerned that when considering an unemployment
121

Rucker Johnson, “Landing a job in urban space: The extent and effects of spatial mismatch,” Regional Science and
Urban Economics (February 2006), pp. 331-372, https://www.ssc.wisc.edu/~gwallace/Papers/Johnson%20(2006).pdf.
122

Fredrik Andersson et al., “Job Displacement and the Duration of Joblessness: The Role of Spatial
Mismatch,” National Bureau of Economic Research (April 2014), pp. 1-50. https://www.nber.org/papers/w20066.pdf.
123

Jangik Jin and Kurt Paulsen, “Does Accessibility Matter? Understanding the Effect of Job Accessibility
on Labour Market Outcomes,” Urban Studies (2018), pp. 92115. https://journals.sagepub.com/doi/abs/10.1177/0042098016684099.

46

rate floor for the purposes of time limit waivers, the Department did not appear to consider whether
job accessibility may limit the potential for childless SNAP participants to obtain a job, even if the
area in which they live has a relatively low unemployment rate.
Childless Adult SNAP Participants Report Other Barriers
That Are Associated With Higher Unemployment Rates
The unemployment rate for the area does not reflect the availability of jobs for adult SNAP
participants because these individuals face many disadvantages compared to the overall labor force.
In addition to some of the characteristics already discussed that are associated with higher
unemployment rates and other worse labor force outcomes, many individuals face barriers to work
that may make it more difficult to find available jobs, complete a job search, be selected by
employers, or maintain a job once employed. Here again, FNS’ proposed rule seemed to either
ignore or dismiss without explanation the considerable research that finds that unemployment rates
do not reflect job availability for the individuals subject to the time limit. We encourage FNS to
carefully review these research, which demonstrates how barriers to employment limit job
availability for people subject to the time limit even if the unemployment rate is low.
• Housing

instability and homelessness. Several studies have reported that some childless
adult SNAP participants lack access to stable housing and some experience homelessness. A
USDA research report looking at individuals first subject to the time limit found that
homelessness was among the barriers that case managers reported. 124 A GAO study that
looked at employment and training programs for childless adults also found that some case
managers reported housing difficulties as a barrier to work; for example, Colorado officials
estimated that about 40 percent of their employment and training participants experienced
homelessness.125 Similarly, a more recent USDA study of employment and training providers
found that over two-fifths of these providers identified a lack of stable housing as a barrier for
at least a quarter of participants in these programs, which include many adults targeted by the
time limit.126
Barriers to work among individuals experiencing homelessness are well-documented,
including limited skills and inconsistent work histories, lack of transportation, or physical or
mental health conditions.127 Those who are homeless may lack consistent access to resources
needed to maintain personal hygiene and meet dress codes, and the sleep deprivation and
stress of lacking housing may also affect these workers. People experiencing homelessness
also lack access to a reliable mailing address and may not have consistent access to a phone or

124

John L. Czajka et al., “Imposing a Time Limit on Food Stamp Receipt: Implementation of the Provisions and Effects
on Food Stamp Program Participation, Volume I,” U.S. Department of Agriculture, Food and Nutrition Service,
September 2001. https://fns-prod.azureedge.net/sites/default/files/abawd.pdf.
125

“Food Stamp Employment and Training Program,” United States General Accounting Office, revised March 2003,
https://www.gao.gov/assets/240/237571.pdf.
126

Gretchen Rowe, Elizabeth Brown, and Brian Estes, “SNAP Employment and Training (E&T) Characteristics Study:
Final Report,” United States Department of Agriculture, Food and Nutrition Services, revised October 2017,
https://fns-prod.azureedge.net/sites/default/files/ops/SNAPEandTCharacteristics.pdf.
127

David Long, John Rio, and Jeremy Rosen. “Employment and Income Supports for Homeless People,” 2007 National
Symposium on Homelessness Research,” https://aspe.hhs.gov/system/files/pdf/180356/report.pdf.

47

computer for job application and communication needs.128 Individuals who experience
evictions are also more likely to be laid off, research finds. 129
• Criminal records. Some

childless adult SNAP participants may face additional barriers to
work due to having a criminal record. For example, a study of childless adults referred to a
work experience program in Franklin County, Ohio found that about one-third reported
having a criminal record.130 Over half of SNAP E&T providers reported that a significant
share of participants reported a criminal record as a barrier to work. 131 An in-depth interview
study of SNAP participants who experienced periods of time with no other income,
approximately half of whom were ages 18 through 49 and who did not have dependent
children, found that about one-fifth of study participants had a criminal record that served as
a barrier to employment. 132 While many of these studies have relatively small study
populations and some are focused on populations who are more likely to have criminal justice
records and are therefore not always representative samples, it is clear that there are
individuals potentially subject to the time limit who have experience with the criminal justice
system.
Formerly incarcerated individuals face steep barriers to work, and as a result, on average face
periods of unemployment following release. Longitudinal studies that have tracked prisoners
upon their release, for example, find that up to half remain without a job 12 months after
release.133 A recent CBPP paper summarizes some of this research:134
o Studies document that the majority of individuals returning from incarceration face
health conditions. For example, one study found that one-half of men and two-thirds
of women had been diagnosed with chronic physical ailments such as asthma,

128

“Taking Away Medicaid for Not Meeting Work Requirements Harms People Experiencing Homelessness,” Center
on Budget and Policy Priorities, revised December 14, 2018, https://www.cbpp.org/sites/default/files/atoms/files/418-18health.pdf.
129

Matthew Desmond and Carl Gershenson, “Housing and Employment Insecurity among the Working Poor,” Social
Problems (January 2016), pp. 1-22.
https://scholar.harvard.edu/files/mdesmond/files/desmondgershenson.sp2016.pdf?m=1452638824.
130

“A Comprehensive Assessment of Able-Bodied Adults Without Dependents and Their Participation in the Work
Experience Program in Franklin County, Ohio,” Ohio Association of Foodbanks,
2014, http://admin.ohiofoodbanks.org/uploads/news/WEP-2013-2014-report.pdf.
131

Gretchen Rowe, Elizabeth Brown, and Brian Estes, “SNAP Employment and Training (E&T) Characteristics Study:
Final Report,” United States Department of Agriculture, Food and Nutrition Services, revised October 2017,
https://fns-prod.azureedge.net/sites/default/files/ops/SNAPEandTCharacteristics.pdf.
132

Claire Wilson and Brian Estes, “Examining the Growth of the Zero-Income SNAP Caseload: Characteristics,
Circumstances, and Dynamics of Zero-Income SNAP Participants,” U.S. Department of Agriculture, revised October
2014. https://fns-prod.azureedge.net/sites/default/files/ops/ZeroIncome-Vol2.pdf.
133

National Research Council, The Growth of Incarceration in the United States: Exploring Causes and Consequences, (Washington,
DC: The National Academies Press, 2014), pp. 233-259. https://www.nap.edu/catalog/18613/the-growth-ofincarceration-in-the-united-states-exploring-causes.
134

Elizabeth Wolkomir, “How SNAP Can Better Serve the Formerly Incarcerated,” Center on Budget and Policy
Priorities, revised March 16, 2018. https://www.cbpp.org/sites/default/files/atoms/files/3-6-18fa.pdf.

48

diabetes, hepatitis, or HIV/AIDS.135 People leaving jail and prison are three to six
times likelier than others to suffer from mental illness, another study found. 136
o Formerly incarcerated individuals also tend to lack education and training sought by
employers. They have an average of fewer than 12 years of education and in some
cases limited cognitive capacity, a history of behavioral problems, or a low level of
functional literacy.137 Furthermore, they miss out on opportunities to gain work
experience while in prison, and often do not have access to training programs.138
o Evidence also suggests that employers are more averse to hiring those with criminal
convictions than any other disadvantaged group, and formerly incarcerated
individuals can also face occupational licensing and other restrictions.139
Unemployment Rates Significantly Overstates Jobs Available
to Childless Adult SNAP Participants, Evidence Suggests
The preamble to the proposed rule suggests that adding an unemployment rate floor to qualify for
a waiver is necessary because “the Department believes a floor should be set for the 20 percent
standard so that areas do not qualify for waivers when their unemployment rates are generally
considered to be normal or low.”140 The Department proposes this unemployment floor to interpret
the statute which provides that states can waive areas that lack a “sufficient number of jobs to
provide employment for the individuals,” therefore suggesting that areas with unemployment rates
below the proposed threshold do have enough jobs to provide employment for the individuals who
are subject to the time limit.
A significant body of research, provided above, demonstrates why FNS’ reasoning is flawed and
lacks transparency. The area unemployment rate is a poor proxy for employment opportunities
available to adult SNAP participants without dependent children. These individuals on average are
more likely than other workers to have limited education and skills, experience discrimination, lack
geographic access to jobs, face housing instability, and experience other barriers to employment.
Many of these individuals likely experience multiple barriers that affect their ability to find a job. For
example, an African American worker with less than a high school education living far from
available jobs with no reliable transportation options will likely have access to far fewer jobs than
their area unemployment rate suggests. A rural area may have a low unemployment rate in part
because of low labor force participation, where many have given up searching for work due to few
135

Kamala Mallik-Kane and Christy A. Visher, “Health and Prisoner Reentry: How Physical, Mental, and Substance
Abuse Conditions Shape the Process of Reintegration,” Urban Institute Justice Policy Center, revised February 2008.
https://www.urban.org/sites/default/files/publication/31491/411617-Health-and-Prisoner-Reentry.PDF.
136

Henry J. Steadman et al., “Prevalence of Serious Mental Illness Among Jail Inmates,” Psychiatric Services (June 2009),
pp. 1-5. https://ps.psychiatryonline.org/doi/pdf/10.1176/ps.2009.60.6.761.
137

National Research Council, The Growth of Incarceration in the United States: Exploring Causes and Consequences, (Washington,
DC: The National Academies Press, 2014), pp. 233-259. https://www.nap.edu/catalog/18613/the-growth-ofincarceration-in-the-united-states-exploring-causes.
138

Bruce Western, “Collateral Costs: Incarceration’s Effect on Economic Mobility,” The Pew Charitable Trusts, revised
2010. https://www.pewtrusts.org/~/media/legacy/uploadedfiles/pcs_assets/2010/collateralcosts1pdf.pdf.
139

Elizabeth Wolkomir, “How SNAP Can Better Serve the Formerly Incarcerated,” Center on Budget and Policy
Priorities, revised March 16, 2018. https://www.cbpp.org/sites/default/files/atoms/files/3-6-18fa.pdf.
140

NPRM, p.984.

49

job opportunities. Not only did FNS not provide any evidence on whether it considered research
showing how the unemployment rate overstates jobs available based on different demographic
characteristics, it also did not provide research that shows how these economic conditions may
interact with each other and affect the opportunities to find work for the very disadvantaged
population that is subject to the time limit.
Because these adults are in many groups that have significantly higher unemployment rates than
the overall unemployment rate for their area, the proposed unemployment rate floor would likely
disqualify many areas where these individuals have few opportunities. The unemployment rate
among the group of individuals subject to the time limit in a county with a 7 percent unemployment
rate is likely much higher than 7 percent. Under the proposed regulation, states would be much less
effective at identifying areas where there are not enough jobs for the individuals subject to the time
limit, as many of these areas would have overall unemployment rates below the proposed threshold.
The Department did not discuss how the unemployment rate relates to job availability for
individuals subject to the time limit. It appears FNS ignored the considerable research that shows
how the individuals subject to the time limit belong to demographic groups with much higher
unemployment rates than the average or face barriers to accessing jobs. Without this research, it is
difficult to understand how it came to the conclusion that a 7 percent two-year unemployment rate
(or a one-year unemployment rate of 10 percent) accurately captures the number of jobs available to
this population.

C. Citation of Labor Surplus Area Unemployment Floor Inappropriate for
this Population
The Department cites the fact that the Department of Labor (DOL) has an unemployment floor
in its classification of Labor Surplus Areas (LSAs) as support for its proposal to impose a similar
floor for the purposes of waiver criteria, implying its approach is consistent with DOL’s. The
Department, however, proposes a higher floor than DOL uses without providing evidence, when
research suggests if anything the floor for this population would be substantially lower than DOL’s.
The Department also fails to acknowledge that DOL uses a 10 percent ceiling in its identification of
LSAs, demonstrating that its citation of LSAs is either misleading or based on incomplete
information.
In the preamble for the rule, the Department explained how the DOL has an unemployment rate
floor for Labor Surplus Areas, implying that implementing an unemployment rate floor for an area
to qualify for a waiver based on having unemployment rates 20 percent above the national average
would be appropriate to be consistent with DOL’s approach.141 Labor Surplus Areas are areas that
DOL identifies as having a “surplus of labor” based on having an unemployment rate of 20 percent
higher than the national average for a designated 24-month period. Federal, state, and local agencies
use LSA designations for multiple purposes. Executive Order 12073 required executive agencies to
“emphasize procurement set asides in order to strengthen our Nation's economy.”142 DOL lists
several other agencies that use Labor Surplus Areas, such as “The Small Business Administration
141
142

NPRM, p.983.

Executive Order 12073, Federal procurement in labor surplus areas, 43 FR 36873, 3 C.F.R., 1978 Comp., p. 216,
https://www.archives.gov/federal-register/codification/executive-order/12073.html.

50

uses the LSA list for bid selections for small business awards in Historically Underutilized Business
Zones (HUBZones).”143
DOL has an unemployment rate floor so that the minimum unemployment rate used for
identification of LSAs is at least 6 percent. DOL also has an unemployment rate ceiling; when
national unemployment is high enough that 20 percent above the national average exceeds 10
percent unemployment over two years, DOL will designate LSAs that have unemployment rates
above 10 percent.144 DOL also allows states to demonstrate that areas meet alternative criteria to
demonstrate an exceptional circumstance, such as a recent three-month unemployment rate at the
threshold required for LSA designation.
The Department established identification of Labor Surplus Areas as one of a non-exhaustive list
of methods of demonstrating “insufficient jobs” in its original 1996 guidance, and it was codified as
an example in the final 2001 regulation.145 The guidance and final rule also allowed states to use
similar data demonstrating unemployment rates that are 20 percent above the national average for a
24-month period.
FNS decided, in its 1996 guidance and 2001 regulation, to include LSA designation as one of
many ways that states can demonstrate that an area lacks sufficient jobs. This decision likely reflects
the fact that unemployment rates are readily available on a monthly basis and are statistically reliable
for sub-state areas. There are few alternative measures available at the state and local level that states
can use to demonstrate a lack of jobs. For example, at the national and state level there are measures
such as “alternative measures of labor underutilization,” a broader set of metrics that include
“discouraged workers,” who are workers who want a job but have not recently searched for work
because they believe no jobs are available to them, as well as others such as workers who would like
to work full time but can only find a part-time job.146 Those metrics cannot be reliably calculated at
the sub-state level. It is reasonable to use metrics developed by DOL for the express purposes of
identifying areas with excess labor compared to jobs, given that these metrics can facilitate the
process for states. It is also reasonable, however, to adapt these criteria to more accurately capture
the intent of the law, which is to capture jobs available for a sub-population. Current regulations
make such an adaptation by adopting the “20 percent standard” without a floor.
In proposing to use a similar unemployment floor to that used by DOL in the LSA designation,
the Department does not consider that while a specific floor may be appropriate for DOL’s
purposes in establishing LSAs, the waiver criteria are meant to represent a fully distinct concept
from LSAs and therefore adapting them to the different purpose of waiver criteria is necessary and
appropriate. The Department correctly notes that waiver criteria do not currently require an area to

143

U.S. Department of Labor, “Labor Surplus Areas: Frequently Asked Questions,”
https://www.doleta.gov/programs/lsa_faq.cfm.
144

Ibid.

145

USDA, “Guidance for States Seeking Waivers for Food Stamp Limits,” December 3, 1996 and Federal Register, Vol.
66, No. 11, January 17, 2001. P. 4462. https://www.federalregister.gov/d/01-1025/p-205.
146

Bureau of Labor Statistics, “Alternative Measures of Labor Underutilization for States, 2018 Annual Averages.”
https://www.bls.gov/lau/stalt.htm

51

meet a specific threshold if the area’s unemployment rate is above the national average, unlike the
LSA criteria.
These measures are meant to serve different purposes, however:
• The

suggested criteria for waivers of the time limit are meant to establish a lack of “sufficient
jobs” for a specific sub-population, childless adult SNAP participants, which faces labor
market disadvantages. As noted in section B above, childless adult SNAP participants belong
to groups that have higher unemployment rates than their local area, which makes defining a
specific unemployment rate at which those individuals have access to enough jobs difficult, if
not impossible. For example, since this group may experience unemployment rates that are
double the rates of their area, a 6 percent or 7 percent unemployment rate “floor” would
mean a 12 or 14 percent unemployment rate for this group of individuals. At any
unemployment rate threshold, it is likely that large groups of childless adult SNAP participants
would not have access to available jobs given their serious barriers to work. Given the
uncertainty and difficulty in establishing whether there are jobs available for this population,
not requiring a specific unemployment rate threshold appropriately allows for greater
flexibility in determining areas with insufficient jobs.

• LSAs,

on the other hand, establish the economic condition of an area to enable prioritization
of procurement contracts and economic development purposes. In creating the threshold for
LSAs, DOL does not need to consider whether jobs are available for specific types of
individuals, and instead is focused on understanding the macroeconomic conditions of an area
in order to direct economic stimulus. It therefore may be reasonable for the Department of
Labor to establish specific thresholds that meet those criteria for those purposes, given that
the criteria attempt to establish levels at which economic stimulus is needed.

In citing Labor Surplus Areas as a reason to implement an unemployment rate floor, the
Department also fails to acknowledge the unemployment rate ceiling that DOL uses in LSA
designation. Not only does DOL have an unemployment rate floor of 6 percent unemployment in
its criteria for LSA designation, it also has an unemployment rate ceiling of 10 percent
unemployment. This means that when designating LSAs, the unemployment rate threshold DOL
uses is 20 percent above the national average but not more than 10 percent. The Department not
only does not acknowledge this fact, it suggests this unemployment rate as a possible floor for the time
limit:
The Department would also like to receive comments on establishing a floor of 10 percent for
the 20 percent standard. A 10-percent floor would allow for even fewer waivers than the other
options and would result in the work requirements being applied in almost all areas of the
country.147
DOL establishes this unemployment rate ceiling for LSA designation in recognition that a
sustained level of unemployment at 10 percent is adequately high to demonstrate that an area has
severe labor market weakness. These criteria ensure that during times of widespread elevated
147

52

NPRM, p.984.

unemployment, areas can qualify without having exceptionally high unemployment rates. For
example, the LSA list in fiscal year 2012 was based on unemployment rates between January 2009
and December 2010, during the height of the Great Recession when national unemployment was 9.5
percent, and 20 percent above that would have been 11.4 percent.148 Areas were also eligible in fiscal
year 2013 and fiscal year 2014 with 10 percent unemployment for the same reason. 149 Without a
ceiling, many areas that were struggling during the height of the recession and recovery would have
been ineligible for LSA designation.
FNS therefore is proposing to pick and choose which features of LSA designation to adapt to the
proposed regulation without discussion of why it made this choice, or even acknowledgement of
this choice. The Department proposes on one hand to implement an unemployment rate floor for
an area to qualify under the “20 percent standard,” but does not propose a similar unemployment
rate ceiling. In fact, the Department proposes what DOL recognizes as a sufficiently high
unemployment rate to qualify for a ceiling, 10 percent unemployment, as a possible unemployment
rate floor. The Department therefore proposes to ensure that unemployment rates in an area must
meet a standard to demonstrate they are high but is not proposing a means of limiting this threshold
during a recession to ensure that the unemployment rate threshold does not provide too high a bar
that it would substantially bar areas suffering from a recession. The Department does not explain
why it chose to adopt an unemployment rate floor similar to that used in LSA criteria but not an
unemployment rate ceiling, and it does not mention the LSA ceiling. This oversight is particularly
perplexing given our research review which indicates if anything, the LSA criteria as is are very
stringent for waiver criteria given the barriers to employment childless adult SNAP participants face,
which would recommend no floor or a very low floor if any at all.
Without any explanation, it appears that by imposing a floor and not a ceiling for the 20 percent
standard, the Department considers for the purposes of measuring whether an area lacks adequate
jobs for childless adult SNAP participants that there is a level of unemployment that is low enough
to ensure that adequate jobs are available, but not a level high enough to signify that there likely are
not enough jobs. Again, without this information, it is difficult to assess the evidence that the
Department used in proposing an unemployment rate floor, especially given that the Department’s
choice seems to contradict all available economic evidence indicating that unemployment rates are a
poor proxy for jobs for this population.

D. Citation of “Natural Rate of Unemployment” Incorrectly Assumes This Is
a Fixed and Accurately Measurable Concept
The Department uses the macroeconomic concept of the “natural rate of unemployment” to
justify its proposed unemployment rate threshold. This use of this concept inappropriately applies a
macroeconomic concept and inaccurately displays economic consensus. The preamble states:

148

Department of Labor, “Labor Surplus Area Classification under Executive Orders 12073 and 10582 2012,”
https://www.doleta.gov/lsa/Archived/2011-2012/Federal_Register_2012_Final.pdf
149

Department of Labor, “Labor Surplus Area Classification under Executive Orders 12073 and 10582 2013”
https://www.doleta.gov/lsa/Archived/2012-2013/Federal_Register_2013_Final.pdf; Department of Labor, “Labor
Surplus Area Classification under Executive Orders 12073 and 10582 2014,”
https://www.doleta.gov/lsa/Archived/2013-2014/2013-2014_LSA_Federal_Register_Notice.pdf

53

The Department believes a floor should be set for the 20 percent standard so that areas do
not qualify for waivers when their unemployment rates are generally considered to be normal
or low. The “natural rate of unemployment” is the rate of unemployment expected given
normal churn in the labor market, with unemployment rates lower than the natural rate
tending to result in inflationary pressure on prices. Thus, unemployment rates near or below
the “natural rate of unemployment” are more indicative of the normal delay in unemployed
workers filling the best existing job opening for them than a “lack of sufficient jobs” in an
area. Generally, the “natural rate of unemployment” hovers around 5 percent. The
Department believes that only areas with unemployment rates above the “natural rate of
unemployment” should be considered for waivers.150
The Department appears to be suggesting that the natural rate of unemployment is a specific
unemployment rate figure that can be used in setting a waiver floor to examine job opportunities for
childless adult SNAP participants. This reasoning is deeply flawed.
First, the so-called natural rate of unemployment is not a known or even an observable jobless
rate. It is a concept that derives from the theoretical construct that there exists an unemployment
rate that is consistent with stable inflation. If unemployment falls below this “natural rate,” inflation
would rise, and vice versa. More colloquially, too low an unemployment rate, where “too low”
means the actual rate is below the natural rate, and the economy will overheat; too high a jobless rate
relative to the natural rate and inflation will fall.
In theory, an estimate of the natural rate should be derivable from observing the (negative)
correlation between changes in the rate of unemployment and that of inflation. However, because
this correlation appears to have moved toward zero over time, our ability to reliably identify a
policy-relevant natural rate, meaning one that could fruitfully be referenced as the Department
suggests in terms of their proposal, is much diminished.
Note, for example, a recent article about this problem by economics journalist Neil Irwin. In the
article, former Fed Vice-Chairman Alan Blinder notes that the “confidence interval” — the band of
statistical uncertainty around the estimate — is such that the concept cannot be usefully employed as
a policy benchmark: “If your range is 2.5 to 7, that doesn’t tell you anything.” 151
Figure 3.3 below reveals the problem using a standard statistical procedure to measure the
inflation/unemployment correlation. The figure represents the coefficient from a regression of core
inflation on lagged inflation and the gap between the unemployment rate and the Congressional
Budget Office’s estimate of the natural rate. The estimates are made using “rolling regressions,”
meaning we estimate the model over 20-year periods, beginning with 1959-79, and advance the
sample one year at a time. We then plot the coefficient on the unemployment gap variable.
In this area of economics, the measure is considered to be the slope of the Phillips Curve, which
is the curve that plots the unemployment/inflation tradeoff. The two lines surrounding the estimate
represent the bounds of a 95 percent confidence interval around the estimate. When these lines
150
151

NPRM, p.984.

Neil Irwin, “How Low Can Unemployment Really Go? Economists Have No Idea,” New York Times, Feb. 28. 2018,
https://www.nytimes.com/2018/02/28/upshot/how-low-can-unemployment-really-go-economists-have-no-idea.html.

54

include zero, as they do for most of the figure, the estimate of the slope is insignificant. In other
words, in these years, the “natural rate” cannot be reliably distinguished from a range of rates that
includes zero.
FIGURE 3.3

In other words, at least by this conventional approach, the Department cannot reliably use 5
percent (or any other level) as an estimate of the natural rate, because the diminished correlation
between unemployment and inflation renders such an estimate statistically insignificant.
Department Proposes Arbitrary Unemployment Rate Floor
Not only does the Department improperly support its proposed unemployment rate floor with a
flawed discussion of the “natural rate of unemployment,” it proposes a floor that is arbitrarily and
significantly higher than what it states is the unemployment rate consistent with the natural rate of
unemployment. The Department proposes an unemployment rate floor that is 40 percent above
what it states is “generally” considered the natural rate of unemployment, of 5 percent. This
difference is not insignificant. With a labor force of about 160 million people, a difference of 2
percentage points is equivalent to more than 3 million people nationwide, employed or not. In a
recent 24-month period of January 2017 through December 2018, 795 counties (or equivalent
entities) had unemployment rates above 5 percent, while only 155 counties had 24-month
unemployment rates above 7 percent.
Considering the difference between those two metrics, it is not clear how the Department used
the concept of the natural rate in developing this floor. These two unemployment rates, 5 and 7
percent, are so different that it is difficult to understand how they are linked without more
information. Furthermore, if the Department were actually basing the proposed unemployment rate
55

floor on a concept that describes the level of unemployment at which inflation increases, it would
need to demonstrate how this concept relates to the specific population or the purpose of
establishing waiver criteria, which is to interpret the “insufficient jobs” criterion in the law targeted
towards a disadvantaged group of individuals.
The Department’s proposal for this 7 percent unemployment floor therefore suggests that it did
not in fact use the natural rate of unemployment to develop the 7 percent unemployment rate floor
proposal. Either the Department used economic data relating the goals of the unemployment rate
floor to the natural rate, in which case it lacked transparency by not providing this research, or the
rate is an arbitrary selection unrelated to the statute that the rule is interpreting, in which case the
discussion of the natural rate is irrelevant to the actual proposal. Without an explanation of how and
why the Department used the natural rate concept to come up with a 7 percent unemployment rate
floor that is related to the purposed of the underlying statute, it is impossible to meaningfully
comment.

E. Evidence Suggests That a 7 Percent Unemployment Floor Is
Inappropriately High for This Population
Not only does the Department not provide economic evidence to support its proposed 7 percent
floor, evidence shows why this floor would be inappropriate for this population. While this
comment argues that we cannot assume that any particular unemployment floor will provide the
necessary labor market opportunities to some groups of workers, the proposed floor of 7 percent is
surely too high. As Figure 3.4 below shows, using national BLS data, there were 106 months since
1972 when the overall unemployment was between 6.5 and 7.5 percent. The average rate was 7.1
percent, about the level of the proposed floor. But unemployment for African American and Latino
workers was a much higher 13.9 percent and 10.2 percent. White unemployment was 6.2 percent.152

152

The results are very similar: blacks, 13.5 percent; Hispanics, 10 percent — when looking at minority unemployment
rates conditional on the two-year average of overall unemployment centered on 7 percent to more closely simulate the
proposed rule.

56

FIGURE 3.4

Consider that at the depth of the Great Recession (2009-10), broadly recognized as the deepest
recession since the Depression, the overall unemployment rate hit 10 percent. This was widely, and
correctly, seen as evidence of a huge, negative demand shock, one requiring an aggressive response
from both fiscal and monetary authorities. And yet, the proposal suggests an unemployment floor
that corresponds historically to a black unemployment rate well above the overall rate at the worst of
the recession.
Other rates that have been floated suffer from the same problem that even in the best of overall
labor markets, certain groups face a much less welcoming set of job opportunities. Table 3.1 below
shows the results from a simple regression of total rates on a constant and rates for black and Latino
workers. Even at 5 percent unemployment, the African American jobless rate is predicted to be
almost 10 percent, and the Latino rate is at 7 percent. At rates above 7 percent, both black and
Latino workers have jobless rates close to and in double digits.
TABLE 3.1

Unemployment Rates and Predicted Black and Latino Unemployment
Predicted Unemployment Rates
Unemployment
5%
7%
10%

Black
9.6%
13.2%
18.5%

Latino
7.0%
9.8%
13.9%

Note: Black and Latino unemployment rates are predicted by regressing total unemployment rate on race-specific rates.

57

Turning from minorities to other less advantaged groups in the labor market reveals similarly wide
disparities between their unemployment rates and the floor in the proposed rule.
• Since

2008, the BLS has tracked unemployment among those who self-report as disabled. In
months when the unemployment rate averaged 7 percent, the disabled rate was 13 percent.

• Doing the

same comparison by education level (for job seekers 25 and older, per BLS
published data), yields jobless rates of 11 percent for those with less than high school degrees.

F. Local Areas With High Unemployment Rates for Sub-Groups
We can also see how in local areas, when the overall unemployment rate is 5, 6, or 7 percent,
some sub-groups face much higher unemployment rates. Of the 239 large labor market areas
(metropolitan areas) with average overall unemployment rates below 7 percent,153 85 have
unemployment rates that are at least 14 percent for particularly vulnerable groups, such as adults
ages 25 to 64 with very low education, people with self-reported disabilities, and/or black and Latino
residents. The data this analysis includes are published Census Bureau figures from the American
Community Survey and average together five years of data from 2013 through 2017 in order to
increase reliability.154
15 metropolitan areas in 2013-2017 had unemployment rates of 7 percent or less overall but
at least 14 percent for the least educated workers. For example, in the Springfield, IL metro
area, for example, the overall unemployment rate was 6.7 percent, but was 21.2 percent for those
with less than a high-school diploma. (Table 3.2.)
TABLE 3.2.

Metropolitan Areas With 5-year Average Unemployment Rates Less Than 7
Percent Overall but Greater Than or Equal to 14 Percent for Workers With Less
Than a High School Education

State
FL
ID
IL
IL
IN
KY

Metropolitan Area
Pensacola-Ferry Pass-Brent, FL
Pocatello, ID
Springfield, IL
Peoria, IL
Kokomo, IN
Elizabethtown-Fort Knox, KY

Overall Unemployment
Rate for Population 16
Years and Over
6.9
6.3
6.7
6.6
6.5
6.7

Unemployment Rate
for Workers With Less
Than High School
Education
16.3
15.2
21.2
15.6
14.2
16

153

U.S. Census Bureau, American Community Survey Table S2201. The United States has 389 metropolitan areas
overall, without regard to unemployment rate.
154

To further increase reliability, we impose two additional requirements on the comparisons. First, subgroup
unemployment rates in a given metropolitan area are only included in our analysis if they are statistically significantly
higher than the overall unemployment rate for the area. Second, we only include estimated unemployment rates that are
at least twice as large as their margins of error.

58

TABLE 3.2.

Metropolitan Areas With 5-year Average Unemployment Rates Less Than 7
Percent Overall but Greater Than or Equal to 14 Percent for Workers With Less
Than a High School Education

State
MI
MI
MO-IL
NY
NY
OH
OHPA
VA
WV

Metropolitan Area
Kalamazoo-Portage, MI
Monroe, MI
St. Louis, MO-IL
Syracuse, NY
Elmira, NY
Canton-Massillon, OH
Youngstown-Warren-Boardman,
OH-PA
Blacksburg-ChristiansburgRadford, VA
Parkersburg-Vienna, WV

Overall Unemployment
Rate for Population 16
Years and Over

Unemployment Rate
for Workers With Less
Than High School
Education

6.6
5.9
6.3
6.6
5.3
6.6

17.2
16.7
14.7
14.4
14.5
15.7

6.9

14.1

5.2

15.5

6.1

17.3

59 metro areas in 2013-2017 had unemployment rates of 7 percent or less overall but at
least 14 percent for workers with any disability. For example, in the Peoria, IL metro area, the
overall unemployment rate was 6.6 percent, but was 18.4 percent for those with any disability. (Table
3.3.)
TABLE 3.3

Metropolitan Areas With 5-year Average Unemployment Rates Less Than 7
Percent Overall but Greater Than or Equal to 14 Percent for Workers With Any
Disability

State
AL
AL
AL
AL
CA
CA
CA
CT
DE

Metropolitan Area
Florence-Muscle Shoals, AL
Decatur, AL
Birmingham-Hoover, AL
Daphne-Fairhope-Foley, AL
Santa Maria-Santa Barbara, CA
Salinas, CA
Napa, CA
Hartford-West Hartford-East Hartford,
CT
Dover, DE

Overall
Unemployment Rate
for Population 16
Years and Over
6.7
6.6
6.8
5.5
6.6
6

Unemployment Rate
for Workers With a
Disability
16.3
17.6
15.4
14.3
14.5
14.4

5.4

16

6.8

15.6

6.7

14.9
59

TABLE 3.3

Metropolitan Areas With 5-year Average Unemployment Rates Less Than 7
Percent Overall but Greater Than or Equal to 14 Percent for Workers With Any
Disability

State
FL
FL
FL
FL
FL
FL
ID
IL
IL
IL
IN
KY-IN
LA
LA
LA
MA-CT
MD
ME
ME
MO
MO-IL
MO-IL
NC
NC
NC
NY
NY
NY
NY
NY-NJ-PA
OH
OH-KY-IN
60

Metropolitan Area
Deltona-Daytona Beach-Ormond
Beach, FL
Gainesville, FL
Pensacola-Ferry Pass-Brent, FL
Tampa-St. Petersburg-Clearwater, FL
Orlando-Kissimmee-Sanford, FL
North Port-Sarasota-Bradenton, FL
Pocatello, ID
Springfield, IL
Peoria, IL
Champaign-Urbana, IL
Fort Wayne, IN
Louisville/Jefferson County, KY-IN
Houma-Thibodaux, LA
Shreveport-Bossier City, LA
Baton Rouge, LA
Worcester, MA-CT
Baltimore-Columbia-Towson, MD
Lewiston-Auburn, ME
Bangor, ME
Springfield, MO
St. Louis, MO-IL
Cape Girardeau, MO-IL
Burlington, NC
Greensboro-High Point, NC
Durham-Chapel Hill, NC
Utica-Rome, NY
Syracuse, NY
Rochester, NY
Ithaca, NY
New York-Newark-Jersey City, NY-NJPA
Canton-Massillon, OH
Cincinnati, OH-KY-IN

Overall
Unemployment Rate
for Population 16
Years and Over

Unemployment Rate
for Workers With a
Disability

6.7

15.6

6.7
6.9
6.8
6.8
6.1
6.3
6.7
6.6
5.2
5.9
6
6.5
6.8
6.7
6.3
6
5.2
6.5
5.2
6.3
5.4
6
6.6
6
6.5
6.6
6.3
4.8

14.4
15.1
16.2
15.1
14.9
15.3
17.4
18.4
14.2
14
14
16.2
15.6
14.9
14.5
14.8
15.2
18.1
14
14.2
14.7
15.2
15.2
14.1
17.2
15.4
14.8
14

6.9

14.5

6.6
5.8

14.9
14

TABLE 3.3

Metropolitan Areas With 5-year Average Unemployment Rates Less Than 7
Percent Overall but Greater Than or Equal to 14 Percent for Workers With Any
Disability

State
OH-PA
OR
OR-WA
PA
PA-NJ
RI-MA
TX
TX
TX
VA
VA
VA
WA
WA
WI
WI
WI-MN
WV-KYOH

Metropolitan Area
Youngstown-Warren-Boardman, OH-PA
Corvallis, OR
Portland-Vancouver-Hillsboro, OR-WA
Pittsburgh, PA
Allentown-Bethlehem-Easton, PA-NJ
Providence-Warwick, RI-MA
Sherman-Denison, TX
Tyler, TX
Waco, TX
Richmond, VA
Roanoke, VA
Blacksburg-Christiansburg-Radford, VA
Walla Walla, WA
Kennewick-Richland, WA
Sheboygan, WI
Janesville-Beloit, WI
La Crosse-Onalaska, WI-MN
Huntington-Ashland, WV-KY-OH

Overall
Unemployment Rate
for Population 16
Years and Over

Unemployment Rate
for Workers With a
Disability

6.9
6.7
6.2
5.7
6.7
6.9
6.5
6.5
5.2
6.3
5.4
5.2
6.1
5.9
4.4
6.4
4.4

14.5
14.1
14.5
14.4
14.1
15.9
14.7
15.8
14.7
14.3
14.6
16.1
14
15.1
15.4
14.1
14.9

6.6

16

29 metro areas in 2013-2017 had unemployment rates of 7 percent or less overall but at
least 14 percent for African American residents. For example, in the Canton-Massillon, OH
metro area, the overall unemployment rate was 6.6 percent, but was 16.9 percent for black workers.
(Table 3.4.)

61

TABLE 3.4

Metropolitan Areas with 5-year Average Unemployment Rates Less Than 7 Percent
Overall but Greater Than or Equal to 14 Percent for Black/African American
Subgroup

State

Metropolitan Area

AR
IA
IA
IL
IL
IN
IN
IN
IN-MI
KY
MA
ME
ME
MI
MN
MN
MN
NY
NY
NY
OH

Jonesboro, AR
Dubuque, IA
Waterloo-Cedar Falls, IA
Springfield, IL
Peoria, IL
Elkhart-Goshen, IN
Terre Haute, IN
Fort Wayne, IN
South Bend-Mishawaka, IN-MI
Owensboro, KY
Pittsfield, MA
Lewiston-Auburn, ME
Bangor, ME
Kalamazoo-Portage, MI
Mankato-North Mankato, MN
Rochester, MN
St. Cloud, MN
Utica-Rome, NY
Syracuse, NY
Rochester, NY
Canton-Massillon, OH
Youngstown-Warren-Boardman,
OH-PA
Altoona, PA
Scranton-Wilkes-BarreHazleton, PA
Pittsburgh, PA
Williamsport, PA
Janesville-Beloit, WI
Green Bay, WI
Wheeling, WV-OH

OH-PA
PA
PA
PA
PA
WI
WI
WV-OH

Overall Unemployment Rate for
Population 16 Years and Over

Black/African
American
Unemployment Rate

5.9
3.9
4.9
6.7
6.6
5.6
6.9
5.9
6.6
6.1
6.8
5.2
6.5
6.6
4.2
3.9
4.5
6.5
6.6
6.3
6.6

14.3
17
19.7
16.4
18.1
15
16.9
15.2
14.3
15.6
19.6
17.7
28.9
15.3
23.3
20.2
17.2
15.2
15.1
14.8
16.9

6.9

17.2

5.2

18.7

6.4

19.2

5.7
6.2
6.4
4.4
6.1

14.1
23
17
18.1
15.8

Six metro areas had unemployment rates of 7 percent or less overall but at least 14 percent
for Hispanic or Latino adults. For example, in the Bloomsburg-Berwick, PA metro area, the
overall unemployment rate was 4.9 percent, but it was 26.9 percent for Hispanic/Latino workers
(who may be of any race). (Table 3.5.)
62

TABLE 3.5

Metropolitan Areas With 5-year Average Unemployment Rates Less Than 7
Percent Overall but Greater or Equal to 14 Percent for Latino/Hispanic Subgroup
State
NY
OH
PA
PA
PA
PA

Metropolitan Area
Utica-Rome, NY
Canton-Massillon, OH
Bloomsburg-Berwick, PA
Erie, PA
ChambersburgWaynesboro, PA
Lebanon, PA

Overall Unemployment Rate for
Population 16 Years and Over
6.5
6.6
4.9
6.5
5.7
5.8

Hispanic or Latino
Unemployment Rate
15.5
20.1
26.9
15.8
14.8
14.2

Fourteen percent unemployment, averaged over five years, is a strikingly high rate that matches
the unemployment rate estimated for the overall labor force in 1937 during the Great Depression.155
Under the proposed rule, some communities would be ineligible for a waiver where some
individuals subject to the time limit are likely to face unemployment rates of this level.
A recent analysis by the Center on Poverty and Social Policy at Columbia University also
demonstrates how sub-populations face significantly higher unemployment rates than the area
average. This analysis looked at over 200 metropolitan areas that could potentially lose waiver
eligibility: areas with unemployment rates 20 percent above the national average, but below 7
percent unemployment rates. As the authors stated:
The median unemployment rate for non-white individuals is closer to 10 percent, and in some
metro areas the unemployment rate is greater than 20 percent for non-white individuals. For
those with a high school education or less, over three quarters of all metropolitan areas have a
higher unemployment rate than the 7-percent floor, which is a particularly dire statistic for the
relevant population of “lower-skilled” workers. Given that close to threequarters of
“ABAWDs” have a high school education or less and close to half are non-white, this
evidence suggests that the proposed rule would disqualify many areas where the individuals
subject to the time limit face substantially higher unemployment rates than 7 percent.
This analysis also looked at complementary labor force metrics, finding that non-white individuals
and workers with a high school education or less had significantly lower employment-population
ratios, with over half of individuals with less education living in areas with employment-population

155

The U.S. Department of Labor estimates 7.7 million persons (or about 14.2 percent) were unemployed in 1937 out of
a labor force of 54.0 million. U.S. Department of Labor, “Labor Force, Employment, and Unemployment, 1929-39:
Estimating Methods,” Technical Note, Monthly Labor Review, July 1948,
https://www.bls.gov/opub/mlr/1948/article/pdf/labor-force-employment-and-unemployment-1929-39-estimatingmethods.pdf.

63

ratios lower than 50 percent. This research shows how the unemployment rate hides the variation
for sub-groups. 156 We strongly encourage FNS to review these data.
Given this evidence, the proposal to restrict states’ ability to waive areas except with very high
overall unemployment rates will have a disproportionate impact on subgroups with rates much
higher than overall unemployment, including groups belonging to protected classes under 7 U.S.C. §
2020(c).
Many “Distressed Communities” Have Relatively Low Unemployment
Another way of considering how areas with relatively low unemployment may provide insufficient
jobs for individuals subject to the time limit is to look at other economic indicators, which provide
other information that could indicate a paucity of jobs. The non-profit organization Economic
Innovation Group calculates a measure of community well-being called the “Distressed Community
Index” that combines seven metrics for the 2012-2016 period: the share of adults ages 25 and up
without a high school diploma; the percent of habitable housing that is unoccupied; the share of the
prime-age (25-64) population that is not employed; the poverty rate; median household income as a
percent of the state’s median households income; the change in employment; and the change in
business establishments.
While these measures do not strictly measure job availability, they do provide a snapshot of
economic health, and present a snapshot of how divergent the economic conditions are, and
recovery from the Great Recession has been, at the local level. For example, most of the job growth
from 2007 to 2016 has occurred in the zip codes in the top quintile of the index. Over two-thirds of
zip codes in that quintile, termed “prosperous,” added jobs since 2007 (adding an average of 1,300
jobs); meanwhile, over two-thirds of zip codes in the lowest quintile, called “distressed,” have fewer
jobs since 2007, and those that did add jobs only added an average of 400 over the period studied.
While about one-fifth of prime-age adults were out of work in “prosperous” zip codes, that share
was double for adults in “distressed” zip codes.157
We looked at counties that had a waiver in 2018 but would not have qualified if the proposed rule
were in place because they did not meet the 7 percent unemployment rate threshold. Of these over
600 counties, over 100 were considered “distressed.” Todd County, South Dakota, for example, had
a 6.6 unemployment rate for the January 2016-December 2017 period. According to this index,
nearly half of the residents in this county lived in poverty in the 2012-2016 period, and close to half of
prime-age adults were not employed. The number of jobs in this county declined by over two-fifths
from 2012 to 2016, and the number of business establishments also declined by over 5 percent.
Another example is Stewart County, Georgia, where over one-third of adults have less than a college
degree, household median income is only about two-fifths of the state’s median income, and over two-

156

Robert Paul Hartley, Christopher Wimer, and Jane Waldfogel, “Limiting States’ Ability to Waive Federal SNAP Work
Requirements: A Closer Look at the Potential Implications,” Columbia University Center on Poverty and Social Policy
Research Brief, Vol.3 No.4, March 25, 2019.
https://static1.squarespace.com/static/5743308460b5e922a25a6dc7/t/5c9a2d43652dea22023c9492/1553608003521/P
overty+%26+Social+Policy+Brief+3_4_+SNAP+Work+Requirements+and+Unemployment.pdf.
157

Economic Innovation Group, “From Great Recession to Great Reshuffling: Charting a Decade of Change Across
American Communities,” October 2018, https://eig.org/wp-content/uploads/2018/10/2018-DCI.pdf.

64

thirds of prime-age adults are not employed. Stewart County also lost both jobs and business
establishments between 2012 and 2016.
While an area’s unemployment rate may mask differences between unemployment rates within
that area, it also may fail to reflect economic conditions more broadly, which may contribute to job
availability for the individuals potentially subject to the time limit. Here again, we are concerned that
FNS did not appear to offer any evidence to support its contention that unemployment rates are a
reliable predictor of jobs available for low-income individuals.
Underemployment Rates Also Higher For Sub-Groups
Along with the impossibility of identifying an unemployment rate that reliably implies the absence
of available jobs, it is also the case that the unemployment rate is an insufficient indicator of labor
market slack. For one, it leaves out those who have left the labor market, in some cases due to slack
labor demand or to personal labor market barriers, including skill deficits and discrimination.
Second, the unemployment rate leaves out a significant group of part-time workers who would
prefer to be full-timers. Such workers are literally under-employed, as they want to work more hours
than their current job offers them. For families with low incomes, working too few hours can put
pressure on family budgets and lead to nutritional hardship.
Table 3.6 shows underemployment rates associated with unemployment rates of 5, 7, and 10
percent for all workers and by race/ethnicity (see note under the table for methodology). At the
proposed rule’s suggested level of 7 percent unemployment, overall underemployment is predicted
to be 12.5 percent, with rates of about 20 and 18 percent for African American and Latino workers,
respectively. In other words, the rule suggests that SNAP waivers should be disallowed in places
where about a fifth of black and Latino workers could be un- or underemployed.
Higher unemployment of course corresponds to even higher underemployment rates, but even at
5 percent unemployment, black and Latino underemployment is around 15 and 13 percent,
respectively.
TABLE 3.6

Predicted Underemployment Rates at Different Unemployment Rates, by
Race/Ethnicity
Predicted Underemployment
Unemployment

All

White

Black

Hispanic

5%

9.1%

7.4%

14.9%

13.0%

7%

12.5%

10.2%

20.4%

17.9%

10%

17.7%

14.4%

28.8%

25.2%

Note: Rates for “all” are derived from regression of u-6 underemployment rate on the overall unemployment rate. Racial
underemployment rates are then derived from ratios of the overall unemployment to underemployment rates by race
using Economic Policy Institute data from 1994-2018

While the “20 percent standard” currently uses unemployment rates, which do not capture aspects
such as labor force participation or part-time work, FNS proposes making these criteria even less
65

responsive to economic conditions by requiring a specific unemployment rate. Given the severe
racial disparities that exist in labor force measures, the fact the Department did not address whether
it considered how an unemployment rate varies in relation to other labor force metrics is another
reason why we cannot comment on how it supported this rule.

G. Unemployment Rate Floor of 7 Percent Fails to Protect Areas During
Recessions
The Department does not discuss how the 7 percent unemployment rate floor would affect
waiver eligibility during an economic downturn, or at any other point in the business cycle besides a
time of relatively low unemployment. An unemployment rate of 7 percent is relatively high for any
area; for example, during the 2001 recession, the national unemployment rate never reached 7
percent.158 An area with a 24-month unemployment rate averaging at least 7 percent signifies that an
area has experienced a prolonged depression. The Department does not acknowledge the
unemployment rate of 7 percent in relation to other economic indicators, but also does not discuss
how the length of time its proposal would require such a high unemployment rate to qualify for a
waiver would affect states entering an economic recession.
When unemployment rates rise rapidly when the economy is entering into a recession and jobs are
quickly declining, individuals likely face many challenges finding or keeping work. By requiring a
very high two-year average unemployment rate, the proposed rule, however, would keep many areas
from qualifying for a waiver during this time. The proposed rule would continue to allow an area to
qualify for a waiver when it qualifies under any of the criteria (including optional criteria) for
Extended Benefits (EB) in the Unemployment Insurance program, which would often allow states
with rapidly rising unemployment to qualify for a waiver.159 Among other criteria, under EB, states
can qualify for a waiver if they meet optional indicators that include a three-month unemployment
rate of 6.5 percent that is at least 110 percent of the same three-month period in either of the
previous two years.160 We agree with the Department’s proposal to continue to allow states to
request waivers when they qualify for EB, as these are times when unemployment rates are high and
rising and individuals likely have difficulty finding jobs. Because the 7 percent unemployment rate
floor is so high and because the NPRM would prohibit states from requesting statewide waivers
based on the 20 percent standard, many states would experience a gap between when their
unemployment rates begin to rise during a recession and when they qualify for a waiver based on
meeting the EB criteria.
For example, consider the experience of two states, South Carolina and Oregon, who would have
been left with a gap between their unemployment rates rising and their qualification for a waiver
based on the EB criteria in the beginning of the Great Recession. During the beginning of the Great
Recession, which officially started in December 2007, these states both had unemployment rates that
158

“National Bureau of Labor Statistics, U.S. Business Cycle Contractions and Expansions,”
https://www.nber.org/cycles.html. Monthly unemployment rates are from BLS.
159
160

NPRM, p.992.

While the criteria mentioned are the criteria to establish additional weeks of Extended Benefits under an optional
trigger (called the Total Unemployment Rate trigger), FNS guidance establishes that states only have to qualify based on
the unemployment measures, even if the state chooses not to provide EB benefits under these criteria.

66

were rising and would have qualified for 2007 statewide waivers under the existing 20 percent
standard (as in, without the 7 percent unemployment rate floor). 161 These states, like others, would
have had to wait several months before they would have qualified for a waiver under Extended
Benefits: South Carolina qualified for a waiver based on Extended Benefits beginning in August
2008 and Oregon qualified for a waiver based on Extended Benefits in November 2008.162 Under
the proposed rule, they could not have requested statewide waivers based on the “20 percent
standard,” as we discuss in Chapter 5, as the rule would only allow statewide waivers based on EB.
Even if they could have requested statewide waivers, however, these states would have been well
into the recession before they qualified for a waiver based on having 24-month unemployment rates
above 7 percent. Therefore, under the proposed rule, at least some areas in both states would have
been ineligible for a waiver at a time when unemployment was high and rising.
The Department repeatedly explains how its rulemaking would prevent states from requesting
waivers when unemployment is low. For example, it states:
Right now, nearly half of ABAWDs live in areas that are covered by waivers despite a strong
economy. The Department believes waiver criteria need to be strengthened to better align
with economic reality. These changes would ensure that such a large percentage of the country
can no longer be waived when the economy is booming and unemployment is low. 163
The Department has clearly considered the role of waivers at a time when national unemployment
is low, though this analysis of course does not take into account the fact that unemployment rates
can vary across the country and even with low unemployment rates, individuals subject to the time
limit may lack available jobs. Even more concerning, however, is that the Department did not
indicate whether it considered the effect of the proposed rule at different parts of the business cycle,
such as entering into a recession, and how climbing unemployment rates affect job availability. (The
Regulatory Impact Analysis also fails to include analyses of the impact of the provision using
historical data to assess how it would fare differently in different economic times.) Without such a
discussion, it is impossible to assess the economic considerations it made in proposing a policy that
would result in many areas remaining ineligible for waivers at a time of rising unemployment.

161

Oregon would not have qualified for a 2008 waiver.

162

In November 2008, Congress passed a temporary expansion of Extended Benefits, called the Emergency
Unemployment Compensation program, that in January 2009 the Bush Administration stated could qualify states for a
1-year statewide waiver. These states would have qualified for statewide waivers immediately then, as both qualified
beginning with the first EUC trigger notice November 23, 2008
(https://oui.doleta.gov/unemploy/euc_trigger/2008/euc_112308.pdf). In addition, Congress passed the American
Recovery and Reinvestment Act (ARRA), which waived all states statewide beginning in April 2009. Because these are
both temporary measures requiring action by Congress, there is no guarantee under the proposed rule that they would
be available in the future, and as we note, they weren’t available in the beginning of the recession in early 2008.
163

NPRM, p.981.

67

H. Department Does Not Explain Claim That Suggested Floor Is “Designed
Specifically for ABAWDS”
The Department proposes a 7 percent unemployment floor for areas to qualify for a waiver under
the 20 percent standard, significantly higher than the floor used by Labor Surplus Areas. The
Department suggests this floor would be more “targeted” towards the specific individuals subject to
the time limit, but provides no evidence to support this assertion. The preamble states:
The Department believes that amending the waiver regulations to include an unemployment
floor is a critical step in achieving more targeted criteria. While the 20 percent standard is
similar to the calculation of an LSA, the Department believes it is appropriate to request
public comment to explore a floor that is designed specifically for ABAWD waivers.164
The Department suggests that having a higher unemployment rate floor than that used by DOL in
its identification of LSAs would be more appropriate for this population than the general LSA floor
of 6 percent unemployment. Evidence shows that the childless adult SNAP participants face labor
market disadvantages, and likely experience higher unemployment rates than their area. This
evidence would recommend against a specific unemployment rate floor, given the difficulty in
assessing a specific rate that would reflect available jobs for this population. For example, a city or
county may have an unemployment rate of 7 percent, but the unemployment rate for childless adult
SNAP participants is 12 or 14 percent. The difficulty in establishing available jobs is especially true
in local areas where local labor market conditions may yield differing opportunities for this
population for different levels of unemployment. For example, even if two areas had the same
unemployment rate, an area where individuals live close to jobs that match their skills will have more
opportunities than an area where there is considerable spatial mismatch.
Given this evidence, if the Department did want a floor “designed specifically for ABAWD
waivers,” it would follow that they would want to explore a floor that is considerably lower than that
used by the Department of Labor in designating Labor Surplus Areas. By suggesting that a higher
floor would be “designed specifically for ABAWD waivers,” the Department is suggesting that
unless unemployment is at a relatively high level, substantially higher than what the Department of
Labor considers sufficiently high in designating Labor Surplus Areas, there are sufficient jobs
available for childless adult SNAP participants. All available evidence suggests the opposite is true:
unemployment has to fall to very low levels before more disadvantaged workers can find jobs. The
Department does not provide any evidence to support its conclusion that a 7 percent unemployment
rate bears any relationship to available jobs for this specific population. By referencing that such a
floor would be “targeted,” the Department indicates that there are considerations that it took when
establishing a floor substantially higher than the LSA floor. Without any discussion of those
considerations, however, it is impossible to follow the Department’s logic and thus meaningfully
comment on it. The robust review we did to understand the availability of jobs as related to the
unemployment rate finds that the unemployment substantially overstates job opportunities for the
individuals subject to the time limit, which would recommend flexibility, not imposing a specific
unemployment rate floor. The Department is claiming an opposite finding without providing any
evidence to support its conclusions.

164

68

NPRM, p.984.

I. Reasoning for Specific Unemployment Rate Floor Not Consistent With
Congressional Intent
The Department’s stated rationale for proposing the 7 percent unemployment rate floor for
waivers for areas with unemployment rates 20 percent above the national average is to ensure that
time limit waivers cover a reduced population compared to current standards. This rationale is
inconsistent with the intent of Congress. Congress intended for FNS to develop criteria to measure
a lack of jobs, and did not specify intended limits to the usage of waivers, provided they reflect
economic conditions.
In the preamble, the Department establishes that the justification for the proposed rule is not
based on an analysis of the relationship of the unemployment rate floor to job availability for
childless adult SNAP participants. The Department states it instead weighed the effect of the
proposal on the breadth of waiver coverage:
The Department seeks to establish a floor that is in line with the Administration’s effort to
encourage greater engagement in work and work activities. The Department believes that the
7 percent floor for the 20 percent standard would strengthen the standards for waivers so that
the ABAWD work requirement would be applied more broadly and fully consider the “lack of
sufficient jobs” criteria in the statute.165
The Department therefore states that the goal of the NPRM is to apply the time limit to more
childless adults. The Department states that applying the time limit to more unemployed adults
would “fully consider the ‘lack of sufficient jobs’ criteria in the statute,” but does not explain how
restricting areas and would better reflect employment opportunities for this population. Moreover,
this reasoning is completely contrary to Congressional intent, which was to allow states to waive
areas with insufficient jobs without imposing limits on the share of areas covered by waivers by state
or nationally. We are confused as to why FNS believes it has the authority to purposefully expose
more people to the time limit as a rationale.
Available evidence suggests that restricting waivers to only areas with very high unemployment
would actually make waivers less likely to reflect available jobs for this population, given that it
would exclude from eligibility many areas where these individuals lack jobs (such as an area with an
unemployment rate of 6.7 percent, but unemployment rates well above 10 percent for childless adult
SNAP participants).
Congress intended for the Administration to develop economic criteria to measure job
opportunities for childless adult SNAP participants. Congress did not propose any measure to limit
waivers based on the number or share of individuals subject to the time limit. If Congress had
intended for waivers to be limited so that a specific share of childless adults live in an area with a
waiver, it could have written legislation to achieve that goal. For example, in the 2017 Tax Cuts and
Jobs Act (P.L. 115-97), Congress created Opportunity Zones, which are low-income Census tracts
designated by the chief executive of a state that are eligible for tax incentives for investment. While
Congress created several criteria to identify areas that are nominated, such as the poverty rate or
median family income, Congress also limited the number of potential eligible Opportunity Zones
165

NPRM, p.984.

69

each state is allowed to designate based on the total number of low-income communities in the
state. For example, in areas with over 100 low-income census tracts, no more than 25 percent of the
number of those low-income tracts can be designated as Opportunity Zones.166 This law serves as
an example of one way that Congress can establish criteria to limit a particular sub-state designation,
if that is indeed its intent. In establishing waiver criteria in the welfare reform law, Congress did not
establish any mechanism to limit the scope of waivers, which it could have done by various means,
had that been its goal. Instead, the law allows for the DEPARTMENT to develop measures to
evaluate available jobs for childless adult SNAP participants, which are not limited in scope. The
number of areas lacking jobs can expand or contract with economic conditions, and Congress
allowed for states to waive areas in response to these changing economic conditions. Congress has
not changed this approach since the original 1996 legislation.
Furthermore, the Department uses provisions of the House-passed version of H.R. 2 to support
its proposed unemployment rate floor, ignoring that Congress ultimately rejected such provisions. In
providing support for the 7 percent unemployment rate floor, the preamble states, “Furthermore,
this aligns with the proposal in the Agriculture and Nutrition Act of 2018, H.R. 2, 115th Cong.
§ 4015 (as passed by House, June 21, 2018).”167 While this bill did contain a similar provision, the
Senate bill did not include this provision, and the Conference Committee chose to align with the
Senate version, passing both chambers without any restrictions on waivers. While the Department
may consider Congressional bills, offering this as support while ignoring that these provisions were
ultimately excluded from the final bill, the Department is offering an incomplete interpretation of
Congressional intent.
Department Provides Little Explanation
to Support Stated Goal to Limit Waiver Coverage
The Department establishes that its intent is to limit waiver coverage and therefore expand the
time limit to the extent possible. The Department does not explain how this goal is related to the
intent of the statute to identify areas that lack jobs for childless adults. Even if it had clarified how
its stated goal related to the underlying statute it is interpreting, the Department also does not
provide clear explanation of the assumptions used in determining the metric used repeatedly to
support its conclusions, the share of “ABAWDs” living in a waived area. Without any explanation of
the analysis used to understand what it believes the relationship between the unemployment rate
floor and waiver coverage is, and how waiver coverage relates to the “insufficient jobs” law, the
Department has limited our ability to comment on these specific assertions.
The Department explains that the principal criteria it considered when proposing the specific 7
percent unemployment rate floor was not based on an economic argument about the relationship
between the general unemployment rate and jobs available for disadvantaged individuals, but rather
a desire to limit waivers of the time limit:
As stated previously, the Department seeks to make the work requirements the norm rather
than the exception to the rule because of excessive use of ABAWD time limit waivers to date.
Using the proposed rule’s 7 percent floor for this criterion and eliminating waiver approvals
166

Congressional Research Service, “Tax Incentives for Opportunity Zones: In Brief,” R45152, November 20, 2018,
httlps://fas.org/sgp/crs/misc/R45152.pdf.
167

70

NPRM, p.984.

based on an LSA designation (as well as utilizing the proposed limit on combining areas
discussed below), an estimated 11 percent of ABAWDs would live in areas subject to a waiver.
Currently, approximately 44 percent of ABAWDs live in a waived area. The Department
views the proposal as more suitable for achieving a more comprehensive application of work
requirements so that ABAWDs in areas that have sufficient number of jobs have a greater
level of engagement in work and work activities, including job training. 168
The Department suggests that not only would limiting waivers be preferable to keeping the
current regulations, but also suggests that the more the rules result in limited waivers, the more
SNAP participants will be led towards self-sufficiency. (The Department also makes the claim that
current waiver coverage is “excessive without providing explanation of the criteria used to judge
appropriate waiver coverage.) It states that a 7 percent floor, which it indicates would result in a
decline from 44 percent of “ABAWDs” living in a waived area to 11 percent, would be “more
suitable” than the current rules. In the proposed rule, the Department asks for feedback on
alternative floors to the 7 percent unemployment rate floor of 6 percent or 10 percent (see Section I,
below), explaining how the higher the unemployment rate floor, the more preferable according to
their standards:
Based on the Department’s analysis, nearly 90 percent of ABAWDs would live in areas
without waivers and would be encouraged to take steps towards self-sufficiency if a floor of 7
percent was established. In comparison, a 6 percent floor would mean that 76 percent of
ABAWDs would live in areas without waivers and a 10 percent floor would mean that 98
percent of ABAWDs would live in areas without waivers. A higher floor allows for the
broader application of the time limit to encourage self-sufficiency.169
The Department therefore believes that expanding the time limit to more people is a desirable
outcome. The Department states that the greater the unemployment rate threshold, the fewer
childless adults will live in waived areas, suggesting that its goal is to minimize waiver coverage to the
extent possible. Setting aside the issue that there is no evidence that applying the time limit more
broadly encourages self-sufficiency, which we address comprehensively in Chapters 6 and 11, the
Department leaves several unanswered questions with regards to how this rulemaking will further
the intent of the law in defining areas with insufficient jobs:
• The

Department does not explain how imposing a higher unemployment rate floor would
better approximate a lack of jobs. The purpose of the regulation is to define areas that lack “a
sufficient number of jobs to provide employment” to childless adult SNAP participants. To
interpret this regulation, it would follow that the specific waiver criteria the Department
develops would best allow states to identify areas lacking jobs, and best enable the
Department to approve those waivers based on consistent criteria. Operating under the
framework that these regulations interpret the statute, the appropriate amount of waiver
coverage is related to the share of this population facing limited employment opportunities
(i.e., a lack of sufficient jobs) in their area. For example, if about one-third of counties did not
have sufficient job opportunities for childless adults, then about one-third of counties would
be eligible for a waiver, if there were a way to perfectly capture job availability for this

168

NPRM, p.984.

169

NPRM, p.984.

71

population. If this share rises during a recession to 75 percent, then the share of the country
eligible for a time limit waiver could also rise accordingly.
The Department is therefore proposing an alternative interpretation of the statute, though it is
not clear what this interpretation is or what the authority it has to drastically change this
interpretation. The Department does not explain whether it believes that there is an economic
argument supporting limiting waivers, or instead if it believes that the goal of limiting waivers
is separate from establishing areas with insufficient jobs, and if so, the authority under which it
can establish new criteria for waivers that are not found in the statute. Without more
explanation as to why its goal for limiting waivers is relevant to this rulemaking, it is difficult
to assess the merits of the underlying arguments.
• The

Department does not explain whether there are any parameters to its stated goal of
limited waivers, and under what criteria it judges the appropriate level of waiver coverage. In
regard to the proposed 6 percent unemployment rate floor, the preamble states that “the
Department is concerned that too many areas would qualify for a waiver of the ABAWD time
limit with a 6 percent floor and that too few individuals would be subject to the ABAWD
work requirements.”170 The language of “too many” or “too few” implies that there is a
desired level of waiver coverage, and that the waiver coverage that they estimate a 6 percent
unemployment rate floor would yield (24 percent of “ABAWDs” living in waived areas) is too
high. The Department therefore has implicit criteria by which it is judging an appropriate
share of individuals living in a waived county that it does not explain. While it is not clear how
the share of “ABAWDs” living in a waived area is relevant to the rulemaking, even if it were,
the Department does not allow commenters the ability to provide input on this metric without
establishing the criteria it is using to judge the appropriate level.

• Relatedly,

the Department does not explain if it believes that limiting waivers would be an
equally important goal during an economic recession, when the share of areas with limited
jobs would expand considerably. Again, if the Department states that 24 percent of
“ABAWDs” in waived areas is “too high,” would that also be true during an economic
recession, if most areas of the country offered few jobs to those individuals, and a majority of
childless adults lived in an area covered by a waiver? Without explaining how its stated goal of
limiting waivers is related to assessing the economic conditions in an area, it is impossible to
tell if the Department considers this goal to be a relative goal (as in, it believes it is acceptable
to expand the time limit in response to higher unemployment), or if it believes there is a
desired percentage of “ABAWDs” living in waived areas regardless of economic conditions.

• Finally,

the Department’s calculation of the share of “ABAWDs” living in a waived area does
not take several important factors into account.
o When the Department calculates the share of what it terms “ABAWDs” living in waived
areas under the different scenarios it lays out, it is not clear if it is considering how this
share will change as the overall denominator changes and what other assumptions are
embedded in its analysis. In a time when fewer areas are waived, childless adults will be
more disproportionately concentrated in waived areas, as they will lose benefits in nonwaived areas. At any time, childless adults include a combination of participants who are
living in waived areas; exempt from the time limit (but who the data does not allow us to
identify as exempt); in their first three months of SNAP participation; or working or

170

72

NPRM, p.984.

complying with the requirements through training. Other variables at the local level, such
as state or county implementation of the time limit, the composition of childless adults
(for example, in some areas, there may be proportionately more disadvantaged
individuals), and the amount of job or training opportunities that are suitable for childless
adults, will also affect how likely childless adults are to continue participating in SNAP in
non-waived areas. Therefore, what share of childless adult SNAP participants living in
waived areas is not just a function of the share of counties covered by a waiver, but also
how they are distributed among those counties. It is not clear if the Department is using
this statistic as a proxy for measuring overall waiver coverage, or if it is meant to convey
an analysis modeling these dynamic variables.
Consider a simplified example. Here, we will look to see how the distribution of childless
adults living in certain waived areas changes as overall nationwide waiver coverage
changes, using eight states, Guam and Virgin Islands that had statewide waivers in 2017,
and 13 states that had statewide waivers at least from 2010 through 2013 but had dropped
them by 2017. (This illustrative analysis therefore is not looking at childless adults living
in all waived areas, but rather choosing to look at states when they had a statewide waiver
or no waiver in 2017 to simplify the analysis.) In 2010, almost all areas of the country
were waived in the aftermath of the Great Recession, which prompted Congress to
include a provision in the Recovery Act (P.L. 111-5) that waived the time limit
nationwide. (About five states continued to implement the time limit in parts of their
state, but they offered work opportunities for individuals subject to the time limit.) About
89 percent of the general population lived in an area that was waived.171
Because of this widespread waiver coverage, the share of childless adult SNAP
participants living in waived areas would be expected to be very similar to the share of all
SNAP participants living in those areas. To the extent that this distribution differed, it
would likely be due to compositional differences, such as areas with greater shares of
children or elderly individuals. Waiver coverage would not be the main driver of
differences, as waiver coverage was similar nationwide. Indeed, in 2010, about 20 percent
of the total U.S. population lived in states that had statewide waivers continuously
through 2017, and a slightly smaller share, 17 percent, of SNAP participants lived in those
states. (This may be because these states’ populations had slightly smaller shares of
individuals with income below SNAP’s income eligibility limits, reduced access to SNAP,
or increased barriers for eligible people, or other reasons.) The share of all adults ages 1849 in childless households who lived in those states in 2010 was similar to the distribution
of SNAP participants; about 18 percent of childless adults lived in those eight states.
Similarly, for states that had earlier had statewide waivers, but dropped them by 2017, the
share of childless adults living in those states was similar to the share of SNAP
participants living in those states in 2010. Those states had about 24 percent of the total
U.S. population, but a slightly higher share of SNAP participants (26 percent), and a
slightly higher share of childless adults (28 percent). (Table 3.7.)

171

Center on Budget and Policy Priorities, “States Have Requested Waivers from SNAP’s Time Limit in High
Unemployment Areas for the Past Two Decades,” https://www.cbpp.org/research/food-assistance/states-haverequested-waivers-from-snaps-time-limit-in-high-unemployment.

73

TABLE 3.7

Distribution of Childless Adults Living in Waived Areas Changes as Overall Waiver
Coverage Changes

Total
Population
(millions)

Share of
Total
Population

SNAP
Participants
(program
data,
millions)

Share of
SNAP
Participants

SNAP
Participants
Ages 18-49,
Without
Disabilities,
in Childless
Households
( millions)

Share of
Participants
Ages 18-49,
Without
Disabilities,
in Childless
Households

Fiscal Year 2010: 89% of U.S. Population Lives in Waived Area
States with
statewide
waivers in at
least 2010-2013
and no waivers
2017
States with
statewide
waivers in at
least 2010-2013
and 2017
Total all states

72.9

24%

10.4

26%

1.1

28%

62.1

20%

6.7

17%

0.7

18%

309.3

100%

40.3

100%

3.9

100%

Fiscal Year 2017: 36% of U.S. Population Lives in Waived Area
States with
statewide
waivers in at
least 2010-2013
and no waivers
2017
States with
statewide
waivers in at
least 2010-2013
and 2017
Total all states

77.0

24%

10.6

25%

0.6

19%

64.7

20%

8.3

20%

0.9

29%

325.1

100%

42.1

100%

3.2

100%

Notes: The states with statewide waivers from the time limit in 2013 but no waivers at all in 2017 (which represented
about a quarter of SNAP participants in 2013) were Alabama, Arkansas, Florida, Indiana, Iowa, Kansas, Maine,
Mississippi, Missouri, North Carolina, Oklahoma, South Carolina, and Wisconsin. The states with statewide waivers in
both 2013 and 2017 (which represented about 20 percent of SNAP participants) included Alaska, California, District of
Columbia, Illinois, Louisiana, Nevada, New Mexico, Rhode Island, Guam, and Virgin Islands.
Sources: CBPP analysis of FY2010 and FY2017 SNAP household characteristics data; USDA program data; Census
population estimates as of July 1st 2010 and 2017

By 2017, the share of waived areas had declined dramatically as the economy improved.
About 36 percent of the U.S. population lived in an area that was waived.172 Given that
172

74

Ibid.

the eight states who continued to waive the time limit statewide were the only states
remaining with statewide waivers and overall waiver coverage was much lower, it would
be likely that childless adults would be disproportionately living in states that continued to
be waived statewide, as they would be more likely to be subject to the time limit in other
states with partial or no waivers. Similarly, we would expect the share of childless adults
living in states without waivers to have declined relative to the share of all SNAP
participants in those states. As table 3.7 shows, the share of childless adult SNAP
participants who lived in states with statewide waivers (29 percent) was about 50 percent
greater than the share of overall SNAP participants who lived in those areas (19 percent).
This share also represents a significant increase from the 2010 share of 18 percent. We
see the opposite trend for states that had no waiver by 2017: while in 2010, states that had
no waivers by 2017 had a slightly greater share of SNAP participants and childless adults
than they did of overall U.S. population, by 2017, proportionately fewer childless adults
lived in those states. While about 25 percent of SNAP participants lived in those states in
2017, only 19 percent of childless adults lived in states with the time limit statewide in
2017.
The distribution of childless adults essentially flipped between these two groups of states
between 2010 and 2017, as the overall number of areas waived declined and childless
adults became more concentrated in states with waivers and much less so in areas without
waivers.
It is not clear if the Department incorporated these factors into its analysis, or assumed a
more static relationship. The complexity of analyzing childless adults in waived areas
raises the question of why the Department chose this particular statistic to measure
waiver coverage, particularly given that it provided little explanation of some of these
assumptions behind this calculation. Without more information, it is difficult to evaluate
how relevant this statistic is to their overall point, which is to measure childless adults in a
waived area as a measure of the proposed rule’s effect.
o Similarly, as we explain in our comments on the Regulatory Impact Analysis in Chapter
11, the term “ABAWD” lacks specificity, particularly when describing changes in waiver
coverage. The data do not allow us to tell which of the larger group of adults without
dependent children, ages 18 to 49, without disabilities, might be exempt from the time
limit, so many of these adults are not subject to the time limit. Others are only subject to
the time limit if they live in an area without a waiver. When fewer areas are waived, more
of these adults will be subject to the time limit and lose benefits, and the overall number
of these adults participating in SNAP will decline. Therefore, when the Department
describes how 11 percent of “ABAWDs” would live in a waived area with a 7 percent
floor and 24 percent would live in a waived area with a 6 percent floor, it is unclear if the
Department is considering the decline in overall childless adults participating in SNAP
that would occur with the reduction in waivers.
o As we also discuss in our comments on the Regulatory Impact Analysis in Chapter 11, it
is unclear why the Department used the number of participants in non-public assistance
households in the FNS-388 form as a proxy for childless adults in estimating the share of
“ABAWDs” living in waived areas under different scenarios.

75

Instead of providing evidence that the Department’s proposal will interpret the statute in a more
effective manner by improving on the measurement of jobs available for low-income childless adult
SNAP participants, the Department instead uses a confusing and unexplained metric to support its
proposal, the share of childless adults living in an area covered by a waiver. This metric is seemingly
unrelated to the intent of the statute. Because the Department provided little evidence to support
the assumptions made in estimating this metric and to explain why it is relevant to the underlying
law, it is impossible to provide more detailed discussion.

J. Proposed Alternative Unemployment Rate Floors Also Problematic
In addition to the Department’s preferred unemployment rate floor of 7 percent, the Department
also sought comment on unemployment rate floors of 6 or 10 percent. Both of these proposals are
flawed, demonstrating why selecting a specific unemployment rate floor to reflect available jobs for
this population is a misguided approach.
Department’s Proposed 6 Percent Floor Demonstrates Why No Specific Unemployment
Rate Floor Is Appropriate
The Department explains that a 6 percent unemployment rate floor would both be consistent with
Labor Surplus Areas and bears a similar relationship to what it erroneously considers to be the
natural rate of unemployment, stating “As previously noted, the ‘‘natural rate of unemployment’’
generally hovers around 5 percent, meaning that 20 percent above that rate is 6.0 percent.”173 The
Department therefore at least provides evidence that is somewhat more consistent with current
standards such as relying on Department of Labor criteria, though undermines its 7 percent
unemployment rate proposal, for which it does not provide any such evidence.
This unemployment rate floor would still exclude many areas where childless adult SNAP
participants face considerably higher unemployment or underemployment rates and where they will
not have access to jobs, however. As stated above, even at 5 percent unemployment rates, black and
Latino workers nationally face unemployment rates of 9.6 percent and 7 percent, respectively, and
underemployment rates of 14.9 percent and 13 percent, respectively. Some local metropolitan areas
had 2013-2017 average unemployment rates below 6 percent, but unemployment rates for subpopulations well above 14 percent, the unemployment rate of the Great Depression. (See Tables in
Section E above.) For example, Monroe, MI, had an unemployment rate of 5.9 percent, but workers
without a high school degree faced unemployment rates of 16.7. In Fort Wayne, Indiana, the area
unemployment rate was 5.9 percent, but workers with a disability had unemployment rates of 14
percent. In the Pittsburgh, PA metro area, while the unemployment rate was 5.7 percent, the
unemployment rate among African American workers was 14.1 percent. Again, these figures
demonstrate why it is impossible to set a specific unemployment rate threshold at which it can be
reasonably assured that childless adults subject to the time limit can readily find a job with steady
hours. Evidence shows that this group is likely to face unemployment rates much higher than their
local area, and the unemployment rate floor is an inadequate proxy to measure jobs available to
them.

173

76

NPRM, p.984.

The Department’s suggestion that a 10 percent unemployment rate floor is in any way a
reasonable proposal highlights many of the internal inconsistencies and inadequate explanations in
this proposed rule.
10 Percent Floor Inconsistent with Congressional Intent and Based on Obscure and
Inconsistent Reasoning
Congress clearly designated a 10 percent unemployment rate as one way for a state to qualify for a
waiver, and a second, more flexible and targeted criterion of “insufficient jobs” as an alternative to
demonstrating a 10 percent unemployment rate. Had Congress intended for 10 percent
unemployment to be the only way for a state to qualify for a waiver, it would not have included an
alternative. This proposal therefore runs afoul of Congressional intent.
The Department also ignores the LSA standard’s 10 percent ceiling, demonstrating how its
reasoning is inconsistent. As explained elsewhere, the Department picks and chooses when and how
it will aim to be consistent with the DOL’s approach in assessing unemployment. The Department
of Labor clearly considers 10 percent to be a sufficiently high level of unemployment that an area
with 10 percent unemployment over 24 months demonstrates a surplus of labor. The Department
ignores this fact by proposing a 10 percent unemployment rate floor for the 20 percent standard,
while also citing the LSA standard as support for the unemployment rate floor concept in general.
As with the proposed 7 percent floor, the Department suggests that the 10 percent standard
would achieve its goal of greatly curtailing waiver coverage, which as explained above, is a goal not
aligned with the intent of the underlying statute and for which it does not offer a transparent
rationale. The Department states, “the Department estimates that a 10-percent floor would reduce
waivers to the extent that approximately 2 percent of ABAWDs would live in waived areas.”174
While the Department may consider reducing the population of “ABAWDs” living in waived areas a
priority, the Department provides little explanation of how this priority relates to the underlying
statute and identifies areas lacking jobs for childless adults. It also provides little transparency with
regards to the assumptions used in estimating the effects of these unemployment rate floors on the
population living in waived areas.
10 Percent Floor Would be Duplicative of Existing 10 Percent Criteria
This proposal would also be largely duplicative of existing criteria. The Department does discuss
how the time frame used would be different from the existing regulations regarding waivers based
on 10 percent unemployment rates: “the 10-percent unemployment floor would be attached to the
20 percent standard, which would mean an area would require an average unemployment rate 20
percent above the national average for a recent 24-month period and at least 10 percent for the same
period; the other similar, but separate standard requires an area to have an average unemployment
rate of over 10 percent for a 12-month period.”175 Ten percent unemployment is extremely high at
the national level. Only during a deep recession, such as immediately following the Great Recession,
would 20 percent above the national average be close to or above 10 percent, which would require
the national average to be at least 8.4 percent for a 24-month period. Since the BLS began tracking
174

NPRM, p.984.

175

NPRM, p.984.

77

monthly unemployment statistics since 1948, out of 830 total 24-month periods there have been
only 64 24-month periods when the national average would have met this standard. There were 28
periods from November 1980 through January 1985, during and following the 1981-1982 recession,
and 36 24-month periods around the Great Recession of 2007-2009 and its long, slow recovery,
from May 2008 through March 2013.
For most of the time barring these prolonged economic crises, then, this regulation would simply
extend the time frame for demonstrating 10 percent unemployment, essentially eliminating the 20
percent standard most of the time. There may be some areas that are recovering from a deep
economic shock that have more recent unemployment rates just below 10 percent, but
unemployment over the past two years high enough over that rate to nudge the average up above 10
percent. Because 10 percent is such a high level of unemployment, however, it is unlikely that many
areas would qualify that would not have otherwise qualified under the 10 percent criterion.
For example, we analyzed all 12-month time periods that a state could use to examine waiver
eligibility based on either having a 12-month unemployment rate over 10 percent or a 24-month
unemployment rate 20 percent above the national average but at least 10 percent from 2008 to 2019.
With the exception of the years capturing peak unemployment rates immediately following the Great
Recession, from 2014 to 2016, when capturing a longer time frame allowed for more months during
peak unemployment, only a handful of counties would qualify under the 24-month average but not
the 12-month average. (Even during those years from 2014-2016, fewer than 150 counties, or less
than 5 percent of counties, would have qualified under the 24-month but not the 12-month
standard.)
It is not clear if the Department more fully considered the practical differences between these
measures, such as comparing how many areas would have qualified in past years based on having 12
months of 10 percent unemployment or 24 months of 10 percent unemployment. Its rationale for
essentially replacing the 20 percent standard, which measures high unemployment relative to the
national average, with additional criteria for the 10 percent standard, is therefore not transparent.
Department’s Alternative Floors Highlight Arbitrary Choice of 7 Percent Floor
The Department’s discussion of alternate unemployment rate floors also demonstrates how its
proposed floor of 7 percent is an arbitrary figure. When proposing the 6 percent and 10 percent
floors as alternatives to the 7 percent floor, the Department gives little discussion of the relevance to
these floors to the underlying statute, which is to identify areas where individuals subject to the time
limit do not have access to enough jobs. For example, the Department could have provided
economic evidence that indicates how these specific unemployment rates relate to job availability for
childless adult SNAP participants. The only such discussion Department includes is when it explains
a relationship between the natural rate of unemployment and the 6 percent floor by stating that it
the 6 percent floor is roughly 20 percent above 5 percent, which the Department inaccurately states
is consistent with the “natural rate of unemployment” concept. 176 The Department therefore uses no
economic discussion to support its proposed 7 percent floor but gives some discussion to explain
the 6 percent floor, which is its alternate proposal. The discussion of the relationship of the 6
percent floor to the natural rate of unemployment therefore undermines the Department’s proposal
176

78

NPRM, p.984.

for the 7 percent floor, as it demonstrates that the Department either did not consider any economic
evidence, or did not provide any evidence to allow us to meaningfully comment.
The only discussion the Department gives to justify choosing the 7 percent floor (or to support
the alternative of 10 percent) is to limit the share of the population covered by a waiver, which as
discussed above, is not what Congress intended when creating the waiver authority. Without any
discussion to explain how the 7 percent unemployment rate is an appropriate measure of available
jobs for the individuals subject to the time limit, the 7 percent floor appears to be a completely
arbitrary choice.
The Department proposes alternative unemployment rate floors for the “20 percent standard.”
These proposed floors would also be problematic, as would any specific unemployment floor,
because it is impossible to demonstrate that an area with an unemployment rate below a specific
threshold lacks jobs for the individuals subject to the time limit.

K. Conclusion: Proposal for Unemployment Rate Floor Is Deeply Flawed
The Department proposes to change one of the most frequently used standards for waiver
approval, the “20 percent standard,” to require a minimum unemployment rate. With this proposal,
a state could request a waiver with an unemployment rate 20 percent above the national average for
a 24-month period only if it was above this unemployment rate floor. The Department proposed an
unemployment rate floor of 7 percent, but also sought input on proposed floors of 6 or 10 percent.
The Department provides little economic evidence to support the claim that imposing this floor
would better interpret the statute, which allows states to request waivers for areas that lack “a
sufficient number of jobs to provide employment” for the individuals subject to the time limit. The
Department instead appears to work backwards from its stated goal of applying the time limit to
more SNAP participants by limiting waivers, and proposes an unemployment rate floor as a means
of achieving this goal. The Department does not explain how this goal relates to the purpose of the
law it is interpreting, or how the specific floors it proposes would better reflect available jobs for
participants.
Research shows that the childless adults who may be subject to the time limit if they are not
exempt or living in a waived area tend to have many characteristics that are associated with higher
unemployment rates. The majority have lower levels of educational attainment, they are
disproportionately people of color, many have health conditions or barriers such as unstable housing
that limit their ability to work, and many likely experience spatial mismatch and lack access to the
jobs that are available in their communities. Because of these features, it is difficult to find a labor
force metric that accurately portrays the job opportunities available to these individuals. Current
regulations allow states to show that an area has elevated unemployment compared to the national
average. The current “20 percent standard” therefore already disqualifies many areas with
unemployment similar to or below the national average where there are not enough jobs for these
individuals to find employment that are not reflected in the unemployment rate.
While current regulations could be improved, this proposal would substantially worsen the
existing inadequacies. The proposal would require an area has an unemployment rate consistent with
weak labor markets for the overall labor force, 7 percent, to qualify for a waiver. Given that the
unemployment rates for the group of these individuals are likely substantially higher than their area,
79

this proposal would disqualify many areas where individuals face much higher rates than 6 or 7
percent unemployment. The proposal would not align with Congressional intent, which purposefully
did not specify a specific unemployment rate to signify that an area lacks jobs in recognition that the
unemployment rate cannot capture job availability for this specific group.
States frequently request waivers based on the current “20 percent standard,” given that data are
readily available and consistent across states, and FNS standards for approval are transparent and
consistently applied. While the current standard falls short of accurately reflecting jobs available for
this population, we believe the proposed unemployment rate floor would be inconsistent with the
intent of Congress and would make the current criteria substantially less effective at measuring
insufficient jobs. We therefore urge FNS to drop the unemployment rate floor proposal, keep the 20
percent standard as it is, and explore metrics based on evidence that would more effectively reflect
jobs available to the population in recognition of the likely substantially higher unemployment rates
they face.

80

Chapter 4. Dropping Several Key Criteria From Waiver
Criteria Is Inconsistent With the Statute
The NPRM proposes several significant changes to longstanding SNAP policy that would restrict
states to one limited measure of labor market conditions, the unemployment rate, when providing
evidence of lack of sufficient jobs. It would eliminate the ability of states to use valuable, readily
available labor market indicators, such as a low and declining employment-to-population ratio, a lack
of jobs in a declining industry, or an academic study or other publication(s) that describes an area’s
lack of jobs. The NPRM fails to provide reasons for limiting states’ ability to use widely accepted
labor market measures to support requests for waivers, to discuss the implications of relying on a
single measure of labor market conditions, or to acknowledge the valuable information provided by
other measures. Without knowing what evidence justifies this change in longstanding policy and an
adequate discussion of alternative methods for assessing labor market conditions, it is impossible to
assess the potential impact of the changes on SNAP participants and their ability to achieve selfsufficiency. The sections below provide an overview of existing statutes, regulations, and guidance,
address limitations of the general unemployment rate, and discuss alternative measures of labor
market conditions.

A. Current Statute, Regulations, and Guidance Acknowledge That There Is
No Perfect Measure of an Area’s “Lack of Sufficient Jobs”
According to the statute, a state may waive the applicability of the work requirement “to any
group of individuals in the state if the Secretary makes a determination that the area in which the
individuals reside has an unemployment rate above 10% or does not have a sufficient number of
jobs to provide employment for the individuals.” The statute does not limit the type of information
that can be used to support a claim of lack of sufficient jobs.
According to the current rule (7 C.F.R. § 273.24 (f)(2)(ii)), states are not limited to using
unemployment rates to support a claim of lack of sufficient jobs. States may provide evidence that
an area has a low and declining employment-to-population ratio, has a lack of jobs in declining
occupations or industries, or is described in an academic study or other publications as an area
where there are lack of jobs. In the preamble to the current rule, FNS stated that “State agencies
could submit requests with no limit on the supporting documentation, and every request would be
weighed on its own individual merits.” The final rule included a non-exhaustive list of the kinds of
information a state agency may submit to support a claim of “lack of sufficient jobs.”
Below are excerpts from FNS guidance and rulemaking that give states flexibility in using other
types of data to provide evidence of a lack of sufficient jobs, acknowledging that unemployment
rates may not adequately capture the local labor market prospects of individuals subject to the time
limit.
• December

3, 1996 guidance: According to FNS, the statute “recognizes that the
unemployment rate alone is an imperfect measure of the employment prospects of individuals
with little work history and diminished opportunities. It provides states with the option to

81

seek waivers for areas in which there are not enough jobs for groups of individuals who may
be affected by the new time limits.”177
“Lack of jobs due to lagging job growth. Job seekers may have a harder time finding
work in an area where job growth lags behind population growth. A falling ratio of
employment-to-population may be an indicator of an adverse job growth rate. When the
number of jobs in an area grows more slowly than the working age population, the local
economy is not generating enough jobs.
The employment-to-population ratio complements measures of unemployment by
taking into account working age persons who may have dropped out of the labor force
altogether. The ratio can be computed by dividing the number of employed persons in
an area by the area’s total population. A decline in this ratio over a period of months
could indicate an adverse job growth rate for the area….
Lack of jobs in declining occupations or industries. “Employment markets dominated by
declining industries could lead to the presence of large numbers of people whose current
job skills are no longer in demand. This can be especially true in smaller, rural areas
where the loss of a single employer can immediately have a major effect on local job
prospects and unemployment rates.”
• 1999 proposed

rule: In the preamble, FNS noted that “the legislative history does not
provide guidance on what types of waivers the Department should approve under this
standard, and there are no standard data or methods to make the determination of the
sufficiency of jobs. States requesting waivers are therefore free to compile evidence and
construct arguments to show that in a particular area, there are not enough jobs for individuals
who are affected by the time limit.”178 FNS reiterated that one possible indicator that an area
has insufficient jobs is a falling ratio of employment-to-population, but that “no particular
approach is required.”

• August 2006 guidance:

“Waivers may also be submitted based on the following criteria: 1)
areas having a low and declining employment-to-population ratio; 2) areas having a lack of
jobs in declining occupations or industries; 3) areas described in an academic study or other
publications as an area where there is a lack of jobs. The state may submit whatever data it
deems appropriate to support requests based on this data. FNS will evaluate the data and
determine if it is acceptable to justify a waiver.”179

• December

2016 guidance: FNS provided additional detail on “other potential types of
waiver requests” beyond those based on the LSA designation or unemployment rates:180
A low and declining employment-to-population ratio. Employment-to-population (ETP)
ratio can be a meaningful economic indicator for an area where the unemployment rate

177

USDA, “Guidance for states Seeking Waivers for Food Stamp Limits,” December 3, 1996.

178

Food Stamp Program: Personal Responsibility Provisions of the Personal Responsibility and Work Opportunity
Reconciliation Act of 1996; 64 Federal Register 242, p. 70946 (December 17, 1999) (to be codified at 7 C.F.R. pts. 272
and 273).
179
180

USDA, “Guidance on Requesting ABAWD Waivers,” August 2006.

USDA, “Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents
(ABAWD),” December 2, 2016.

82

may not provide a complete picture of the labor market due to people leaving the
workforce – but demographic changes, such as an aging population, can influence these
data. Historically, low and declining ETP data have been used successfully to waive
Indian reservations or tribal lands where unemployment statistics and other economic
data are limited or unavailable. ETP data can also be used to request waivers for nontribal areas, such as counties, but it is uncommon because BLS unemployment data is
readily available for these areas. Therefore, FNS has approved requests based on ETP
data for non-tribal areas, such as rural counties, on a limited basis when the state has
demonstrated that the area’s ETP ratio is:
o Low: at least 1 percentage point below the national average for the most recent
year of the reference period;
o Declining: best demonstrated by a decline year after year;
o Covering at least a four-year reference period, ending no earlier than two years
prior to the year in which the waiver is effective; and
o Complemented by a recent 24-month unemployment rate at least 10 percent
above the national average in the requested area.
A lack of jobs in declining occupations or industries. Employment markets dominated by
declining industries could impact large numbers of people whose current job skills are no
longer in demand. This can be especially true in smaller, rural areas in which the loss of a
single job provider, such a major manufacturing plant or mining industry, can have a major
effect on local job availability. The state might consider providing studies, reports, or other
analysis from credible sources in demonstrating that an area has a lack of jobs in declining
occupations or industries.
Description in an academic study or other publication as an area where there is a lack of jobs.
The state might consider providing an academic study or other credible publication that
documents a lack of sufficient jobs in an area.
The state may submit whatever data or evidence it deems appropriate to support these types
of requests. FNS will evaluate such requests on a case-by-case basis and will approve those
that provide compelling support of a lack of sufficient jobs in the area. FNS strongly
encourages the state to work closely with its regional offices for technical assistance if it is
considering requesting a waiver based on the less common support mentioned above.

B. The Proposed Rule Would Restrict the Evidence to Support Lack of
Sufficient Jobs to a Single, Imperfect Measure of Labor Market
Conditions
The NPRM says the proposed core standards would not include other labor market information,
such as a low and declining employment-to-population ratio, a lack of jobs in a declining industry, or
an academic study or other publication(s) that describes an area’s lack of jobs. It would eliminate the
ability of states to support a waiver request using other available information about the labor market,
unless BLS unemployment data for the area is limited or unavailable, such as a reservation area or
U.S. territory. FNS proposes to eliminate these other criteria on the grounds that they are “rarely
used, sometimes subjective, and not appropriate when other more specific and robust data are
available,” but does not provide further substantiation of this claim.
83

The proposed rule would replace an approach that allows for multiple measures to capture labor
market conditions experienced by individuals subject to the time limit with a single, limited, and
imperfect measure, the unemployment rate. FNS has stated in its guidance that using the
unemployment rate is an imperfect measure for the job prospects for individuals subject to the time
limit. Labor market researchers routinely use other labor market measures in addition to, or instead
of, the unemployment rate, such as the employment-to-population ratio.
Other Measures, Including the Employment-To-Population Ratio,
Provide Important Information About Labor Market Conditions
That the General Unemployment Rate Does Not
The employment-to-population ratio is a well-defined and widely used measure that is far from
subjective. The employment-to-population ratio is the proportion of the civilian noninstitutional
population aged 16 and over that is employed. As the 1996 guidance describes, employment data for
areas is available from BLS. Population estimates for areas are available from the Bureau of Census.
The calculation of the employment-to-population ratio is a standard BLS procedure, which is a
measure it reports on a regular basis at the regional and state level.181 In many instances, researchers
use employment-to-population ratio as a more appropriate measure for labor market conditions for
low-skill workers who face serious barriers to employment.
Current regulations allow states to demonstrate that an area lacks sufficient jobs by showing that it
has a low and declining employment-to-population ratio. The rule proposes eliminating this criterion
as a means for an area to qualify for a waiver. This would be a mistake, as it would throw away
valuable information about the state of the labor market and the likely availability of jobs that cannot
be gleaned from the unemployment rate alone. The unemployment rate is the number of people
actively looking for a job as a percentage of the labor force (the number of people who have a
job plus the number of people who don’t have a job but are actively looking for one). In a job
market with limited job opportunities for any of a number of reasons, such as weak demand due
to a national economic recession, a local business slump, or the closing of a major plant, there could
be a number of people who would like to work but for reasons such as discouragement due to a
failed job search, experience with discrimination, or a general sense that their job prospects are
limited haven’t looked recently enough to be counted as in the labor force but unemployed.
These individuals are classified as “marginally attached to the labor force” and are included in
broader measures of labor market underutilization, including the U-6 measure, which
includes the unemployed, the marginally attached to the labor force, and those who are working
part-time but want to be working more hours.182

181
182

See for instance: https://www.bls.gov/news.release/srgune.nr0.htm

See for instance “How the Government Measures Unemployment,” Bureau of Labor Statistics, online
at https://www.bls.gov/cps/cps_htgm.htm#unemployed

84

Figure 4.1. Job Market Indicators in the Great Recession

In a national recession, a local economic slump, or in localities with limited job opportunities, the
unemployment rate can paint a very incomplete picture of the availability of jobs. This was
illustrated dramatically at the national level in the Great Recession. Between the start of the
recession in December 2007 and early 2010, the share of the population with a job (the
employment-to-population ratio) fell sharply. That was mostly due to the sharp rise in the
unemployment rate, but some of it reflected a drop in labor force participation as the number of
people marginally attached or otherwise not in the labor force rose.
The unemployment rate then began a long decline but the labor force participation rate continued
to fall as well. As a result, the share of the population with a job remained depressed and did not
begin to rise again until 2014 (see Figure 4.1, above).
The U-6 measure of unemployment came down more slowly than the official unemployment rate
as jobs, especially full-time jobs, remained scarce. Even as the unemployment rate dropped below 7
percent, the employment-to-population ratio remained well below where it was at the start of the
recession.
Researchers Routinely Use Employment-to-Population Ratio
to Measure Local Labor Market Conditions
Researchers routinely use the employment-to-population ratio in addition to, or instead of, the
unemployment rate to measure labor market conditions. According to Bartik, it is “unclear whether
the availability of labor is best measured by employment-to-population ratios or employment to

85

labor force ratios.”183 Bartik finds that employment-to-population ratios are more strongly related to
job growth than employment to labor force ratios.184
For individuals subject to the time limit, the employment-to-population ratio may be more
appropriate than the unemployment rate. According to Western and Pettit, for groups who are
weakly attached to the labor market and who face significant barriers to labor force participation,
like young men with little education, economic status is often measured by the employment-topopulation ratio. This measure counts as jobless those who have dropped out of the labor market
altogether. The unemployment rate is more restrictive and does not account for individuals who are
not currently in the labor force. 185 A study by Cadena and Kovak illustrates this approach, using
employment-to-population ratios to estimate the probability of employment in the less-skilled labor
market.186
An improved (or deteriorating) unemployment rate does not directly correspond to an
improvement (or deterioration) of the employment situation, because it does not take into account
changes in the labor force participation rate due to the movement of discouraged jobseekers in and
out of the labor market. Only a stable participation rate allows for unambiguous conclusions from a
rising (or falling) unemployment rate. Unemployed people who have been adversely affected by
economic restructuring may give up hope of working again and withdraw from the labor force. Job
booms may only be a boom for certain kinds of workers. Watson argues that a more useful
indication of the quantity of employment in the economy is provided by employment-to-population
ratios, which remove the confounding influence of labor force participation and give a more
accurate indication of the amount of employment available to the population. 187
Hoynes estimated the effect of local labor markets on Aid to Families with Dependent Children
participation in California using several measures of labor market conditions, including
unemployment rates, log of employment, employment-to-population ratios, and earnings. Results
showed that higher unemployment rates, lower employment growth, lower employment-topopulation ratios, and lower wage growth are associated with longer welfare spells and shorter
periods off welfare. Models that controlled for labor market conditions using employment-based
measures, such as employment-to-population ratios, performed better than unemployment rates.

183

The unemployment rate can be derived from the employment to labor force ratio by subtracting the latter from 1.

184

Timothy J. Bartik, “How Do the Effects of Local Growth on Employment Rates Vary with Initial Labor Market
Conditions,” Upjohn Institute Staff Working Paper 09-148 (Nov. 4, 2006), pp. 1-35,
https://www.econstor.eu/bitstream/10419/64401/1/607052678.pdf.
185

Bruce Western and Becky Pettit, “Incarceration and Social Inequality,” Daedalus Journal of the American Academy of Arts
& Sciences (Summer 2010), pp. 8-19, https://www.mitpressjournals.org/doi/pdf/10.1162/DAED_a_00019%20.
186

Brian C. Cadena and Brian K. Kovak, “Immigrants Equilibrate Local Labor Markets: Evidence From the Great
Recession,” National Bureau of Economic Research (August 2013), https://www.nber.org/papers/w19272.pdf.
187

Ian Watson, “Beyond the Unemployment Rate: Building a Set Indices to Measure the Health of the Labour Market,”
Australian Bulletin of Labour (September 2000), pp. 175-190,
http://www.ianwatson.com.au/pubs/health%20of%20labour%20market.pdf.

86

“Unemployment rates are less desirable measures of labor market opportunities because they
fluctuate not only with employment but also with changes in labor force participation.” 188
Dranove, Garthwaite, and Ody used employment-to-population ratio to examine the impact of
the economic slowdown that began in 2007 on the rate of growth in health spending. They used the
employment-to-population ratio, rather than unemployment rate, because it is not affected by
decisions to enter the labor force and instead provides a local measure of changes in economic
activity resulting from the slowdown. Their results were broadly consistent with results using the
local unemployment rate instead of employment-to-population ratio.189
The General Unemployment Rate May Not Adequately Measure
Weak Labor Demand at the State and Sub-State Level
During this period when the national employment-population ratio was flat, there were many local
and regional labor markets where labor market conditions remained weak even as the general
unemployment rate fell.
In a 2017 speech that partially focused on the geographical variance of labor markets across the
country (and on policies to ameliorate such differences), then Federal Reserve Chair Janet Yellen,
pointed out the following: 190
While the job market for the United States as a whole has improved markedly since the depths
of the financial crisis, the persistently higher unemployment rates in lower-income and
minority communities show why workforce development is so essential. For instance,
unemployment rates averaged 13 percent in low- and moderate-income communities from
2011 through 2015, compared with 7.3 percent in higher-income communities…. The
challenges for workers in minority communities are even greater. The average unemployment
rate across all census tracts where minorities made up a majority of the population averaged
14.3 percent from 2011 through 2015.
Labor economist Danny Yagan added an important insight about the geographical dispersion of
employment conditions following the historically large, negative demand shock from the Great
Recession.191 As Figure 4.2 below shows, states that were harder hit by the downturn saw
significantly larger losses in employment rates, even years after the recession was over. Yagan argues
188

Hilary W. Hoynes, “Local Labor Markets and Welfare Spells: Do Demand Conditions Matter?” The Review of
Economics and Statistics (August 2000), pp. 351-368, https://gspp.berkeley.edu/assets/uploads/research/pdf/HoynesRESTAT-2000.pdf.
189

David Dranove, Craig Garthwaite, and Christopher Ody, “Health Spending Slowdown is Mostly Due to Economic
Factors, not Structural Change in the Health Care Sector,” Health Affairs (Aug. 2014), pp. 1399-1406,
https://www.healthaffairs.org/doi/pdf/10.1377/hlthaff.2013.1416.
190

Janet Yellen, “Addressing Workforce Development Challenges in Low-Income Communities,” Federal Reserve
Board of Governors, March 28, 2017, https://www.federalreserve.gov/newsevents/speech/files/yellen20170328a.pdf.
191

Danny Yagan, “Employment Hysteresis from the Great Recession,” NBER Working Paper No. 23844, August 2018,
https://www.nber.org/papers/w23844.

87

that his findings provide evidence of “hysteresis,” meaning lasting economic damage to persons and
communities from periods of economic weakness. As he summarizes, “These findings reveal that
the Great Recession imposed long-term employment and income losses even after unemployment
rates signaled recovery.”192
As shown in the next section, even at low rates of national and regional unemployment (meaning
rates well below 7 percent), there are areas of the country where economic weakness persists.
Yagan’s findings suggest that these areas may suffer from more lasting damage to workers’ ability to
find gainful jobs. In the context of the proposed rule, such dynamics speak to the importance of
taking a much more nuanced approach to the waiver process, examining local labor markets from
both the demand side (i.e., the extent of job availability, both quantity and quality) and the supply
side (i.e., the skills and abilities of members of the local workforce to respond to labor demand).
Figure 4.2. State-Level Great Recession Employment Shocks
and 2007-2015 Employment Rate Changes

Note: Yagan defines employment shocks as the sum of state-level employment growth forecast errors
for 2008 and 2009. These forecast errors represent the difference between each state's actual
employment growth and its predicted employment growth based on pre-recession trends. Values on the
x-axis represent the inverse of 2007-2009 employment growth forecast errors.

Geographical Variation of Weak Labor Demand, Even at Low Unemployment
Echoing the Yagan findings referenced above, a recent paper by Austin et al. illustrates that labor
demand, particularly for low-wage workers, varies significantly from place to place. 193 In their recent
192
193

Ibid.

Benjamin Austin, Edward Glaeser, and Lawrence Summers, “Jobs for the Heartland: Place-Based Policies in 21st
Century America,” March 8, 2018, https://www.brookings.edu/wp-content/uploads/2018/03/AustinEtAl_Text.pdf.

88

analysis of regional disparities, these authors find pockets of persistently weak labor markets across
America, citing what they label: “a hardening of America’s geographic divisions.” Their paper
identifies three findings particularly germane to the shortcomings of the new rule: “the decline of
geographic mobility,” “increased sorting by skills across space,” and “persistent pockets of
nonemployment.” The combination of these three negative developments imply a larger share of
lower-wage workers stuck in various locations without enough work. We find these disparities very
much present in the labor market over the current expansion, even at historically low rates of
unemployment.
Many labor economists consider the prime-age employment rate to be a proxy for labor demand.
As part of their “Distressed Community Index,” the Economic Innovation Group (EIG) provides
county-level data on non-employment rates, or 1 minus the employment rate. Thus, higher nonemployment rates correspond to weaker labor demand.
Between 2012 and 2016, the average national non-employment rate for prime-age workers was 23
percent, meaning 77 percent of such workers had jobs. EIG’s data, to which we appended countylevel unemployment data from the BLS, reveal that in counties with unemployment rates between
6.5 and 7.5 percent, the average non-employment rate for prime-age adults was about 34 percent,
more than 10 percentage points above the national average.194 Note that even at the worst of the
Great Recession, the non-employment rate peaked at about 25 percent.195
The scatterplot in Figure 4.3 below shows the correlation between un- and non-employment at
the county level. Note that the scatterplot expands at higher unemployment, implying greater
dispersion of labor demand across counties at higher rates of unemployment. For example, the plot
shows that at 10 percent county unemployment, there are some counties with quite low nonemployment rates and some with very high rates. This dispersion further underscores the need to
avoid the single number approach proposed in the rule. Second, the scatterplot shows that at 7
percent unemployment, as noted above, non-employment is above 30 percent.

194

To be clear, employment (and non-employment) rates are mechanically correlated as higher unemployment means
lower employment. Our focus here, however, is on the levels of these variables and what they imply for labor demand.
195

EIG and BLS have slightly different definitions for “prime-age” – BLS uses adults 25-54, and EIG uses adults 25-64.

89

Figure 4.3. County-Level Unemployment and Non-Employment

Using the same procedure employed in the previous section, a regression of county-level nonemployment rates on the county’s unemployment rate, predicts that at 5, 7, and 10 percent
unemployment rates, county-level non-employment rates would range from 27 to 41 percent. (Table
4.1.) In other words, such high levels of non-employment demonstrate significant labor market slack
at the jobless rates proposed by the Department.
TABLE 4.1

Predicted County-Level Prime-Age Non-Employment Rate
Unemployment Rate

Predicted Prime-Age Non-Employment Rate

5%

27%

7%

33%

10%

41%

Note: County-level prime-age non-employment rates are predicted by regressing county-level unemployment rates on non-employment
rates.

The Federal Reserve recognized that there was still considerable “slack” in the labor market not
captured by the unemployment rate and kept short-term interest rates effectively at zero until
December 2015 before it began to raise them cautiously in small increments.

90

C. Information About Declining Occupations or Industries Can Help Identify
Smaller Areas Experiencing a Lack of Sufficient Jobs
According to current regulations and guidance, states can support a claim of lack of sufficient jobs
by providing evidence of a lack of jobs in declining occupations or industries. This can be especially
true in smaller, rural areas in which the loss of a single job provider, such a major manufacturing
plant or mining industry, can have a major effect on local job availability. In the December 1996
guidance, FNS suggested that states could use BLS monthly data published in the “Employment and
Earnings” report on state and sub-state employment figures by major industry.196 A declining trend
within a particular industry or sector may be taken as evidence of declining employment prospects
for persons with experience in or skills appropriate to that sector.
Although states have not frequently used occupation or industry employment data to support
claims of lack of sufficient jobs, FNS has approved them on a limited case-by-case basis. For
example, FNS approved waivers for a county (Polk) in Arkansas and a county (Coos) in New
Hampshire that were significantly affected by plant closures during the recession that started in
2001. The state agencies provided evidence of the adverse labor force impacts due to a major factory
or plant closing, such as the number of workers affected by layoffs and rapidly increasing
unemployment rates (10 percent and higher) over a short period of time. The impact of a plant
closure may not show up in 24-month unemployment rates until several months, or even a year,
have passed. Information indicating the decline of particular industries, such as significant plant
closures, gives states the ability to quickly adapt their waiver policy to respond to rapidly
deteriorating labor market conditions.

D. Eliminating Criteria of Three-Month Average Unemployment Rate Over
10 Percent and Historical Seasonal Unemployment Rate Over 10
Percent Is Inconsistent With the Statute
The proposed rule would restrict states’ ability to use an unemployment rate over 10 percent as
the basis for waiver approval. It would limit the use of the criterion of a recent three-month average
unemployment rate over 10 percent to “exceptional circumstances” and eliminate the criterion of an
historical seasonal unemployment rate over 10 percent.197 This would leave just one criterion —
having a 12-month average unemployment rate over 10 percent — as the basis for approval using an
average unemployment rate over 10 percent. These changes are inconsistent with the statute and
regulations that clearly establish that areas with an unemployment rate over 10 percent qualify for a
waiver. If the Department proceeds to publish a final rule it must reject these changes to be
consistent with the statute.
According to the statute, a state may waive the applicability of the work requirement “to any
group of individuals in the state if the Secretary makes a determination that the area in which the
individuals reside has an unemployment rate above 10% or does not have a sufficient number of
jobs to provide employment for the individuals.”198 The statute clearly establishes the 10 percent

196

https://www.bls.gov/opub/ee/home.htm

197

NPRM, p. 985, 987.

198

Food and Nutrition Act, 7 U.S.C. § 2015(o)(4). This language is identical to the language in P.L. 104-193, PRWORA.

91

unemployment rate criterion as a basis for approval. The statute does not specify requirements
regarding the duration of time that an area must have an unemployment rate above 10 percent.
Three-Month Average Unemployment
In guidance issued in December 1996 and then reinforced in the preamble of the 1999 proposed
rule,199 200 the Department stated that it would not require a 12-month average to approve a waiver
because of two shortcomings. “A 12-month average will mask portions of the year when the
unemployment rate rises above or falls below 10 percent. In addition, requiring a 12-month average
before a waiver could be approved would necessitate a sustained period of high unemployment
before an area became eligible for a waiver.” To address these shortcomings and to ensure that
waivers are granted as quickly as possible where needed, the Department explained that “states have
several options. First, a state might opt to use a shorter moving average. A moving average of at
least three months is preferred. In periods of rising unemployment, a three-month average provides
a reliable and relatively early signal of a labor market with high unemployment. A state might also
consider using historical unemployment trends to show that such an increase is not part of a
predictable seasonal pattern to support a waiver for an extended period (up to one year).”201
In the preamble to the proposed rule, the Department expressed its preference that waivers reflect
current economic conditions.202 Yet by eliminating the ability of states to use a recent three-month
average unemployment rate over 10 percent as the basis for waiver approval, it is eliminating one of
the criteria that most closely aligns with current economic conditions and signals deteriorating labor
market conditions in an area.
Historical Seasonal Unemployment
In guidance issued in December 1996 and in the preamble of the 1999 proposed rule,203 204 the
Department confirmed the applicability of waivers to “areas with predictable seasonal variations in
unemployment.” The Department provided a detailed example:
States may use historical trends to anticipate the need for waivers for certain periods. For
example, if the pattern of seasonal unemployment is such that an area’s unemployment rate
typically increases by two percentage points in January, February, and March, and the area’s
unemployment rate is currently 9 percent, a state may request a waiver for this area based on
its current rate and historical trends. The period covered by the waiver will then coincide with
the period of high unemployment.

199

USDA, “Guidance for states Seeking Waivers for Food Stamp Limits,” December 3, 1996.

200

Food Stamp Program: Personal Responsibility Provisions of the Personal Responsibility and Work Opportunity
Reconciliation Act of 1996; 64 Federal Register 242 (December 17, 1999) (to be codified at 7 C.F.R. pts. 272 and 273).
201
202
203
204

USDA, “Guidance for states Seeking Waivers for Food Stamp Limits,” December 3, 1996.
NPRM, p. 986.
USDA, “Guidance for states Seeking Waivers for Food Stamp Limits,” December 3, 1996.

Food Stamp Program: Personal Responsibility Provisions of the Personal Responsibility and Work Opportunity
Reconciliation Act of 1996; 64 Federal Register 242 (December 17, 1999) (to be codified at 7 C.F.R. pts. 272 and 273).

92

Aligning the Period Covered by the Waiver and the Period of Projected High Unemployment
Does Not Require Data of a Particular Duration
The 2001 final rule codified criteria related to unemployment rates over 10 percent at 7 C.F.R. §
273.24 (f)(2)(i) and provided flexibility to meet these criteria using data of varying duration. “To
support a claim of unemployment over 10 percent, a state agency may submit evidence that an area
has a recent 12-month average unemployment rate over 10 percent; a recent three-month average
unemployment rate over 10 percent; or an historical seasonal unemployment rate over 10 percent.”
The intent of current regulations was to align the period covered by the waiver to the period when
unemployment is high, rather than designate an arbitrary duration requirement:
Therefore, the Department is proposing that in general, the duration of a waiver should bear some
relationship to the documentation provided in support of the waiver request. FNS will consider
approving waivers for up to one year based on documentation covering a shorter period, but the
State must show that the basis for the waiver is not a seasonal or short term aberration.205
In the preamble of the NPRM, the Department arbitrarily adds a duration requirement of 12
months to the 10 percent criterion.206 Only areas with a recent, 12-month average unemployment
rate over 10 percent would be considered for approval. Under the proposed rules, the Department
may approve a waiver for an area with a recent three-month average unemployment rate over 10
percent only if an “exceptional circumstance has caused a lack of sufficient number of jobs.” 207
The Department does not discuss the rationale for restricting the 10 percent criterion to a 12-month
duration. It does not adequately explain what represents an exceptional circumstance and what
economic measures might signal this circumstance. It does not discuss what economic measures a
state might be able to use as an alternative to the three-month average unemployment rate, which it
has described as a “reliable and relatively early signal of a labor market with high unemployment” in
past guidance.
The Department argues for eliminating the historical unemployment rate criterion because it does
not demonstrate “prolonged” lack of sufficient jobs, that it is “limited to a relatively short period of
time each year,” and that it is “cyclical rather than indicative of declining conditions.”208 The
Department acknowledges that, by definition, historical seasonal unemployment is contradictory
with prolonged duration. Rather than drop the newly introduced and contradictory requirement on
duration (which is inconsistent with existing statute and regulation), the Department argues for the
elimination of the historical seasonal unemployment criterion (which is upheld in existing statute
and regulations).
The Department also proposes to eliminate the historical seasonal unemployment criterion
because it has not approved a waiver using this criterion. This is not sufficient ground for the
proposed change, as the Department has no way of knowing if states intend to use this criterion in
205

Food Stamp Program: Personal Responsibility Provisions of the Personal Responsibility and Work Opportunity
Reconciliation Act of 1996; 64 Federal Register 242 (December 17, 1999) (to be codified at 7 C.F.R. pts. 272 and 273).
206

NPRM, p. 983.

207

NPRM, p. 985, 992.

208

NPRM, p. 987.

93

the future. To maintain consistency with the statute, we urge the Department to leave the regulation
as is and retain the three-month average unemployment rate over 10 percent and historical seasonal
unemployment rate over 10 percent as criteria for waiver approvals.

E. We Recommend Rejecting Proposed Changes That Would Ignore
Important Information About Labor Market Conditions Not Captured by
the General Unemployment Rate
These studies and analyses illustrate how the unemployment rate alone may not tell the full story
of how abundant or scarce jobs are in the labor market. FNS would be mistaken to rely solely on
unemployment rates as the basis for demonstrating that an area has a lack of sufficient jobs. The
unemployment rate does not account working-age persons who may have dropped out of the labor
force altogether. Other labor force measures, such as the employment-to-population ratio or
industry-specific employment data, complement unemployment rates in capturing the labor market
conditions faced by individuals subject to the time limit, who often face significant barriers to labor
force participation. In the Great Recession a low or depressed employment-to-population ratio was
often a better measure of labor market slack and lack of job opportunities than the unemployment
rate. Thus, a low or falling employment-to-population ratio is a valuable indicator and data are
available for local areas.
The proposed rule is based on insufficient reasons to change current regulations by prohibiting
states from using average unemployment rates over 10 percent and other available information
about labor market conditions, except for areas that have limited or unavailable unemployment data
from BLS or a BLS-cooperating agency. It fails to discuss the reasons why it is restricting the use of
average unemployment rates over 10 percent during periods of acute or seasonal high
unemployment or the limitations of the general unemployment rate in assessing the labor market
conditions, particularly those faced by individuals subject to the time limit. It does not acknowledge
the valuable information that will be lost if measures such as the employment-to-population ratio are
excluded as evidence of lack of sufficient jobs. Given the lack of supporting information, the public
has an insufficient opportunity to comment meaningfully on the proposed rule and we recommend
rejecting the proposed changes to the rules.

94

Chapter 5. Restricting State Flexibility on Grouping
Areas Is Counter to Evidence
The NPRM proposes several significant changes to longstanding SNAP policy that
would significantly restrict state flexibility to develop and implement waiver policy that aligns with
state operations, priorities, and resources. The NPRM fails to provide reasons for limiting states’
ability to consider relevant factors when grouping areas covered by waivers, fails to acknowledge
decades of state discretion in grouping areas (including statewide areas) for waivers, and fails to
identify the data and evidence that justify the elimination of statewide waivers and the use of one
narrow, inflexible, federally prescribed method for grouping areas. Without knowing what evidence
justifies such a drastic change in longstanding policy and an adequate discussion of alternative
methods for grouping, it is impossible to assess the potential impact of the changes on SNAP
participants and their ability to achieve self-sufficiency. The sections below provide an overview of
existing statutes, regulations, and guidance, and discuss factors that states consider when grouping
areas, alternative grouping methods used by states to group areas, and limitations of the method for
grouping proposed by the Department.

A. States Have Had Broad Discretion to Define Areas for More Than Two
Decades
Since the passage of the 1996 welfare law (P.L. 104-193) and the three-month time limit, FNS has
given states broad discretion to determine which geographic areas the state would like to waive from
the three-month time limit, including an area spanning the entire state or sub-state areas. While every
state-defined area as a whole must meet the waiver eligibility criteria as set forth in 7 C.F.R. § 273.24
(f), states may define areas that best align with local and regional labor force conditions, resources,
and administrative needs. The federal rules do not limit waivers to specific sub-state areas, such as
cities or counties. As states define areas to request waivers for, they often consider a range of factors
within geographic regions, such as labor market characteristics, job opportunities, availability of
SNAP Employment and Training (E&T) services, housing and transportation, workforce and
economic development resources and strategies, and SNAP agency administrative capacity.
States have had discretion to define areas in accordance with the law, regulation, and guidance
over the past 22 years. For nearly as long, USDA has approved waivers for entire states or those that
group sub-state areas.209 The proposed rule would take flexibility away from states to define what
areas they wish to waive and restrict states to one narrow, inflexible, federally prescribed criteria.
The proposed criteria are likely outdated and disconnected from local and regional factors that states
consider when developing and implementing policies to connect low-skilled workers to job and
training opportunities. By prohibiting states from grouping areas according to their needs, FNS
would hamper their ability to deliver integrated support to SNAP participants in gaining the skills
and work experience needed to secure jobs leading to self-sufficiency. The proposed rule would
severely restrict, and potentially eliminate, state flexibility to define areas for waivers, without
providing evidence that the changes would help increase self-sufficiency among SNAP participants.

209

In 2000, FNS approved a waiver requested by Florida for the combined area of Broward and Dade Counties, which
belong to the same Metropolitan Statistical Area.

95

According to the statute, a state may waive the applicability of the work requirement “to any
group of individuals in the state if the Secretary makes a determination that the area in which the
individuals reside has an unemployment rate above 10% or does not have a sufficient number of
jobs to provide employment for the individuals.” The statute does not identify or require a
geographic definition of “area.” The 2018 farm bill did not change this and the House bill proposal
that sought to limit states’ ability to define areas was rejected.
According to the current rule (7 C.F.R. § 273.24 (f)(6)), “States may define areas to be covered by
waivers. We encourage state agencies to submit data and analyses that correspond to the defined
area. If corresponding data does not exist, state agencies should submit data that corresponds as
closely to the area as possible.” The current rule gives states broad discretion in defining regions,
requiring only that the data and analysis that states submit to support the waiver request correspond
to the defined area.
Below are excerpts from USDA guidance and rulemaking that uphold state flexibility in defining
areas:
• December

3, 1996 guidance: The initial USDA guidance on waivers from the three-month
time limit gives states flexibility to define areas and goes a step further by encouraging states
to consider combining sub-state areas. “USDA will give states broad discretion in defining
areas that best reflect the labor market prospects of program participants and state
administrative needs.”210 While USDA encouraged states to consider sub-state waivers over
statewide, the flexibility was left completely to states.

• 1999 proposed

rule: In the preamble, USDA noted its intent to “balance the competing goals
of ensuring consistent national application of these requirements, and providing state agencies
with appropriate implementation flexibility” to implement the time limit. 211 “The Department
is allowing states broad discretion in defining areas that best reflect the labor market prospects
of Program participants and state administrative needs. In general, the Department encourages
states to consider requesting waivers for areas smaller than the entire state. Statewide averages
may mask slack job markets in some counties, cities, or towns. Accordingly, states should
consider areas within, or combinations of, counties, cities, and towns. The Department also
urges states to consider the particular needs of rural areas and Indian reservations. Although
the Department is proposing to allow states flexibility in defining areas to be covered by
waivers, the supporting data must correspond to the requested area (e.g., a county-wide waiver
must be supported by county-wide data). In other words, states may define areas to be
covered by waivers, but the data and analysis used to support the waiver must correspond to
the defined area.” [Emphasis added.]

• 2001 final rule:

In the preamble of the final rule, USDA noted that it had “proposed that state
agencies have complete discretion to define the geographic areas covered by waivers so long
as they provide data for the corresponding area” and that most of the comments received
supported this proposal. USDA explained that, “for simplicity sake, we encourage states to

210
211

USDA, “Guidance for states Seeking Waivers for Food Stamp Limits,” December 3, 1996.

Food Stamp Program: Personal Responsibility Provisions of the Personal Responsibility and Work Opportunity
Reconciliation Act of 1996; 64 Federal Register 242 (December 17, 1999) (to be codified at 7 C.F.R. pts. 272 and 273).

96

define areas for which corresponding data exists. We believe this is very easily done, especially
since unemployment data goes down to the census tract level.”212
• August 2006 guidance:

“Jurisdictions or a cluster of areas or counties may be combined to
waive an area larger than one county. States have authority to define the cluster of areas to be
combined. If a state defines its own jurisdiction or cluster of areas, the boundaries or clusters
must be thoroughly documented to expedite review of the waiver request. The Department of
Commerce, Bureau of Economic Analysis (BEA) is one source that can be used to identify
economic areas. This data may be found at the website
www.bea.gov/bea/regional/docs/econlist.cfm. These areas define the relevant regional
markets surrounding metropolitan or micropolitan statistical areas. They consist of one or
more economic nodes — metropolitan or micropolitan statistical areas that serve as regional
centers of economic activity — and the surrounding counties that are economically related to
the nodes. Other sources or methods may be used to combine a cluster of areas.”213 The
guidance also provided an example illustrating the use of the Department of Commerce
economic areas to create groups of counties in Montana. It also explained that “the state could
request a waiver for all counties or a sub-area of the economic areas as long as the data for the
combined area meets the waiver criteria.”

• December

2, 2016 guidance: In its most recent guidance on waivers, USDA repeated that
“the state agency has discretion to define the area(s) in which it requests to waive the time
limit.” According to this latest guidance, “the state can request that a waiver apply statewide or
at the sub-state level, as statewide averages may mask slack job markets in some counties,
cities, or towns. However, in order to receive FNS approval to waive the ABAWD time limit
the state must support its request with evidence that corresponds to the requested area (e.g., a
county-wide waiver must be supported by county-wide data). The state must also clearly
identify which areas are being requested and under which criteria. Unemployed and labor
force data from individual areas can be combined to waive a larger group of areas, whether
based upon a recent unemployment rate over 10 percent or a 24-month unemployment rate
20 percent above the national average.”214
USDA provided guidance on how states could combine areas. The guidance requires that
combined areas must be contiguous or must belong to an economic region. The guidance
provides flexibility in defining an economic region. “In order to be combined, the areas must
be contiguous or considered parts of the same economic region. For example, two or more
contiguous counties could be grouped together in order to consider their aggregate average
unemployment rate. If the counties in the sub-area all belong to the same region, they do not
need to be contiguous to be defined as an area. The state has discretion to define the group of
areas to be combined, provided that the areas are contiguous or can be considered to be part
of an economic region. If the state defines its own group, the rationale for the boundaries of
the group must be thoroughly documented. For example, state or local labor departments
often have defined economic regions based upon shared industries or other factors. Other

212

Food Stamp Program: Personal Responsibility Provisions of the Personal Responsibility and Work Opportunity
Reconciliation Act of 1996; 66 Federal Register 11 (January 17, 2001). (to be codified at 7 C.F.R. pts. 272 and 273).
213

USDA, “Guidance on Requesting ABAWD Waivers,” August 2006.

214

USDA, “Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents
(ABAWD),” December 2, 2016.

97

sources, methods, or rationale to support that areas share an economic region may also be
considered.” The guidance repeated the example from the August 2006 guidance of using
BEA economic areas as a guide for grouping counties.
Without providing justification or rationale, the proposed rule would end over two decades of
consistent guidance and support for state flexibility to determine the geographic scope of waivers
that best aligns with state SNAP policies and capacity, training and workforce service delivery,
funding and resources, and regional planning and strategies. The Department did not discuss or
reference over two decades of consistent regulation and guidance it has issued on grouping areas.
The Department did not go back to review the comments it received on the 1999 proposed rule
supporting the proposal to give states complete discretion to define areas. The Department did not
elaborate on any shortcomings it believes exist with the current flexibility that states have to define
geographic areas. This makes it difficult for people who wish to comment to critique the
Department’s proposal to restrict the ability of states to define areas they would like to waive from
the three-month time limit.
The proposed rule significantly restricts the ability of states to waive groups of areas. Without
providing a discussion, the Department arbitrarily eliminates the ability of states to waive the entire
state even when statewide unemployment rates have risen significantly during an economic
downturn, except for situations when the state qualifies for extended unemployment benefits.
The Department introduces a specific definition of labor market areas as the only acceptable
method for grouping areas and does not acknowledge past guidance it has issued that encouraged
states to explore different sources and methods for grouping areas. It proposes these labor market
areas to ensure that grouped areas are economically tied, yet this approach only captures one way
(commuting patterns) that areas might be economically tied. This proposal ignores all the other ways
areas may be economically integrated, such as through workforce development initiatives, economic
development investments, employer recruiting practices, and migration patterns.
For example, guidance issued in 2006 and 2016 both use the BEA economic areas (either entire
economic areas or sub-areas) to illustrate how a state can combine unemployment data to support a
waiver for grouped areas. The guidance suggests that states explore other sources or methods for
combining areas, including economic regions defined by state or local labor departments. Even if
the Department had provided reasons for requiring a very specific method for combining areas, it is
difficult for the public to understand why the Department would disregard or minimize other
economic or administrative factors, such as SNAP E&T service provision, that it currently gives
great consideration to in other aspects of program operations.

B. States Use the Current Flexibility to Align SNAP Policies With
Administrative Needs, Job Opportunities, Training Funding and
Resources, and Economic and Workforce Development Strategies
States consider multiple factors when grouping areas for waivers to align resources, administrative
policies and capacity, and service delivery. A state may consider a range of local, sub-state (regional),
and statewide factors:
•
98

SNAP E&T service delivery

• TANF

work programs

• Workforce

Innovation and Opportunity Act (WIOA) regional workforce development
funding and strategies

• Office

locations (SNAP, workforce development centers)

• Community
• Employer
• Regional

college locations

and industry recruiting patterns

economic development funding and strategies

• Commuting

patterns

• Housing and

transportation patterns
Aligning With SNAP E&T Coverage

By limiting state flexibility to define areas, the proposed rule would restrict a state’s ability to
allocate and coordinate SNAP E&T resources and service delivery to meet the needs of its SNAP
participants. States have used their discretion to define areas to help align the geographic scope of
waivers with areas where they are unable to provide sufficient work or training opportunities to
work registrants, including those subject to the time limit. A state that can only provide SNAP E&T
slots in certain counties may request waivers for (eligible) counties where SNAP E&T slots are not
available or guaranteed. As Maryland was preparing to lose its statewide waiver in January 2016, the
state agency requested a waiver for multiple sub-state areas, including the nine-county Eastern Shore
recognized by the state as an economic region. In areas not covered by waivers, Maryland offered
SNAP E&T services to individuals subject to the time limit.215 Similarly, Colorado operated a
mandatory SNAP E&T program for all work registrants, including individuals subject to the time
limit, in 40 of its 64 counties. Individuals in the remaining counties were not subject to the time limit
because of waivers or the use of individual exemptions, but could still participate in the E&T
program on a voluntary basis.216
Aligning With Workforce and Economic Development Regions
Restricting or eliminating waivers for grouped areas would deny states the ability to align SNAP
and Workforce Innovation and Opportunity Act (WIOA) regional service delivery, funding, and
planning efforts. Coordinating service delivery with WIOA can help SNAP agencies make more
qualified work activities available to SNAP participants because participation in a WIOA program is
considered a qualifying activity for purposes of meeting work requirements for individuals subject to
the time limit.

Maryland Department of Human Resources, “Maryland Supplemental Nutrition Assistance Program (SNAP)
Employment and Training (E&T) Program: state Plan of Operations,” Revised September 22, 2015,
https://dhr.maryland.gov/documents/Data%20and%20Reports/FIA/YR2016%20SNAP%20E&T%20State%20Plan%
20of%20Operations%20(revised).pdf.
215

Colorado Department of Human Services, “Colorado SNAP E&T State Plan: Federal Fiscal Year 2017,” 2016,
http://coemploymentfirst.org/wp-content/uploads/2017/01/Colorado-SNAP-E_T-State-Plan-FFY17.pdf.
216

99

Both USDA and the Department of Labor (DOL) recognize the opportunity to coordinate these
two programs to integrate services and resources and avoid duplication. In a joint letter issued in
March 2016, USDA and DOL encouraged SNAP and the workforce system to work together to
develop shared strategies to better connect SNAP participants, specifically individuals subject to the
time limit, to job and training services through WIOA American Job Centers (AJCs). 217 The letter
cited the shared goal of helping low-skilled, low-income, or low-wage individuals find work through
training activities and workforce programs.
A state may want to align its waivers and SNAP E&T service delivery with WIOA regions,
workforce development regions, or economic development regions in order to best plan and
coordinate service delivery related to training and job opportunities for the population subject to the
time limit. States may be able to maximize administrative capacity by aligning service delivery, case
management, and data tracking by multi-county regions, such as WIOA Local Workforce
Development Areas.
For example, Tennessee SNAP E&T services are delivered through the workforce system. SNAP
E&T participants are referred to the WIOA program for training provided through partnerships
with technical and community colleges. SNAP participants have access to on-the-job training (OJT)
opportunities not available outside of the WIOA-SNAP E&T partnership.218 Tennessee organizes its
workforce activities into three regions: East, Middle, and West. These regions are further broken
down into Local Workforce Development Areas (LWDAs). In 2007, before Tennessee eventually
became eligible for a statewide waiver during the most recent recession, the state requested waivers
for groups of counties belonging to the same LWDAs (known as Local Workforce Investment
Areas before the passage of WIOA in 2014).
States sometimes adjust the regional alignment of programs to reflect changes to the labor force,
resources, service delivery, and administrative capacity. Federal agencies may not be aware of these
changing circumstances or be able to make adjustments in a timely manner. In 2018, Tennessee
realigned its LWDAs by consolidating 13 areas into nine.219 The Tennessee Workforce Development
Board realigned the LWDAs to match the regional organization of other programs, such as the
Tennessee Department of Economic and Community Development base camps, Tennessee
Reconnect Communities, and Tennessee Pathways regions. 220

U.S. Department of Agriculture and U.S. Department of Labor, “Partnering to Help Connect Low-Income Ablebodied Adults to the Public Workforce System,” March 31, 2016, https://fnsprod.azureedge.net/sites/default/files/snap/USDA-DOL-joint-ABAWD-letter.pdf.
217

U.S. Department of Agriculture, “SNAP to Skills: Policy Brief 8,” June 2018,
https://snaptoskills.fns.usda.gov/sites/default/files/2018-06/Brief_June2018_508comp.pdf.
218

Tennessee Department of Labor and Workforce Development, “Map of Realignment of Local Workforce
Development Areas,” 2018,
https://www.tn.gov/content/dam/tn/workforce/documents/ProgramManagement/RealignmentMaps.pdf.
219

220

“TN Realigns Workforce Development Areas,” The Chattanoogan, June 28, 2018,
https://www.chattanoogan.com/2018/6/28/371092/TN-Realigns-Workforce-Development-Areas.aspx.

100

Alignment With Other Regional Factors
Beyond the alignment opportunities between SNAP E&T and WIOA, there are many other
reasons why a state might group sub-state areas. States may combine different streams of funding to
delivery SNAP E&T services regionally. Some funding opportunities may be available as regional
grants, such as some CDBG grants or workforce development grants. In Portland, Oregon, the
regional Workforce Development Board integrates WIOA, SNAP E&T, Community Development
Block Grant, and other funding streams, to provide workforce development activities serving SNAP
recipients and others in the Portland region.221
States may have administrative reasons for grouping areas. According to a 2016 report by the
USDA Office of Inspector General, the requirements related to the time limit are difficult to
implement.222 Some state officials said that waivers can help reduce the burden of tracking
individuals subject to the time limit. A state may request a waiver to cover areas that have reduced
administrative capacity and give areas more time to acquire staff, training, or upgrade case
management or data systems. For example, San Francisco County in California used the time while
it was covered by the waiver to upgrade its data systems and secure new E&T partnerships.223
States may want to align waivers with the geographic scope of other resources. A state may align
Information Technology (IT) systems, such as eligibility, case management, or data tracking systems
within geographic regions. Counties in California are grouped into eligibility system consortia (with
40 counties belong to the CalACES consortium and 18 counties belonging to the CalWIN
consortium). Within each consortium, counties are further organized into regions (eight CalACES
regions and four CalWIN regions). Each of these consortia systems support TANF work programs,
SNAP E&T activities, and county-specific employment programs. Waivers for groups of counties
could be organized by consortia regions to help align service delivery, case management, and data
tracking.
The Department did not explain why it was eliminating states’ ability to use relevant methods for
grouping areas, such as workforce development service delivery, to inform how they group areas
covered by waivers. From two decades of experience reviewing state waiver requests, the
Department is aware of how states use their existing flexibility to balance multiple priorities,
resources, and policies, such as SNAP E&T policies and services, housing and transportation
planning, and workforce and economic development strategies. The Department did not provide
reasons for ignoring these other considerations, making it difficult for the public to comment on the
proposed changes. As a federal agency, USDA may not be aware of all the local and sub-state
factors that impact the development and delivery of employment and training services. The
proposed rule makes sweeping and arbitrary changes that will hamper states’ ability to integrate and
coordinate resources to provide employment and training to SNAP recipients. By prohibiting states
221

U.S. Department of Agriculture, “SNAP to Skills: Policy Brief 8,” June 2018,
https://snaptoskills.fns.usda.gov/sites/default/files/2018-06/Brief_June2018_508comp.pdf.
222

U.S. Department of Agriculture, Office of Inspector General. “FNS Controls Over SNAP Benefits for Able-Bodied
Adults Without Dependents,” Audit Report 27601-0002-31, https://www.usda.gov/oig/webdocs/27601-0002-31.pdf.
U.S. Department of Agriculture, “State Highlights: California,” Retrieved Feb. 15, 2019,
https://snaptoskills.fns.usda.gov/state-highlights/state-highlights-california.
223

101

from grouping sub-state areas, the agency would limit states’ ability to coordinate and align SNAP
ABAWD policies with training opportunities and resources, workforce and economic development
strategies, and other factors within the state.

C. Eliminating a State’s Ability to Adjust to Rising Unemployment Across
the State
The proposed rule would eliminate statewide waivers when sub-state unemployment data is
available, except for situations when a state qualifies for extended unemployment benefits. The
Department provides no discussion of the rationale for eliminating this flexibility, other than
asserting that the use of sub-state unemployment data helps target particular areas with high
unemployment. This ignores the statistical principle of weighted averages — in order for an entire
state to qualify under current rules, unemployment rates throughout the states must have risen
dramatically, particularly in the most populous areas of the state.
While we commend the Department for retaining the qualification of a state for extended
unemployment benefits (EB) as a core standard for approval, this criterion does not adequately
detect states with high unemployment rates that are not rising rapidly. States must meet both the
minimum three-month unemployment threshold of 6.5 percent and have rising unemployment over
at least one year in order to qualify for extended benefits. States that only meet one of these
conditions would not be able to obtain a waiver. For example, a state with a consistent
unemployment rate of 8 percent over time would not qualify for extended benefits because its rate,
by definition, is not rising. Similarly, a state with rapidly rising unemployment, but whose rate has
not yet reached 6.5 percent, would also not qualify. In both examples, states can have high or
worsening unemployment and would not be able to obtain a waiver to help SNAP participants
working in these economic conditions.
To illustrate this, consider the experience of South Carolina and Oregon in 2007, prior to the
Great Recession. These states had high unemployment in the months preceding the recession, but
under the proposed rule would have had no options for statewide waivers until well into the
recession. South Carolina qualified for extended benefits in August 2008, eight months into the
recession that started in December 2007. Had the state sought a waiver for 2007 based on statewide
unemployment rates,224 it would have qualified based on 24-month average unemployment rates that
ranged between 6.6 and 6.8 percent for the various periods relevant for such a waiver. Similarly, the
state would have qualified for a 2008 statewide waiver under existing rules, based on 24-month
average unemployment rates that ranged between 6.1 and 6.6 percent for the various periods
relevant for such a waiver.
Oregon qualified for extended benefits in November 2008, 11 months after the start of the
recession. Had Oregon sought a waiver for 2007 based on statewide unemployment rates,225 it would
have qualified based on 24-month average unemployment rates that hovered between 5.8 and 6.7
percent for the relevant period.

224

South Carolina had a two-year statewide waiver that expired in February 2009.

225

Oregon was waived under a two-year statewide waiver that ended in April 2008.

102

The proposed rule only allows waivers for sub-state geographies, and not the entire state, until
statewide labor market conditions become so dire that the state qualifies for extended benefits. The
Department argues that statewide unemployment figures may include areas in which unemployment
rates are relatively low and that eliminating statewide waivers will help target areas in which
unemployment rates are high. The Department does not discuss the dynamic nature of labor market
conditions across time and across geographic areas. Unemployment rates do not change uniformly
within a state. A state may include areas with persistently high unemployment, areas with relatively
low but rapidly rising unemployment rates, areas with high unemployment rates that are slowly
creeping higher, as well as areas with relatively low unemployment rates.
Current rules allow a state to detect deteriorating economic conditions across the state even
before it qualifies for extended unemployment benefits. The Department makes an arbitrary
decision to eliminate statewide unemployment analysis because of potential variation in
unemployment rates within the state. Variation in unemployment rates exists at all geographic levels,
including at small scale Census tract and Census block group levels. The Department’s failure to
provide a robust assessment of the impact of this change on the ability of states to cushion the blow
of deteriorating economic conditions across their borders makes it difficult to comment on the
proposed rule.
Arbitrary Standard for Grouping Areas
The proposed rule would limit the ability of states to request waivers for groups of geographic
areas, such as multi-county areas, except for areas that are “economically tied.” The Department
provides a limited definition of an “economically tied” area based on commuting patterns — an area
within which individuals can reside and find employment within a reasonable distance or can readily
change employment without changing their place of residence. The preamble says that “existing
general conditions for grouping of areas — that the areas must be either contiguous and/or share
the same economic region — were intended to ensure that the areas grouped together are
economically tied.” Yet in statutes, regulations, and guidance over the past two decades, USDA has
given states broad discretion to define areas and has never expressed the requirement of that
grouped areas be “economically tied” based solely on commuting patterns. The proposed rule
arbitrarily imposes this requirement without providing justification or acknowledging the many other
ways areas can be economically tied apart from commuting patterns, such as employer recruiting
practices, regional workforce development strategies, and regional economic development and
investment patterns.
More specifically, the proposed rule would limit grouping to areas that are designated labor
market areas (LMAs) based on a narrow statistical definition used by the Bureau of Labor Statistics
(BLS). USDA has requested public comment on whether it should be even more restrictive and
prohibit grouping entirely. USDA proposes taking away state discretion to define areas, arguing that
the application of waivers on a more limited basis will encourage more individuals subject to the
time limit to take steps towards self-sufficiency, but does not explain how restricting the grouping of
areas will help achieve this goal. USDA did not offer any other alternative frameworks for grouping
areas for discussion, even the Bureau of Economic Analysis economic areas that it has used as an
example for grouping in past guidance.

103

D. Using a Narrow, Statistical Definition of Labor Market Areas
The Department uses the BLS definition of a labor market area, which is an area within which
individuals can reside and find employment within a reasonable distance or can readily change
employment without changing their place of residence. 226 It defines an “economically tied area” the
same way. By using the same definition for “labor market area” and “economically tied area,” the
Department is conflating the two concepts and makes it confusing and challenging for the public to
respond. The Department appears to be taking the BLS definition of labor market areas and
applying it to the more general concept of “economically tied” areas. Using the relatively narrow
definition of labor markets to also define economically tied areas ignores the various ways areas can
be connected economically beyond commuting patterns. For example, areas can share economic ties
through investment patterns, service delivery models, and migration patterns.
The proposed rule establishes BLS labor market areas as the only standard by which sub-state
areas may be grouped together and covered by a waiver from the three-month time limit. To
delineate the entire United States into mutually exhaustive and exclusive labor market areas, BLS
uses a narrow operational definition that relies solely on measures of population size and commuting
flows between counties. BLS LMAs include both the metropolitan and micropolitan areas defined
by the Office of Management and Budget (OMB) and the small labor market areas maintained by
the BLS Local Area Unemployment Statistics (LAUS) program.
Major labor market areas include Core Based Statistical Areas (CBSAs), which can be either
Metropolitan Statistical Areas or Micropolitan Statistical Areas. Metropolitan Statistical Areas have at
least one urbanized area with a population of 50,000 or more, along with adjacent territory that has a
high degree of social and economic integration with the core (as measured by commuting patterns).
Micropolitan Statistical Areas have at least one urban cluster with a population of at least 10,000 but
less than 50,000, along with adjacent territory that has a high degree of social and economic
integration with the core (as measured by commuting patterns).
An outlying county is combined with the central county or counties of the CBSA if it meets the
following commuting requirements:
• At

least 25 percent of the workers living in the county work in the central county or counties
of the CBSA; or

• At

least 25 percent of the employment in the county is accounted for by workers who reside
in the central county or counties of the CBSA.

Metropolitan and Micropolitan Statistical Areas are delineated in terms of whole counties (or
equivalent entities) and counties can only belong to one CBSA.227

226

Bureau of Labor Statistics, “Local Area Unemployment Statistics Geographic Concepts: Labor Market Areas,” 2018,
https://www.bls.gov/lau/laugeo.htm#geolma.
Office of Management and Budget, Revised Delineations of Metropolitan Statistical Areas, Micropolitan Statistical
Areas, and Combined Statistical Areas, and Guidance on Uses of the Delineations of These Areas, 2015.
https://www.whitehouse.gov/sites/whitehouse.gov/files/omb/bulletins/2015/15-01.pdf
227

104

For counties that do not belong to metropolitan or micropolitan areas, counties are combined
into a small LMA if either or both of the following conditions are met:228
• At

least 25 percent of the employed residents of one county commute to work in another
county; and

• At

least 25 percent of the employment (persons working) in one county are accounted for by
workers commuting from another county.

Labor market areas can vary in geographic scope, ranging from a single county to multi-county
metropolitan areas. LMAs can also span multiple states and in New England, they are composed of
cities and towns. The proposed rule requires states to group all areas within an LMA together
(leaving no areas out), but multi-state LMAs would require states to treat areas within their state
borders separately from the rest of its LMA.
Based on the conditions described earlier, the BLS labor market area definition only considers
aggregated commuting patterns between county of residence and county of employment — and
does not take into account sub-county variations by industry or by an individual’s socioeconomic
and demographic characteristics. It also does not take into account the many other dynamics beyond
commuting patterns that may impact an individual’s ability to find and secure a job, such as housing
and transportation, the location of new or future employment opportunities, the location of training
providers (such as community colleges), industry-specific recruitment practices, or regional
workforce or economic development strategies. It also does not reflect the ability of some
individuals to relocate within a state to pursue a job opportunity.
According to OMB guidance, the purpose of the Metropolitan and Micropolitan Statistical Area
standards is to provide nationally consistent delineations for collecting, tabulating, and publishing
federal statistics for a set of geographic areas. OMB establishes and maintains these areas solely for
statistical purposes and does not take into account or attempt to anticipate any non-statistical uses
that may be made of the delineations, nor will OMB modify the delineations to meet the
requirements of any nonstatistical program. OMB cautions that Metropolitan Statistical Area and
Micropolitan Statistical Area delineations should not be used to develop and implement federal,
state, and local non-statistical programs and policies without full consideration of the effects of
using these delineations for such purposes.229 While this does not preclude the Department or states
from using LMAs to inform the grouping of areas covered by waivers, the Department did not
explain the rationale for, and effect of, using LMAs as the only framework for grouping areas and its
reasons for excluding other methods for grouping.
In the following section, we discuss some of the other approaches for grouping areas that capture
regional dynamics that BLS LMAs don’t account for and that FNS did not explain whether it

228

Bureau of Labor Statistics, “Local Area Unemployment Statistics Geographic Concepts: Labor Market Areas,” 2018,
https://www.bls.gov/lau/laugeo.htm#geolma.
229

Office of Management and Budget, “Revised Delineations of Metropolitan Statistical Areas, Micropolitan Statistical
Areas, and Combined Statistical Areas, and Guidance on Uses of the Delineations of These Areas,” 2015,
https://www.whitehouse.gov/sites/whitehouse.gov/files/omb/bulletins/2015/15-01.pdf.

105

considered. We also discuss how FNS failed to address why it believes LMAs are preferable to the
many possible alternatives.
BLS LMA Definition Based on Outdated Data
BLS LMAs are limited in their ability to capture current and local workforce dynamics. They are
relatively static and do not account for sub-county variation. The BLS LMAs are revised each decade
following the census. The current list of BLS LMAs are based on population data from the 2010
Census and commuting data from the American Community Survey five-year dataset for 2006-2010
(issued on February 28, 2013, through OMB Bulletin No. 13-01).230 In other words, current BLS
LMAs reflect population data from nine years ago and commuting data from nine to 14 years ago.
The BLS LMAs are not updated frequently enough to capture current or recent labor market trends
and may not line up with more current labor force patterns.
The BLS LMAs are based on population and commuting data aggregated at the county level. The
commuting patterns between counties may vary depending on the industry or type of occupation.
For instance, commuting flows for workers working at an automotive assembly plant (which may be
relatively focused around the plant location) will tend to vary from commuting flows for workers in
food retail (relatively dispersed). Goetz and Han note that a given county may belong to multiple
commuting sheds and give the example of a commuter county on the east coast with residents who
commute to Washington, D.C., Philadelphia, and Baltimore and may be located in the border region
between the cores of multiple LMAs.231 Another example is Mercer County, Pennsylvania, which
could be considered part of Philadelphia’s LMA or New York’s LMA based on the commuting
patterns of residents. In these situations, it is not obvious which commuting regions or labor market
areas the county should be considered a part of and will vary depending on the industry or type of
worker.
Some BLS LMAs include areas from more than one state. For example, the Philadelphia LMA
includes counties belonging to Pennsylvania, New Jersey, Delaware, and Maryland. The proposed
regulatory language says that “the state agency may only combine data from individual areas that are
collectively considered to be a Labor Market Area by DOL.”232 The Department did not discuss
how states should handle multi-state LMAs, making it difficult for the public to comment on the
implications of using only BLS LMAs to group areas. Consider an LMA that spans two states, but
with 90 percent of its population residing in one state. Suppose the county with 10 percent of the
population has an unemployment rate that is slightly lower than the threshold needed to qualify for a
waiver, but the unemployment rate for the entire LMA exceeds the threshold needed to qualify.
Would that county qualify for a waiver, recognizing that the LMA that it belongs to may lack
sufficient jobs?
230

Office of Management and Budget, “Revised Delineations of Metropolitan Statistical Areas, Micropolitan Statistical
Areas, and Combined Statistical Areas, and Guidance on Uses of the Delineations of These Areas,” 2013,
https://www.whitehouse.gov/sites/whitehouse.gov/files/omb/bulletins/2013/b13-01.pdf.
Stephan J. Goetz and Yicheol Han, “Identifying Labor Market Areas Based on Link Communities,” paper prepared
for presentation at the 2015 Agricultural & Applied Economics Association and Western Agricultural Economics
Association Annual Meeting, San Francisco, CA, July 26-28, 2015, https://aese.psu.edu/nercrd/publications/publishedpapers/identifying-labor-market-areas-based-on-link-communities.
231

232

NPRM, p.992

106

Alternative Definitions of Labor Market Areas
Unlike a county or state, which are political and administrative units with defined borders, a labor
market area is an analytical concept and the definition used by BLS is only one of several ways that
labor economists and other researchers approximate labor market areas. “Researchers examining
labor markets in the United States often turn to one of several standard geographic definitions that
are widely known and compatible with publicly available economic data, including: states,
metropolitan areas, and counties.”233 Definitions of labor market areas that are based on single or
multiple counties, such as the one used by BLS, have the advantage of having unemployment data
readily available. However, the political or administrative boundaries that are used to delineate labor
market areas may not always align well with the notion of a labor market as “a set of relationships
between employers and workers that are spatially bounded by places of work and residence.”234 235
For example, a definition of a local labor market area that attempts to better capture an area in
which individuals both live and work is the Commuting Zone (CZ). CZs group counties based on
commuting flow data and hierarchical cluster analysis.236 Noting that there is no consensus definition
of LMAs, economists at the Economic Research Service, Scherpf et al., tested multiple definitions of
LMAs (BLS LMAs, Commuting Zones, and Workforce Innovation Areas) in their examination of
the relationship between labor market area conditions and length of SNAP participation spell.237
Their preferred definition of LMAs used the CZ definition and had the largest estimated effects.
They found that a 10 percent increase in county-level employment raised the share of recipients who
finished their SNAP spell in 12 months or less by about 5.3 percentage points (or about 8.8 percent).
Using alternative definitions of labor market areas resulted in smaller, but still positive, estimated
effects: a 10 percent increase in county-level employment raised the probability that a SNAP
recipient would finish a spell in 12 months or less by between 1.5 and 2 percentage points (or
between 2 and 3 percent).
Although CZs are delineated using more sophisticated analytical methods, they share a similar
limitation to BLS LMAs. Like BLS LMAs, CZs are based on commuting patterns and do not
account for other administrative or economic factors a state would like to consider when grouping
Andrew Foote, Mark J. Kutzbach, and Lars Vilhuber, “Recalculating... How Uncertainty in Local Labor Market
Definitions Affects Empirical Findings,” Center for Economic Studies Working Paper CES 17-49, August 2017,
https://www2.census.gov/ces/wp/2017/CES-WP-17-49.pdf.
233

Erik Scherpf et al., “Participation in USDA’s Supplemental Nutrition Assistance Program (SNAP): Effect of Local
Labor Market Conditions in Oregon,” United states Department of Agriculture (September 2018), pp. 1-50,
https://www.ers.usda.gov/webdocs/publications/90038/err-257.pdf?v=0.
234

Charles M. Tolbert and Molly Sizer, “U.S. Commuting Zones and Labor Market Areas: A 1990 Update, AGES9614,” U.S. Department of Agriculture, Economic Research Service, 1996,
https://usa.ipums.org/usa/resources/volii/cmz90.pdf.
235

236

Hierarchical cluster analysis is a method for exploring similarities between objects. An algorithm is used to group
similar objects into a cluster. Each cluster is distinct from other cluster and the objects within each cluster share similar
features.
237

Erik Scherpf et al., “Participation in USDA’s Supplemental Nutrition Assistance Program (SNAP): Effect of Local
Labor Market Conditions in Oregon,” United states Department of Agriculture (September 2018), pp. 1-50,
https://www.ers.usda.gov/webdocs/publications/90038/err-257.pdf?v=0.

107

areas covered by waivers. In proposing the BLS LMAs as the only acceptable framework for
grouping areas, the Department did not discuss alternative frameworks for grouping and the
implications of using a framework based on commuting data to shape SNAP policy.
Commuting Patterns Are Not the Only Factor Connecting Labor Markets
BLS LMAs only look at commuting patterns and ignore other economic factors that may be
related to spatial correlation of unemployment. Spatial correlation is a measure of the relationship
between “close” spatial units, such as neighboring counties. Using county-level monthly price data
from the real estate service Zillow, Fogli, Hill, and Perri examined trends in housing prices across
geographic areas. They observed that the housing price decline from early 2007 to early 2009
appeared to follow the same spatial patterns as unemployment. 238 Looking at the example of Florida,
they show that early 2007 prices fell in scattered locations around the coasts and over time prices fell
in nearby locations until they reached a uniformly low level across the state. Spatial diffusion of
housing prices and unemployment were strikingly similar across the whole period of housing boom
and bust, suggesting that housing prices might be one of the factors that states may implicitly or
explicitly consider when grouping areas.
The Department did not provide an explanation why it is restricting states from considering other
rationale that might be relevant to their SNAP population when requesting waivers. We strongly
encourage the Department to review the research we have outlined in this chapter related to
commuting patterns and to explain its rationale relative to the evidence that demonstrates the flaws
of this approach.
WIOA Regions
As discussed earlier, states may seek to align SNAP waivers with workforce development regions.
Under WIOA, states are required to identify regions for regional workforce planning. States shall
identify regions after consultation with elected officials and Local Workforce Development Boards
and take into account the following factors:
1. The extent to which regions are consistent with labor market areas in the state;
2. The extent to which regions are consistent with regional economic development areas in the
state; and
3. An assurance that regions have available the federal and non-federal resources necessary to
effectively administer activities under subtitle B and other applicable provisions of the
WIOA, including whether the areas have the appropriate education and training providers,
such as institutions of higher education and area career and technical education schools.
Under WIOA, states have discretion to define regions and are encouraged to take an integrated
approach to account for a range of different factors. States are encouraged to use LMAs as the
starting point for determining workforce development regions, but also need to consider workforce
and economic development framework, funding streams, and service (training) delivery.

238

Alessandra Fogli, Enoch Hill, and Fabrizio Perri, “The Geography of the Great Recession,” National Bureau of
Economic Research Working Paper 18447, October 2012, https://www.nber.org/papers/w18447.pdf.

108

E. States Have Used Their Discretion to Create Groupings Informed by
Multiple Factors
States use grouping methods, such as BLS LMAs and WIOA Local Workforce Development
Areas, as the starting point for developing workforce development plans and policy, but modify
them based on multiple additional factors. Below are some examples of how and why states group
areas using a variety of factors.
• When designating areas

for workforce development planning, the Virginia Board of
Workforce Development considers the BLS LMAs, regional economic development areas,
funding streams and service providers for training, community college regions, and industryand sector-specific strategies.239

• Rhode Island

has treated the entire state as a region for the purposes of workforce
development planning. The state considered geographic boundaries, LMA analysis, and
funding and resource realities in determining the geographic scope of its workforce
development plan. From its labor market area analysis, it found that of the 39 cities and towns
in Rhode Island, 36 fell within the “Providence-Warwick, RI-MA Metropolitan NECTA”
LMA as determined by the Bureau of Labor Statistics. One additional community was its own
LMA due to the fact it is an island, and two additional communities fall outside of the
Providence-Warwick, RI-MA Metropolitan NECTA. The governor concluded that the entire
state will be a single planning region for workforce development purposes. 240

• As

discussed earlier, Tennessee implements SNAP E&T in coordination with WIOA and
Local Workforce Development Areas. The delineation of these areas is different from the
BLS LMAs. For instance, Lauderdale County, located on the western border of Tennessee
north of Memphis, is its own BLS LMA, but is grouped with Shelby, Fayette, and Tipton
Counties into the Greater Memphis Local Workforce Development Area.

The Department did not explain why it was taking away states’ ability to consider multiple factors
when grouping areas covered by waivers. It did not discuss how the use of BLS LMAs would
improve states’ ability to meet the needs of individuals subject to the time limit and help move them
to self-sufficiency.
BEA Economic Areas
In two separate guidance memoranda (August 2006,241 December 2016242), the Department
provided an example of grouping based on BEA economic areas, yet the 2019 NPRM preamble did
not offer this framework for grouping areas for consideration. In the example provided, the
Virginia Board of Workforce Development, “Designation of Regions and Planning Requirements,” June 2016,
https://virginiacareerworks.com/wp-content/uploads/Policy-200-06-Designation-of-Regions-and-PlanningRequirements-FINAL-Signed.pdf.
239

Rhode Island, Governor’s Workforce Board, “Regional Planning Policy,” March 16, 2017, https://gwb.ri.gov/wpcontent/uploads/2017/06/17-01-3-16-2017.pdf?189db0.
240

241

USDA, “Guidance on Requesting ABAWD Waivers,” August 2006.

242

USDA, “Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents
(ABAWD),” December 2, 2016.

109

Department explained that Montana could group the counties of Blaine, Cascade, Chouteau,
Glacier, Hill, Liberty, Phillips, Pondera, Teton, and Toole that comprise the North Central Montana
Economic Area (or Area 65: Great Falls, MT Economic Area) and analyze the data to see if the
grouped area would qualify for a waiver. The guidance also suggested that the state could consider
grouping a sub-area such as Glacier, Liberty, and Toole. The Department did not explain why it is
eliminating this grouping approach, even though this was a grouping policy it has suggested that
states could use in the past.
Requiring Areas to Be Contiguous
Ignores the Reality That Proximity to Job Opportunities Is Decreasing
The proposed rule establishes BLS LMAs as the only scenario where states can request waivers
for combined geographic areas (counties). In doing so, the agency seeks to limit waivers that
combine areas that are not contiguous. This suggests an assumption that individuals will only
respond to job opportunities in their county or in counties adjacent to their county of residence.
From the perspective of workers in search of job opportunities, requiring contiguity of geographic
areas is an assumption that does not hold up under empirical scrutiny. Using county-level data for
eight states between 1969 to 1994, Khan, Orazem, and Otto found that local county population
responded to economic growth within the county, in adjacent counties, and even two counties
away.243 The effect decreased as the distance from the reference county increased. Workers look
beyond their county and adjacent counties for job opportunities.
Other research has found that proximity to jobs has decreased in metropolitan areas in recent
years and that poor, minority residents experienced a bigger decline in proximity to jobs compared
to non-poor white resident. Kneebone and Holmes looked at the number of jobs within a typical
commuting distance (median within-metro commuting distance) for residents of the 96 largest
metropolitan areas and found that the number of jobs within a typical commuting distance declined
by 7 percent between 2000 and 2012.244 Poor residents experienced a 17 percent decline in nearby
jobs compared to 6 percent for non-poor residents. Hispanic residents had a 17 percent decline and
black residents had a 14 percent decline compared to 6 percent for white residents. Individuals have
to look farther from their local neighborhoods for job opportunities, requiring longer commutes.
Requiring Areas to Be Contiguous or to Comprise Entire LMAs Ignores the Reality That
Unemployment Rates Rise and Fall at Different Rates Even in Neighboring Areas
The general unemployment rate does not account for variations in unemployment rates for subpopulations and for variations in the increase or decrease in unemployment rates across geographic
areas. The proposed rule would require that states request waivers for BLS LMAs in their entirety,
without omitting certain areas. The Department offers no rationale for proposing this and no
discussion of the implications of this arbitrary requirement. This requirement prevents states from
responding to variations in unemployment patterns within an LMA.
Romana Khan, Peter F. Orazem, and Daniel M. Otto, “Deriving Empirical Definitions of Spatial Labor Markets: The
Roles of Competing Versus Complementary Growth,” Journal of Regional Science (2001), pp. 735-756.
243

Elizabeth Kneebone and Natalie Holmes, ““The Growing Distance Between People and Jobs in Metropolitan
America,” Metropolitan Policy Program at Brookings (March 2015), pp. 1-24, https://www.brookings.edu/research/thegrowing-distance-between-people-and-jobs-in-metropolitan-america/.
244

110

Fogli, Hill, and Perri examined how the relationship in unemployment in neighboring areas
changes over time as a recession starts and ends.245 The relationship in unemployment between
neighboring areas (spatial correlation) is high overall, falls at the start of the recession, increases
sharply during the recession, and then stabilizes at the end of the recession. They found that
unemployment does not increase in all counties simultaneously, but initially increases in a few
specific counties, not necessarily located close to each other. As the recession deepens, the
geographic distribution of unemployment follows an epidemic pattern, with unemployment tending
to increase in counties that are closer to counties initially hit with high unemployment, so that
unemployment is high in some concentrated areas and relatively low in others, and this results in an
increase in the degree of spatial correlation and of spatial dispersion. As the recession reaches its
peak, high unemployment is spread all over the country and both the degree of spatial correlation
and spatial dispersion stabilize (and eventually decline).
The Department does not discuss the implications of requiring that states request waivers for BLS
LMAs in their entirety, without omitting areas. This requirement ignores variations in labor market
conditions within a labor market area. Consider an LMA that does not qualify for a waiver, but has
high unemployment everywhere except for one county. The proposed rules would prevent a state
from requesting a waiver for a sub-area of the LMA that lacks sufficient jobs, even if most of the
residents in that LMA reside in that sub-area and the sub-area had unemployment rates that met or
exceeded the threshold to qualify for a waiver. Even if the LMA qualified for a waiver, the state may
have reasons why it only wants to waive a sub-area of the LMA. For instance, the state may be able
to guarantee SNAP E&T slots in most areas of the LMA, but wants to request a waiver to cover
counties in the LMA where it cannot guarantee enough slots. Even if the areas the state wants to
waive qualified for a waiver based on unemployment rates, because it comprises only a part of the
LMA, it would not be eligible for a waiver.

F. Conclusion: We Recommend Rejecting Proposed Changes That Restrict
State Flexibility to Waive Groups of Areas
The proposed rule is based on insufficient reasons to change current regulations by prohibiting
states from seeking statewide waivers and from grouping areas, except for areas that are designated
as BLS labor market areas. It fails to discuss the ways that states have used existing flexibility to align
waiver policy with state operations, policy, and resources. It does not discuss the implications and
limitations of its proposed framework for grouping, nor does it address alternative methods for
grouping, including those suggested by the Department in past guidance. Given the lack of
supporting information, the public has an insufficient opportunity to comment meaningfully on the
proposed rule and we recommend rejecting the proposed changes to the rules.

Alessandra Fogli, Enoch Hill, and Fabrizio Perri, “The Geography of the Great Recession,” National Bureau of
Economic Research Working Paper 18447, October 2012, https://www.nber.org/papers/w18447.pdf.
245

111

Chapter 6. Taking Away Food Benefits from
Individuals Who Cannot Document 20 Hours a Week
of Work Will Not Increase Labor Force Participation
for This Population
USDA offered little rational for changing the decades-old criteria for requesting waivers of the
time limit. The primary reasoning it provided is that fewer waivers will result in more individuals
subject to the time limit. The agency believes that if the state threatens to withhold food benefits
from these individuals, they will work more and have higher participation rates in meaningful work
activities. The NPRM often describes this as a “belief.” For example, the NPRM states, “the
Department believes the local unemployment floor should be set at 7 percent to best meet its goals of
promoting self-sufficiency”246 (emphasis added). But the NPRM provides no evidence to support
the belief that taking away food from unemployed individuals will result in higher labor force
attachment or greater participation in job training. Because the NPRM includes no supporting data
or research, commentators are left to accept as unequivocally true that the time limit has an
instrumental role in moving “ABAWDs” from non-work to work.
Yet the claim that subjecting additional individuals to the time limit will result in more meaningful
work activities is wildly out of synch with what we know from the evidence. Research shows that a
significant share of individuals subject to the time limit work when they can find employment
(including while on SNAP) and will work after leaving SNAP even in the absence of the time limit.
The claim also ignores research showing that time limits generally fail to encourage employment.
And, the NPRM does not account for the particular challenges facing this population — barriers
and challenges to employment that differ from those faced by the general public and justify the
current approach to providing waivers and individual exemptions to unemployed childless adults.
Because the agency did not provide any evidence that would demonstrate the time limit is likely to
increase employment, earnings, or self-sufficiency, the agency’s claim that the time limit should be
applied to more individuals in order to increase labor force attachment is without merit. We offer
findings from numerous studies to illustrate our points. We strongly recommend that the
Department review and reflect on each of these studies before moving forward with a final rule.

A. Individuals Subject to the Time Limit Already Have Significant Work
Effort, Raising Doubt as to Whether the Rate Can Be Increased by
Withholding Food
The NPRM repeatedly claims that a primary goal of the proposed changes is to subject more
individuals to the time limit. Terminating food assistance to more people (some 755,000 more, by
the agency’s own estimates), the NPRM argues, will increase work effort, job placement, and
earnings among those subject to the rule. But it fails to include any information about the
employment of individuals subject, or potentially subject, to the time limit either while on SNAP or
before or after participation in the program. The absence of any information is striking because
research shows that many low-income adults without children who receive SNAP in any given
246

NPRM p. 984.

112

month work while on SNAP or soon after, regardless of the existence of the time limit. Other
factors, such as personal circumstances and local economic conditions, play important roles in an
individual’s employment, but are not taken into account by FNS in this proposed change.
In this section, we review the research that describes both the significant work effort of childless
adults without disabilities, as well as the unique challenges they face in the labor market. Because
workers turn to SNAP during periods of unemployment, employment rates among childless adults
while receiving SNAP might be expected to be relatively low. Still, in a typical month, almost a third
and perhaps as many as half of all SNAP households with childless adults work. USDA’s
administrative data show that 31 percent of all SNAP households with non-disabled childless adults
worked in a typical month of 2017.247 USDA’s administrative data may underestimate earnings
because some work may not be required to be reported for SNAP, either because it is irregular or
not expected to continue or because, under SNAP’s “simplified reporting” rules, changes in
circumstances need only be reported at six-month intervals unless they raise household income
above 130 percent of the poverty level. Work that households aren’t required to report for SNAP
purposes may be captured by the other data but not the SNAP data.
In fact, data from other sources suggest the work rate could be closer to 50 percent. Tabulations
generated by CBPP from the Census Bureau’s 2008 Panel of the Survey of Income and Program
Participation (SIPP) show employment rates for SNAP households with non-disabled childless
adults close to 50 percent in a typical month between 2011 and 2013.248
The 2017 data from USDA does not identify whether a childless adult is subject to the time limit
or not (because he or she is exempt or living in a waived area). But work rates among this
population have been consistent, even as the percentage of the country covered by waivers has
varied widely. Comparing 2017 data (when 37 percent of the population lived in a waived area) to
2014 data (when 75 percent of the population did) shows very similar levels of employment. In
2014, when the national unemployment rate was over 6 percent, 35 states had statewide waivers and
9 had areas waived. That year, 27 percent of all SNAP households with non-disabled adults and no
children worked in a typical month. 249 That is close to the 31 percent of non-disabled adults without
children who were working in 2017, when far more of the country had the time limit in place. This
suggests factors other than waivers or the time limit itself play a more important role in the ability of
adults without children to find work.
Even more important, childless adults have high work rates prior to and after a spell on SNAP,
but the NPRM fails to account for how subjecting more individuals to the time limit would impact
this. About 72 percent of SNAP households with a childless, working-age adult worked in the year

247

Food and Nutrition Service, Characteristics of Supplemental Nutrition Assistance Program Households: Fiscal Year
2017, Tables A.14 and A.16.
248

Center on Budget and Policy Priorities, “Unemployed adults without children who need help buying food only get
SNAP for three months,” https://www.cbpp.org/unemployed-adults-without-children-who-need-help-buying-foodonly-get-snap-for-three-months.
249

See Food and Nutrition Service, Characteristics of Supplemental Nutrition Assistance Program Households: Fiscal
Year 2014, Table A.16.

113

before or after receiving SNAP.250 And many of those not working while receiving SNAP were
actively looking for work. USDA’s administrative data suggest that, in a typical month in 2017, close
to half (46 percent) of all childless adults on SNAP who were not working were looking for work
(and of those who reported not participating in the labor force, many likely had health conditions or
other barriers to employment that prohibited labor force participation). Although this statistic
should be viewed with caution as the data may not be sufficiently reliable to draw firm conclusions,
it’s noteworthy that Urban Institute researchers, using National Survey of American Families data,
found that three-quarters of all low-income, able-bodied adults without dependents (not just those
on SNAP) worked in 1997 and 86 percent were in the labor force (that is, either working or actively
looking for work).251
Given the consistent evidence of work among individuals likely to be subject to the time limit
regardless of waiver status or general unemployment rates, the claim that the time limit itself leads to
increased employment is not supported. Instead, it suggests that the rule’s sole purpose is to take
away food assistance from struggling unemployed or underemployed workers.
Further undermining the assertion that the time limit is necessary to increase work effort among
SNAP participants, most childless adults on SNAP who work have substantial work. Among SNAP
households that worked in a typical month while receiving SNAP or worked at some point during
the following year, nearly half (49 percent) worked full time (at least 35 hours a week) for six months
or more of the following year. Twelve percent worked at least 20 hours per week for six or more
months. Another 24 percent worked full time in at least one month over that period. Only about
15 percent worked 20 or more hours per week for less than six months or worked fewer hours than
that.252
Nonetheless, childless adults participating in SNAP are generally low-income, low-skill workers
with limited job prospects. More than 80 percent have no more than a high school education or
GED. They are more likely than other SNAP participants to lack basic job skills like reading,
writing, and basic mathematics.253 A work experience program in Ohio designed to help individuals
subject to the time limit find work or qualifying work activities found signs of functional illiteracy
250

Center on Budget and Policy Priorities, “Unemployed adults without children who need help buying food only get
SNAP for three months,” https://www.cbpp.org/unemployed-adults-without-children-who-need-help-buying-foodonly-get-snap-for-three-months.
251

See Stephen Bell and Jerome Gallagher, “Prime-Age Adults without Children or Disabilities: The ‘Least Deserving of
the Poor’ — or Are They? Assessing the New Federalism Policy,” Urban Institute, February
2001, https://www.urban.org/sites/default/files/publication/61286/310269-Prime-Age-Adults-without-Children-orDisabilities.PDF .
252

Steven Carlson et al., “Who Are the Low Income Childless Adults Facing the Loss of SNAP in 2016?” Center on
Budget and Policy Priorities, February 8, 2016, p. 9, https://www.cbpp.org/sites/default/files/atoms/files/2-816fa.pdf. An updated analysis that looked at all SNAP households with an adult without disabilities (with and without
children), found similar rates of full-and part-time work among the broader group that the earlier analysis found. See
Brynne Keith-Jennings and Raheem Chaudhry, “Most Working-Age SNAP Participants Work, But Often in Unstable
Jobs,” March 15, 2018, https://www.cbpp.org/research/food-assistance/most-working-age-snap-participants-work-butoften-in-unstable-jobs.
253

“Food Stamp Employment and Training Program,” United States General Accounting Office (GAO—3-388), March
2003, p. 17, http://www.gao.gov/assets/240/237571.pdf.

114

even among those with a high school degree. 254 As a result, wages of childless working-age adults on
SNAP are quite low: one study found that 90 percent of those aged 25-49 earned less than twice the
minimum wage, compared to 47 percent of all workers aged 25-49.255
USDA did not provide any information or analysis in the NPRM that suggested it had reviewed
this evidence. Nor did it offer any research to the contrary — and we believe no such research
exists. That leaves us to wonder whether the Department was aware that the work rates for
individuals subject to the time limit appear to be similar whether or not they live in a waived area.

B. The NPRM Fails to Account for the Distinct Characteristics of
Unemployment Childless Adults on SNAP
Because the group of individuals subject to the rule are, when compared to the general public,
poorer, less educated, and more likely to have medical conditions or other factors affecting their
ability to find employment, states have long found the three-month time limit in statute harsh and
unfair. To mitigate the harm caused by taking food assistance away from this group, states have
routinely relied on the option to request a waiver based on demonstrating a lack of sufficient jobs
for the individuals affected by the time limit.
The NPRM seeks to restrict this state option in order to end food assistance for unemployed
adults without children in the hope that this will result in increased employment among this group.
The NPRM offers no information about the individuals affected by the rule — whether they have
the skills, training, and support to find and keep employment, and whether they disproportionately
face barriers to work like undiagnosed health conditions, a lack of transportation, or a criminal
history. The NPRM neglects to address research findings from multiple sources, including USDA
itself, showing that this population does face a different labor market environment than the general
public. We are surprised that USDA did not draw upon this wide body of research and urge the
agency to review the studies we cite — all of which are included in the appendix.
We review the research to show that the job prospects for these individuals are not accurately
captured by the general unemployment rate. Even when unemployment is low due to a strong
economy, adult SNAP participants without children face a very different labor market than higherincome adults. This population is also underserved by other support programs, often lacks stable
housing, and struggles to be hired into stable jobs. States wisely use the flexibility provided by law
to assess the availability of jobs for this population and request waivers when there are insufficient
jobs.
Childless adults on SNAP are extremely poor. Like many others, childless adults often turn to
SNAP for assistance when they are no longer able to make ends meet, especially if their jobs are lost,
254

See “A Comprehensive Assessment of Able-Bodied Adults Without Dependents and Their Participation in the Work
Experience Program in Franklin County, Ohio,” Ohio Association of Foodbanks,
2014, http://admin.ohiofoodbanks.org/uploads/news/WEP-2013-2014-report.pdf.
255

Stephen Bell and Jerome Gallagher, “Prime-Age Adults without Children or Disabilities: The ‘Least Deserving of the
Poor’ — or Are They? Assessing the New Federalism Policy,” Urban Institute, February
2001, https://www.urban.org/sites/default/files/publication/61286/310269-Prime-Age-Adults-without-Children-orDisabilities.PDF .

115

hours are cut, or wages hover at the federal minimum. While participating in SNAP, their income
averages 33 percent of the poverty line, the equivalent of about $4,000 per year for a single person in
2019. Average incomes are even lower — just 18 percent of poverty — for those not working 20
or more hours a week, who are most likely to be cut off due to the three-month limit.256
The struggles of poor adults are vividly portrayed in $2.00 a Day: Living on Almost Nothing in
America, which draws detailed portraits of households with little access to substantial employment
and public benefits.257 The descriptions illustrate the complexities of poverty – the psychological
and emotional costs, the uncertainty and lack of options available, and the work effort of those in
poverty. We strongly urge the Department to familiarize itself with the real lived experiences of
those portrayed in the book.
Unemployed childless adults have few resources other than SNAP to rely on while looking for
work. In general, adults without children are not eligible for most government assistance. In the
past, state General Assistance programs have provided small monthly cash allotments to singe adults
to meet basic shelter and other needs, but these programs have weakened considerably. Few
childless adults qualify for unemployment insurance, and childless adults are ineligible for Medicaid
in states that haven’t adopted the Affordable Care Act’s Medicaid expansion. In addition, childless
workers are the only demographic group that the federal tax system taxes into, or deeper into,
poverty, in part because they are eligible only for a tiny Earned Income Tax Credit (EITC). Federal
income and payroll taxes pushed about 7.5 million childless workers into or deeper into poverty in
2015.258 Given how little federal support is available to the group subject to the time limit, it is
surprising that USDA believes the group is able to survive while avoiding work. Their SNAP
benefits are minimal to meet basic food needs, let alone housing, health, and other basic expenses.
Many childless adults have disabilities that make working difficult or impossible but don’t meet
the severe disability standard for receiving Supplemental Security Income (SSI) or Social Security
Disability Insurance (SSDI). If not identified as having a physical or mental condition that prevents
them from working 20 or more hours per week, they would be subject to the time limit, yet unable
to realistically find work in many cases.
There’s more evidence that people subject to the time limit face multiple challenges to
independence and self-sufficiency, including homelessness, physical and mental health limitations,
language barriers, unstable employment histories, and criminal records. A detailed study of

256

CBPP analysis of FY 2017 USDA SNAP Household Characteristics data adjusted to FY 2019 dollars.

257

Kathryn J. Edin and H. Luke Shaefer, $2.00 a Day: Living on Almost Nothing in America, 2015, Houghton Mifflin
Harcourt. We especially highlight descriptions of the struggles of finding and keeping work (pp. 42-47 and 112-114), the
challenges facing people of color in the job market (pp.52-56), and barriers to employment like transportation (pp. 5152, 138-139).
258

Chuck Marr et al., “Lone Group Taxed Into Poverty Should Receive a Larger EITC,” Center on Budget and Policy
Priorities, April 19, 2016, https://www.cbpp.org/research/federal-tax/childless-adults-are-lone-group-taxed-intopoverty.

116

individuals identified by the local SNAP agency as ABAWDs subject to the time limit who were
referred to a work experience program in Franklin County (Columbus), Ohio found that: 259
• Many

have extremely unstable living situations, evidenced by residence in short-term shelters
or with friends and family and limited telephone service.

• One-third

have a mental or physical limitation, including depression, post-traumatic stress
disorders, mental or learning disabilities, or physical injuries. Some of these disabilities,
though not severe enough to qualify for federal disability benefits, may still limit a person’s
ability to work more than 20 hours a week.

• Nearly

one-quarter are non-custodial parents, and 13 percent are caregivers for a parent,
relative, or friend.

• More

than 40 percent lack access to reliable private or public transportation; 60 percent lack a
valid driver’s license.

• Fifteen percent

need supportive services like language interpretation or help with
transportation to obtain employment.

• Nearly

one-quarter have been dismissed from a job in the past and others have gaps in their
employment records — both of which can deter potential employers. More than one-third
have felony convictions, making it hard to find jobs and pass background checks.

These individuals face daunting challenges in finding employment even when general
unemployment rates are low. The Ohio study illustrates why Congress gave states the option to
waive the time limit in areas where there are insufficient jobs for those subject to the rule. Without
providing any evidence to the contrary, the NPRM proposes to limit the ways in which a state can
demonstrate a lack of sufficient jobs for the individuals subject to the time limit. It does this by
eliminating Labor Surplus Areas, low and declining employment-to-population ratios, seasonal
unemployment and requiring recent unemployment rates to be at least 7 percent. But the
Department fails to explain how it determined that the proposed new standards relate to
employment opportunities for those subject to the rule, particularly given the characteristics outlined
in the list above.
One reason states were given flexibility to define the areas which could be waived due to a lack of
sufficient jobs for the individuals subject to the rule is that even in the late 1990s, a growing body of
research showed that the labor market situation for low-skilled workers had grown worse over time,
and that low-skill workers faced limited employment options. As summarized by a report
commissioned by USDA in 1998, “a relatively large body of research indicates that the labor market
situation of the low-skilled has become considerably worse in recent decades and that their current
employment prospects are limited. This suggests that even if ABAWDs are willing to work, they

259

See “Comprehensive Report on Able-Bodied Adults Without Dependents, Franklin County Ohio Work Experience
Program,” Ohio Association of Foodbanks,
2015, http://admin.ohiofoodbanks.org/uploads/news/ABAWD_Report_2014-2015-v3.pdf. The Ohio Association of
Foodbanks gathered the information for the report as a result of a partnership with the county SNAP agency to help
place individuals identified as subject to the time limit in qualifying work activities after screening them.

117

may be unable to do so because there are not enough jobs for low-skilled workers.”260 The report,
which reviewed studies on the employment prospects of low-income adults fitting the “ABAWD”
description (but not necessarily participating in SNAP), also found that:
• Job prospects

for ABAWDs do not look promising, due to changes in the U.S. economy that
have resulted in the decline of industries and skill types in which ABAWDs are concentrated.

• Many

ABAWDs face a spatial mismatch between their residence and the location of low-skill
jobs, as well as a skills mismatch, especially for urban residents.

• The

job prospects of ABAWDs depend significantly on local economic conditions, tied not to
county-level unemployment but to the location of employers needing low-skill workers and
the quality and availability of local institutions supporting workforce development. 261

The Department’s commissioned reports as well as other research paint a clear picture of
individuals in this targeted group who have common characteristics that distinguish the group from
other unemployed adults. These characteristics — including high poverty rates, health issues, and
few supports — make finding and keeping employment a unique challenge. The Department simply
asserts that the time limit will increase employment for this population but does not acknowledge its
own research showing that this is not the case. While all aspects of the rule strike us as arbitrary, this
disconnect between the agency’s basic knowledge of the affected population and the assertions
about how the proposed policy would increase employment is particularly surprising. This is one of
numerous reasons why the proposed rule should be withdrawn.

C. Reports Commissioned by USDA Show Subjecting More Individuals to
the Time Limit Will Not Increase Their Likelihood of Gaining
Employment
In order to understand the impact of the time limit on individuals’ well-being, USDA, along with
the U.S. Department of Health and Human Services, funded studies in four states: Arizona, Illinois,
Iowa, and South Carolina. While the studies varied in scope and focus and were not able to identify
individuals who left SNAP because of the time limit, they do reveal the policy’s limited impact on
employment outcomes, coupled with low earnings and increased hardship.
A 2001 study of individuals leaving SNAP in Illinois showed that far more families than
ABAWDs left SNAP due to increased earnings, even though the time limit was in effect. The study
found that the percentage of ABAWDs who left SNAP and remained off the program and
employed did not vary between counties that had waivers and counties that did not. 262

260

Michael Stavrianos and Lucia Nixon, The Effect of Welfare Reform on Able-Bodied Food Stamp Recipients, Mathematica
Policy Research, Inc., July 1998, pp. 56-57.
261

Michael Stavrianos and Lucia Nixon, The Effect of Welfare Reform on Able-Bodied Food Stamp Recipients, Mathematica
Policy Research, Inc., July 1998, p. 56-57.
262

See Exhibit ES-1, page ES-4, Philip Richardson et al., Food Stamp Leavers Research Study – Study of ABAWDs
Leaving the Food Stamp Program in South Carolina, https://naldc.nal.usda.gov/download/45220/PDF.

118

The study looked at those who left SNAP in areas where the time limit was in effect and areas
where it was not (due to a waiver). It found that in the sample who left SNAP in 1998-1999 , a
majority of individuals (53 percent) were off SNAP in 2001 and not working or back on SNAP. Of
those that were working after leaving SNAP, the study did not attempt to identify the role of the
time limit in employment gains. In fact, the study shows that employment rates among ABAWDs
leaving SNAP were highest in areas exempt from the time limit due to high unemployment rates —
higher than the employment rates in areas using the individual exemptions and areas in which the
time limit was in effect.263 This shows that factors other than the time limit have more impact on
employment outcomes for ABAWDs. FNS fails to address this in the NPRM. Here we again
recommend that FNS review and consider this research.
The Arizona study commissioned by USDA to understand the outcomes of people leaving SNAP
showed that ABAWDs had worse outcomes on a number of employment-based metrics. They were
less likely to have achieved self-sufficiency, less likely to have improved their employment situation,
and more likely to be at risk of hardship or deprivation. The study grouped SNAP participants in
Arizona in 1997 into three categories: ABAWDs, non-ABAWD individuals on TANF, and nonABAWD, non-TANF adults. Upon leaving SNAP, employment rates for each subgroup were over
50 percent, but by one year later, ABAWDs who had been cut off SNAP had experienced the
greatest drop in employment. 264 This suggests that: (1) the three-month time limit may not be an
important factor in causing employment, since SNAP participants not subject to the time limit had
similar levels of employment as those subject to the time limit when leaving SNAP; and (2)
ABAWDs struggle to maintain employment. The report identifies several reasons why the ABAWD
group might see a sharp decline in employment. In a survey of ABAWDs who were no longer on
SNAP and not working, more than 60 percent reported being ill or having health problems or a
disability — circumstances not identified by the SNAP agency as qualifying an individual to an
exemption. 265
Across all four studies, many who lost SNAP benefits were employed but had low earnings. Between 41 and 76
percent of the former recipients in the four states were working after leaving SNAP, but earnings
were low and many remained in poverty.266 Most of those who were not working in Arizona and
Illinois were in poor health or caring for a family member in poor health. Given the complexity of

263

See Exhibit IV-I, Philip Richardson et al., Food Stamp Leavers Research Study – Study of ABAWDs Leaving the
Food Stamp Program in South Carolina Final Report, page IV-2, https://naldc.nal.usda.gov/download/45220/PDF.
264

See Exhibit 2-6, Covered Employment, in Gregory Mills and Robert Kornfeld, Study of Arizona Adults Leaving the
Food Stamp Program, Final Report, Dec. 2000, page 36, https://naldc.nal.usda.gov/download/45673/PDF.
265

Gregory Mills and Robert Kornfeld, Study of Arizona Adults Leaving the Food Stamp Program, Final Report, Dec.
2000, page 51, https://naldc.nal.usda.gov/download/45673/PDF. Individuals with a medically certified physical or
mental condition that prevents them from working are exempt from the three-month time limit, but states have
struggled with correctly identifying these individuals. The study did not look at whether some of the ABAWDs who left
SNAP should have been exempt from the time limit.
266

Elizabeth Dagata, “Assessing the Self-Sufficiency of Food Stamp Leavers,” Economic Research Service, USDA,
September 2002, https://www.ers.usda.gov/publications/pub-details/?pubid=46645.

119

SNAP’s rules governing work effort, disability, and time limits, some may not have realized that they
may again qualify for SNAP and failed to reapply for benefits.267
While the studies of individuals leaving SNAP looked at all types of individuals, the studies yielded
important results for ABAWDs that are relevant to this proposed rule but do not appear in the
Department’s rationale for the NPRM. For example, in Illinois in 1997, the single largest category
of individuals losing benefits was ABAWDs. The impact varied widely among groups. For example,
two-thirds of the ABAWDs leaving the program were African American, well above the percentage
of all leavers (which was 50 percent).268 As discussed more fully in Chapter 12, the NPRM
acknowledges the proposed rule would have a disparate impact by noting the “potential for
disparately impacting certain protected groups due to factors affecting rates of employment of
members of these groups.”269 But the Food and Nutrition Act makes clear that
the regulations implementing Title VI and other civil rights statutes are fully applicable to
SNAP. These regulations prohibit actions in federal programs that have disparate impacts on
members of protected groups as well as intentional discriminatory acts. Therefore, the proposed
rule’s disparate impact on these individuals, as demonstrated by the research and acknowledged in
the NPRM itself, is inconsistent with the Act. Given the requirement under 7 U.S.C. § 2020
(c)(2)(D) that the Department ensure the protections of Title VI of the Civil Rights Act of 1964
apply to all SNAP households, we are stunned that the Department did not review its own research
results that clearly suggest that the proposed policy would have widely disparate impacts on African
Americans.

D. The Loss of SNAP Due to the Time Limit Fails to Raise Income and
Increases Hardship
Childless adults who lose SNAP benefits struggle without food assistance benefits. As noted,
USDA’s most comprehensive assessment of former SNAP recipients in four states in the early
2000s suggests that their life circumstances are quite difficult. 270 A significant minority don’t find
work, and among those who are employed after leaving SNAP, earnings are low. Most remain
poor. Many struggle to acquire enough food to meet their needs, lack health insurance, experience
housing problems, and/or have trouble paying their bills. (These studies include people who leave
SNAP because of the three-month time limit or for other reasons, for example, because they found
a job or mistakenly believed they were no longer eligible.)
Despite relatively high levels of work effort, between one-third and roughly two-thirds of SNAP
leavers in the four states had household incomes below the poverty line — well above the overall
267

Elizabeth Dagata, “Assessing the Self-Sufficiency of Food Stamp Leavers,” Economic Research Service, USDA,
September 2002, https://www.ers.usda.gov/publications/pub-details/?pubid=46645.
268

Anu Rangarajan and Philip M. Gleason, Food Stamp Leavers in Illinois: How Are They Doing Two Years Later?,
Mathematica Policy Research, Inc., January 2001, p. 21, https://www.mathematica-mpr.com/our-publications-andfindings/publications/food-stamp-leavers-in-illinois-how-are-they-doing-two-years-later.
269

NPRM, p. 990.

270

Elizabeth M. Dagata, “Assessing the Self-Sufficiency of Food Stamp Leavers,” Economic Research Service, USDA,
September 2002, https://ageconsearch.umn.edu/record/262256/files/31106_fanrr26-8_002.pdf. This is a summary of
in-depth studies in Arizona, Illinois, Iowa, and South Carolina.

120

poverty rate of 13 percent. Many of these households experienced severe poverty after leaving
SNAP: about 40 percent of the leavers in two states were below half of the poverty line.
Many struggled to acquire adequate food. Between 17 and 34 percent of the SNAP leavers in the four
states reported very low food security (meaning they had to skip or reduce the size of their meals or
otherwise disrupt their eating patterns at times during the year because they couldn’t afford enough
food), compared with 3 percent of all households without children.
USDA’s study of adults in Arizona leaving SNAP found that the incidence of moderate or severe
hunger was greatest among the ABAWD subgroup, at 34 percent, compared to 23 percent for the
TANF subgroup and 18 percent for the non-TANF subgroup. 271 By comparison, 3.5 percent of all
Arizona households were classified as facing moderate or severe hunger. The Arizona study
concludes by pointing out that individuals who might appear to be self-sufficient or better off after
leaving SNAP, because they receive fewer public benefits and report less private support, might still
be facing significant hardship:
The high rate of food insecurity with hunger found among ABAWD exiters —
34 percent — is noteworthy. This incidence is more than twice the 1999
national rate of 14 percent estimated by USDA for households at or below 50
percent of the poverty level, even though most ABAWDs have incomes above
the poverty level. The ABAWD finding highlights the importance of
considering (in this and other exit studies) whether exiters who appear selfsufficient, in terms of their reduced reliance on public and private support, are
able to avoid hardship and deprivation.272
This cautionary note is not found in the NPRM. The agency simply asserts that it believes
expanding the number of individuals subject to the time limit will improve their self-sufficiency,
without acknowledging what the agency-funded research revealed — that the outcome for
ABAWDs leaving because of the time limit may be increased hunger and hardship.
The studies also showed that many lacked health insurance, had housing problems, or had trouble
paying their utility bills. About 30 to 40 percent of the SNAP leavers in the four states faced
housing issues, including falling behind on rent, moving in with relatives, or becoming
homeless. Between 20 and 65 percent reported problems paying for utilities. Just over half of the
SNAP leavers in two of the states were uninsured.273 These are all characteristics, as we illustrate
elsewhere in the comments, associated with much higher rates of unemployment. We continue to
be baffled about how or whether USDA factored its own research into the NPRM. It would appear
that the Department ignored its own studies.

271

Gregory Mills and Robert Kornfeld, Study of Arizona Adults Leaving the Food Stamp Program, Final Report, Dec.
2000, page 85, https://naldc.nal.usda.gov/download/45673/PDF.
272

Gregory Mills and Robert Kornfeld, Study of Arizona Adults Leaving the Food Stamp Program, Final Report, Dec.
2000, page 94, https://naldc.nal.usda.gov/download/45673/PDF.
273

Elizabeth M. Dagata, “Assessing the Self-Sufficiency of Food Stamp Leavers,” Economic Research Service, USDA,
September 2002, https://www.ers.usda.gov/publications/pub-details/?pubid=46645. This is a summary of in-depth
studies in Arizona, Illinois, Iowa, and South Carolina.

121

E. Evidence From Other Benefit Programs Shows That Time Limits Do Not
Increase Employment and Have a Disproportionate Impact on Certain
Populations
The stated rationale for proposing a change to the long-standing waiver process is to expose more
individuals to the time limit, in the belief that this will result in increased labor market attachment.
The NPRM provides no evidence to support this assertion. But, research on the Temporary
Assistance for Needy Families (TANF) program, which imposes both work requirements and a time
limit for benefits, undermines the claim that work requirements and time-limited benefits increase
employment.
A review of the many studies on families whose TANF monthly direct financial support was
reduced or taken away due to work requirements shows that these policies harm individuals and
families, most of whom face significant obstacles to employment, while producing few lasting gains
in employment. While the studies described in this section are about families with children — often
single-mother-headed households — the findings are relevant because they review the circumstances
of very poor households, similar in important respects to households on SNAP. In addition, many
ABAWDs are parents of non-minor children or have children not in the SNAP household. TANF
and SNAP households live in similar circumstances and face similar challenges finding employment,
so the outcomes from one group can inform the likely outcomes for the other.
A time limit ignores the fact that public assistance recipients often vary in their needs and
circumstances; many often live with one or multiple significant barriers to employment. Those
barriers range from low cognitive functioning, to mental and physical health problems, to criminal
justice issues, to low measures of human capital. These barriers make it hard to find or keep a job or
fulfill other work requirements. Extensive research on the effect of time limits in TANF, which
provides modest cash assistance and requires a portion of the caseload to engage in work activities,
shows that the time limit did not notably increase employment, but it did result in increased
hardship. We strongly recommend that USDA read these studies and consider the findings when
developing final regulations. We are confident the results demonstrate the problems with the
proposed policy.

F. Most Low-Income People Sanctioned From Public Assistance Due to
Work Requirements Face Barriers Finding Employment
Studies show that many parents who lose TANF benefits due to work requirements have
significant employment barriers. Those losing benefits are more likely than other TANF parents to
have physical, mental health, or substance use issues; to be fleeing domestic violence; to have low
levels of education and limited work experience; or to face significant logistical challenges, such as
lack of access to or funds to pay for child care and transportation.274 Below are summaries of several
TANF studies finding that many adults losing assistance due to sanctions face significant
employment barriers:

274

LaDonna Pavetti, Michelle K. Derr, and Heather Hesketh, “Review of Sanction Policies and Research Studies,”
Mathematica Policy Research (March 2003).

122

•A

2007 study of regional variation of full-family sanctioning practices in Florida’s TANF cash
assistance program gives evidence that sanctions are significantly related to various barriers to
employment. TANF parents with lower income and lower levels of education were more
likely to be sanctioned than those with higher income and education levels. For example,
recipients with a high school degree were more likely to be sanctioned than those with more
education — however, sanctions were the most common among those with less than a high
school degree. Community traits can matter too: families were more likely to be sanctioned in
counties with higher poverty rates than other counties, after controlling for other
characteristics.275

•A

2006 study of women in Wisconsin receiving TANF found that the state was more likely to
sanction mothers with lower levels of education. Specifically, mothers with at least a high
school diploma or equivalent were less likely to be sanctioned than mothers with less than a
high school diploma, and those with education beyond high school were even less likely to be
sanctioned, even when controlling for how long each individual received TANF grants. The
study also concluded that those “who may be least able to succeed in the labor market are
most likely to be sanctioned.” Specifically, the authors examined sanction activity against the
mothers’ employment status in the two years preceding entry to the TANF program. The
authors’ estimates show a monotonic trend with the number of quarters of work: those with
no work in the past two years were most likely to be sanctioned, those with 1-4 quarters of
employment were less likely to be sanctioned, those with 5-7 quarters of employment were
even less likely to be sanctioned, and mothers who had been employed for all eight quarters
were the least likely to be sanctioned.276

•A

2002 comparison of sanctioned and non-sanctioned TANF recipients in Boston, Chicago,
and San Antonio found sanctioned recipients were less likely than other TANF recipients to
have a high school degree or its equivalent, a working telephone at home, or a car. They were
more likely to report being in fair or poor health, have a substance use issue, or have a partner
who interfered with their employment, training, or schooling. They also had less work
experience, lived in neighborhoods with undesirable qualities (such as abandoned houses,
assaults and muggings, gangs, and open drug dealing), and reported living in housing of poor
quality.277

•A

study of women in Michigan receiving TANF shows that those with educational barriers to
employment were more likely to be sanctioned than those without such barriers. Women with
less than a high school education were 2.06 times more likely to be sanctioned than women
with higher levels of formal education.278

275

Richard C. Fording, Joe Soss, and Sanford F. Schram, “Devolution, Discretion, and the Effect of Local Political
Values on TANF Sanctioning,” Social Service Review (June 2007), pp. 285-316.
276

Chi-Fang Wu et al., “How Do Welfare Sanctions Work?,” Social Work Research (March 2006), pp. 33-50.

277

Andrew J. Cherlin et al., “Operating within the Rules: Welfare Recipients’ Experience with Sanctions and Case
Closings,” Social Service Review (September 2002), pp. 387-662.
278

Ariel Kalil, Kristin S. Seefeldt, and Hui-chen Wang, “Sanctions and Material Hardship under TANF,” Social Service
Review (December 2002), pp. 642-662.

123

• The

same study of women in Michigan receiving TANF shows those with transportationrelated barriers to employment were more likely to be sanctioned than those without such
barriers. Specifically, recipients who lacked either a car or a driver’s license were
disproportionately sanctioned relative to recipients without these transportation barriers. In
addition, those with trauma-related barriers to employment were more likely to be sanctioned
than those without such barriers. Specifically, women who reported experiencing severe
domestic violence (being hit, kicked, shoved, etc.) within the past year were disproportionately
sanctioned relative to recipients who did not report experiences of that type.279

•A

California study of a random sample of CALWORKs (TANF) recipients in four counties
(two large urban counties and two large semirural counties) found that recipients without a car
were roughly 1.5 times more likely to incur sanctions than recipients who owned a car.
Recipients who were sanctioned were less likely to own a car (39 percent of respondents) in
comparison to non-sanctioned recipients (52 percent of respondents).280

• In Illinois,

parents who had ever been sanctioned were significantly less likely than those never
sanctioned to have a high school diploma or its equivalent and more likely to have limited
recent work experience. They also were significantly more likely to be dealing with a physical
or mental health issue, to have been arrested multiple times, and to have experienced a child
care issue. In South Carolina, parents ever sanctioned were significantly more likely to have a
physical health problem, show signs of a learning disability, and have a family member or
friend with a health care issue or special need.281

•A

study of 656 TANF leavers from the 1999 and 2002 data of the National Survey of
America’s Family (NSAF) found that TANF leavers who reached their lifetime limits had a
higher chance of having problems with employment due to work barriers or vulnerable
characteristics such as old age, physical or mental health problems. They experienced greater
hardship because they had less income and less EITC receipt, most had experienced a cutoff
of SNAP, and fewer received child care assistance. The author concluded that time limits can
lead low-income families to endure greater economic hardships.282

•A

study that examined time limits on the receipt of welfare in both the United States and
British Columbia, Canada found that time limits are an ineffective policy tool as they increase
barriers to employment and result in recipients needing more support and access to specific
programs. Recipients who exhausted their benefits struggled financially and had a difficult
time finding a job. While some recipients had several barriers to work, others who had fewer

279

Ariel Kalil, Kristin S. Seefeldt, and Hui-chen Wang, “Sanctions and Material Hardship under TANF,” Social Service
Review (December 2002), pp. 642-662.
280

Yeheskel Hasenfeld, Toorjo Ghose, and Kandyce Larson, “The Logic of Sanctioning Welfare Recipients: An
Empirical Assessment,” Social Science Review (June 2004), Vol. 78, No. 2, pp. 304-319,
http://repository.upenn.edu/spp_papers/88/
281

LaDonna Pavetti et al., “The Use of TANF Work-Oriented Sanctions in Illinois, New Jersey, and South
Carolina,” Mathematica Policy Research (April 2004).
282

Kyoung Hag Lee, “Effect on Lifetime Limits and Differences between TANF Leavers Who Had Reached Their
Lifetime Limits and Those Who Had Exited Voluntarily,” Poverty & Public Policy (2010), pp. 1-22.

124

barriers were still unable to find a job. Low cognitive functioning, limited education, and
physical and mental health problems were some of the barriers that recipients faced.283
•A

2006 study of Minnesota’s TANF program explored in detail the circumstances of those at
or near 60 months, the federal time limit. It found about 16 percent of cases had a case head
with an IQ of less than 80, about 20 percent were caring for ill or incapacitated family
members, and about 21 percent were ill or incapacitated themselves for 30 days or more.
Long-term TANF recipients also had mental illness (which is often untreated or inadequately
treated), were developmentally disabled or learning disabled, were leaving domestic violence
situations, or were otherwise “unemployable.” Some recipients suffered from chronic and
debilitating health problems because they had worked physically demanding jobs. 284

• Another

report on Minnesota’s TANF program shows the percent of persons with a severe
mental health diagnosis at 60 months, the federal time limit, to be about 52 percent and those
with a chemical dependency diagnosis to be 27 percent. Among American Indian recipients,
about 60 percent of those near the time limit had a mental health diagnosis and/or a chemical
dependency diagnosis.285

•A

more recent Washington State study compared families who left TANF due to time limits
and those who left for other reasons. The state found that time-limited families were more
likely to be unstably housed in the year prior to losing assistance. They were also more likely
to have chronic health issues and visit the emergency room. And, they were more likely to
have a range of behavioral health needs, from mental health issues to substance abuse
disorders.286

• In a

survey of 276 West Virginia TANF recipients cut off due to time limits, respondents
identified several barriers to employment. Most of the respondents were unemployed after
leaving TANF. More than half (56.2 percent) of respondents were not working because of a
physical or mental illness or disability problem; more than a third (37.1 percent) had no
transportation; a third (32.6 percent) did not have the right education; and a little less than a
third (29.8 percent) simply could not find a job. Most of these respondents had multiple
barriers to employment. 287

283

Dean Herd, Ernie Lightman, and Andrew Mitchell, “Welfare Time Limits: Symbolism and Practice”, U-Toronto: Social
Assistance in the New Economy (SANE) Project (2008), pp. 1-26.
284

LaDonna Pavetti and Jacqueline Kauff, “When Five Years Is Not Enough: Identifying and Addressing the needs of
Families Nearing the TANF Time Limit in Ramsey County, Minnesota,” Mathematica Policy Research (March 2006), pp. 124, https://www.mathematica-mpr.com/our-publications-and-findings/publications/when-five-years-is-not-enoughidentifying-and-addressing-the-needs-of-families-nearing-the-tanf-time-limit-in-ramsey-county-minnesota.
285

Dana DeMaster, “At the Limit: December 2006 Minnesota Family Investment Program (MFIP) Cases that Reached
the 60 Month Time Limit,” Minnesota Department of Human Services (January 2008), pp. 118, https://edocs.dhs.state.mn.us/lfserver/Legacy/DHS-5092B-ENG.
286

Christina McHugh and J. Taylor Danielson, “TANF Time Limit Analysis Comparing Cases Closed Due to Time
Limits with Other Case Closures,” Washington State Department of Social and Health Services (February 2019).
287

Robert Jay Dilger et al., “WV WORKS 2003: Perspectives of Former Recipients Who Have Exhausted Their 60
Months of Program Eligibility,” West Virginia University Interdisciplinary Task Force on Welfare Reform (Summer 2004), pp. 124.

125

•A

California study of the cases that reach the time limit found families often struggle with one
major barrier to work and often multiple barriers. One-third of respondents cited major
health issues as a big barrier to work. A smaller share said they were caring for a family
member with a major health issue. More than one-fifth of respondents said they suffered
from depression or anxiety or had experienced at least one stressful event in the past year.
About 11 percent experienced a domestic violence situation. More than half of all
respondents noted having at least one barrier and 28 percent cited having two or more
barriers. Forty-three percent said they had trouble paying rent; 54 percent said they had
trouble paying utility bills; 39 percent reported having trouble paying for food; and 40 percent
noted they had to use a food bank.288

G. Imposing Time Limits and Sanctions for Failure to Meet Work
Requirements Has a Disparate Impact on Communities of Color
States’ application of work requirements in the TANF cash assistance program has exacerbated
racial inequities, research shows. On the whole, research on TANF suggests that policies to take
away SNAP from individuals who are not working or participating in work activities for a specific
number of hours each month will hurt, not help, the individuals most in need of assistance. Nearly
every study comparing the race and ethnicity of sanctioned and non-sanctioned TANF recipients
finds that African Americans are significantly more likely to be sanctioned than their white
counterparts.289 For example:
•A

2011 study of Minnesota TANF recipients found that the state sanctioned a
disproportionate number of American Indian or Alaskan Native recipients and sanctioned
them more often than other racial groups during a 24-month observation period. While
American Indian or Alaskan Natives only comprised around 10.9 percent of the families
receiving TANF in Minnesota, they made up 12.2 percent of all families that the state
sanctioned. Further, the average number of sanctions was 3.54 for American Indian and
Alaskan Native families, while it was only 3.18 for White, non-Hispanic families.290

•A

2007 study of Florida’s TANF program showed black families were more likely to be
sanctioned than White families after several months of continuous TANF receipt.
Specifically, the study estimated that among families who received TANF benefits for at least
nine months continuously, black families were 22 to 35 percent more likely to be sanctioned
than White families.291

288

Rebecca A. London and Jane G. Mauldon, “Time Running Out: A Portrait of California Families Reaching the
CalWORKs 60-Month Time Limit in 2004,” Welfare Policy Research Project (November 2006), pp.
1- 6. http://escholarship.ucop.edu/uc/item/37g3510z.
289

LaDonna Pavetti, Michelle K. Derr, and Heather Hesketh, “Review of Sanction Policies and Research Studies,”
Mathematica Policy Research (March 2003).
290

Anita M. Larson, Shweta Singh, and Crystal Lewis, “Sanctions and Education Outcomes for Children in TANF
Families,” Child & Youth Services (September 2011), pp. 180-199.
291

Richard C. Fording, Joe Soss, and Sanford F. Schram, “Devolution, Discretion, and the Effect of Local Political
Values on TANF Sanctioning,” Social Service Review (June 2007), pp. 285-316.

126

•A

2006 study of women in Wisconsin receiving TANF found that black women were more
likely to be sanctioned than their white counterparts. This result was statistically significant
under both a simple analysis and an analysis that took into account the duration of each
individual’s receipt of TANF grants.292

•A

2002 study of women receiving TANF in Michigan found that black women were
disproportionately sanctioned compared to white women. The authors found similar results
under two different specifications. One analysis, looking at mean differences, found that
black women made up a disproportionate number of the total number of women that the
state sanctioned. The other model, a multivariate logistic regression, provided similar
evidence: African American women were 1.73 times more likely to be sanctioned than White
women.293

•A

2011 study of TANF recipients in Maryland found evidence that African Americans were
more likely to lose benefits due to sanctions compared to other recipients. The study found
that African Americans were disproportionately represented among families that had been
sanctioned as a result of work requirements relative to respondents of other races.294

•A

New Jersey study found that among TANF recipients entering the program between July
2000 and June 2001, 36 percent of African American recipients had their TANF grants
reduced and 16 percent had their grant eliminated due to a work-related sanction; the
comparable figures for white recipients were 27 percent and 10 percent, respectively.295

Data from the Wisconsin Department of Workforce Development shows a consistent pattern of
racial and ethnic discrepancies in TANF sanctions. Statewide, 42 percent of African American
participants and 45 percent of Hispanic participants were sanctioned, compared to just 24 percent of
white participants.296 Researchers in several states have looked at the demographics of the share of
their caseload approaching or at the time limit. In a number of examples, including a national
survey, families and recipients of color — and particularly black recipients — are more likely to lose
benefits due to the time limit. This evidence suggests that expanding time limits in other programs
will disproportionally affect black recipients. Other research cited below highlight the unique
challenges African Americans face in the labor market. Loss of assistance, paired with the difficulty
of securing or maintaining a job, could make the hardship experienced by this group even worse.
• The

Minnesota TANF agency found that black women, including African Americans, Somali
immigrants, and other African immigrants, made up about half of the adult recipients who

292

Chi-Fang Wu et al., “How Do Welfare Sanctions Work?,” Social Work Research (March 2006), pp. 33-50.

293

Ariel Kalil, Kristin S. Seefeldt, and Hui-chen Wang, “Sanctions and Material Hardship under TANF,” Social Service
Review (December 2002), pp. 642-662.
294

Sarah Williamson, “Full-Family Sanctions & Economic Recession,” University of Maryland Family Research and
Training Group (January 2011). https://familywelfare.umaryland.edu/reports1/sanctionsbrief.pdf
295

LaDonna Pavetti et al., “The Use of TANF Work-Oriented Sanctions in Illinois, New Jersey, and South
Carolina,” Mathematica Policy Research (April 2004).
296

Wisconsin Department of Workforce Development, “Wisconsin Works (W-2) Sanctions Study,” (December 2004).

127

reached 60 months, the federal time limit. African Americans in particular were the most
likely to reach the federal time limit.297
•A

study in Maryland found most of the caseload consisted of black women and they were the
most likely of any racial group to reach the time limit. 298

• In Virginia,

black families were more likely to reach the limit than white families.299

• In Washington State,

Indian.

300

families cut off by the time limit tended to be black or American

• Using the

Women’s Employment Study for one Michigan urban county, researchers analyzed
factors associated with increased time on TANF. They found that black women were far
more likely to have accumulated more months, and thus be closer to the time limit, than white
women.301

SNAP participants of color also face discrimination when looking for work. Investigations in job
discrimination uncovered strong employer preferences for white candidates over candidates of
color. One study found that resumes with white-sounding names are more likely to get call-backs
than resumes with equal qualifications but with black-sounding names. Other research shows that
generally, those with a criminal record are less likely to get call-backs or requests for interviews from
employers. Furthermore, black applicants without criminal records are less likely to receive
favorable treatment than white applicants without criminal records, but white applicants with a
criminal record are more likely to receive favorable treatment than black applicants with no criminal
history.302

H. Households That Lose TANF Benefits Because of Sanctions or Time
Limits Experience Higher Levels of Material Hardship and Increased
Hardship
People with incomes low enough to qualify for SNAP also often have few or no assets to lean on
in difficult times and a limited amount of cash to meet basic needs like rent and utilities, clothes,
Dana DeMaster, “At the Limit: December 2006 Minnesota Family Investment Program (MFIP) Cases that Reached
the 60 Month Time Limit,” Minnesota Department of Human Services (January 2008), pp. 118, https://edocs.dhs.state.mn.us/lfserver/Legacy/DHS-5092B-ENG.
297

Pamela Ovwigho, Kathryn Patterson, and Catherine E. Born, “The TANF Time Limit: Barriers & Outcomes among
Families Reaching the Limit,” Family Welfare Research and Training Group (November 2007), pp. 134, https://familywelfare.umaryland.edu/reports1/tl_barriers.pdf.
298

Anne Gordon et al., “Experiences of Virginia Time Limit Families After Case Closure: An Interim
Report,” Mathematica (2002), pp. 1 – 177, https://www.mathematica-mpr.com//media/publications/pdfs/expvafinal.pdf.
299

Christina McHugh and J. Taylor Danielson, “TANF Time Limit Analysis Comparing Cases Closed Due to Time
Limits with Other Case Closures,” Washington State Department of Social and Health Services (February 2019).
300

Kristin S. Seefeldt and Sean M. Orzol, “Watching the Clock Tick: Factors Associated with TANF
Accumulation,” National Poverty Center Working Paper Series (May 2005),
http://www.npc.umich.edu/publications/workingpaper04/paper9/04-09.pdf.
301

302

Devah Pager, “The Mark of a Criminal Record,” American Journal of Sociology (2003), pp. 937-975.

128

personal care items, and gas or bus fare, among other things. SNAP helps meet food costs, but
when that assistance is taken away, individuals struggle to make ends meet and some are unable to
avoid a downward spiral. Studies examining sanctioned TANF families show that many people
experience increased hardships after facing a sanction: a 2004 longitudinal study of TANF recipients
in Illinois found that sanctioned families who faced grant reductions had higher levels of food
hardship after being sanctioned than those who did not have their grants reduced by sanctions.
Researchers defined food hardship as sometimes or often not having enough to eat. Recipients who
had sanctions in the period January 1999 to March 2001 reported between February 2002 and
September 2002 a higher incidence of food hardship and perceived overall hardship than other
TANF recipients who did not experience grant reductions resulting from sanctions during the
period. A multivariate analysis conducted with the same data indicated respondents who had their
grants reduced due to sanctions were over three times more likely to report food hardship in the
final period of the study (after the sanction) compared to those who were never sanctioned during
the study, after controlling for demographic and other factors.303
A 2004 longitudinal study of TANF recipients in Illinois found those who had their grants
reduced due to sanctions reported higher levels of perceived overall hardship. Perceived overall
hardship was determined by the extent to which respondents agreed with statements like, “I worry
about having enough money in the future.” Respondents who saw grant reductions as a result of
sanctions between January 1999 and March 2001 showed more perceived overall hardship following
the sanctions (between February and September 2002) than those who were never sanctioned during
the former, 51-month period. 304
A 2002 study of women in Michigan receiving TANF suggests those who were sanctioned were
more likely to experience hardship and have to prioritize hardship-mediating activities than those
who were not sanctioned. Specifically, 21 percent of sanctioned families (compared to 9 percent of
non-sanctioned families) reported having their gas or electricity turned off in the previous year
because they could not afford to make their utility payments.305 Researchers found that 34 percent
of sanctioned families (compared to 14 percent of non-sanctioned families) had resorted to
hardship-mitigating activities such as pawning, stealing food, searching in trash cans, or begging. 306
The 2002 study of Michigan TANF households suggests women who were sanctioned are more
likely to expect future hardship — such as inadequate housing, food, or medical care in the next two
months — than those who were not sanctioned. Women who were sanctioned were 2.41 times
303

Bong Joo Lee, Kristen S. Slack, and Dan A. Lewis, “Are Welfare Sanctions Working as Intended? Welfare Receipt,
Work Activity, and Material Hardship among TANF-Recipient Families,” Social Service Review (September 2004), pp. 370403.
304

Bong Joo Lee, Kristen S. Slack, and Dan A. Lewis, “Are Welfare Sanctions Working as Intended? Welfare Receipt,
Work Activity, and Material Hardship among TANF-Recipient Families,” Social Service Review (September 2004), pp. 370403.
305

Ariel Kalil, Kristin S. Seefeldt, and Hui-chen Wang, “Sanctions and Material Hardship under TANF,” Social Service
Review (December 2002), pp. 642-662.
306

Ariel Kalil, Kristin S. Seefeldt, and Hui-chen Wang, “Sanctions and Material Hardship under TANF,” Social Service
Review (December 2002), pp. 642-662.

129

more likely to expect future hardship compared to women who were not sanctioned, the researchers
found.307
There is evidence from a study of people who frequented a food pantry in upstate New York that
people who lost TANF benefits due to sanctions were more likely to experience hardship than
others. Specifically, it was more common for people who had been sanctioned — relative to those
who had not been sanctioned — to report having more difficulty in the past six months paying for
food, rent, adult health care, and other bills. Similarly, the number of sanctioned respondents who
indicated they had moved within the past six months due to inability to pay rent was
disproportionately high relative to the responses from those who were not sanctioned. Access to a
telephone told a similar story of increased hardship, with a disproportionate number of sanctioned
people lacking access to a phone, relative to those who were not sanctioned. 308
A study using data from the Fragile Families and Child Wellbeing survey, which surveys mothers
from 20 cities in 15 states, evaluated the level of hardship among those who had been sanctioned in
the prior 12 months and non-sanctioned mothers who received TANF. Researchers found that
those who had been sanctioned in the prior 12 months were 85 percent more likely to report any
material hardship compared to non-sanctioned mothers receiving TANF (42 percent of those
sanctioned reported one or more material hardships, compared to 27 percent of those not
sanctioned). Researchers found that those sanctioned were 63 percent more likely than nonsanctioned mothers to report maternal or child hunger and 76 percent more likely to report having
their utilities shut off in the 12 months prior to the interview. The study also found that sanctioned
mothers were 79 percent more likely to report being unable to receive medical care, for either
themselves or a child, due to cost. The study controlled for sociodemographic factors known to be
associated with being sanctioned and controlled for hardship levels mothers faced prior to being
sanctioned.309
A Washington State study using predictive modeling to identify the factors likeliest to cause a new
spell of homelessness for TANF parents found that sanctioned recipients were about 20 percent
more likely than non-sanctioned parents to begin a new spell of homelessness in the next month. 310
Underlying the NPRM’s proposal to restrict waivers is the claim that subjecting more individuals
to the three-month time limit will increase employment. But research on public benefit programs
that have time limits demonstrates that arbitrary time limits do not lead to self-sufficiency. Instead,
some research finds families cut off TANF because of time limits have significant barriers to
307

Ariel Kalil, Kristin S. Seefeldt, and Hui-chen Wang, “Sanctions and Material Hardship under TANF,” Social Service
Review (December 2002), pp. 642-662.
308

Jean Oggins and Amy Fleming, “Welfare Reform Sanctions and Financial Strain in a Food-Pantry Sample,” The
Journal of Sociology & Social Welfare (June 2001), pp. 101-123,
https://scholarworks.wmich.edu/cgi/viewcontent.cgi?article=2725&context=jssw.
309

Nancy E. Reichman et al., “TANF Sanctioning and Hardship,” Social Service Review (June 2005) Vol. 79 No. 2, pp. 215236.
310

Melissa Ford Shah et al., “Predicting Homelessness among Low-Income Parents on TANF,” Washington State
Department of Social & Health Services (August 2015).

130

employment and experience hardship. Without cash, the challenge for parents to support their
children becomes even harder and a downward spiral emerges. Finding employment becomes even
harder when parents need to scramble to make ends meet. The studies below offer examples of
time-limited families unable to maintain stable housing and/or pay their bills and, in some instances,
afford enough food.
•A

Washington State study comparing families who left TANF due to time limits and those
who left for other reasons found that time-limited families were more likely to be unstably
housed one month after losing assistance.311

• According to a

survey of 276 former West Virginia TANF recipients cut off because of time
limits, 59 percent reported that they were either worse off or much worse off financially since
leaving WV WORKS; 65 percent did not have enough money to pay the electric, gas, or water
bill; and 51 percent did not have enough money to pay for heat. These percentages were
much lower when the recipients were on WV WORKS. After losing assistance because of the
time limit, 61 percent of respondents reported that the amount of stress in their lives was
either worse or much worse since being removed from WV WORKS. They were also more
pessimistic about their own personal and financial futures. These financial burdens stem from
their very low level of employment: only 26.9 percent of recipients were employed and more
than half of the employed were working part-time.312

•A

survey of dozens of Maine TANF recipients cut off by the time limit found that families
experienced increased reliance on food banks, inability to pay utilities and other bills, and
overcrowded housing conditions or reliance on homeless shelters.313

I. Work Requirements in TANF Do Not Work
The rationale for reducing or eliminating benefits for not meeting a work requirement is that this
will compel unemployed adults to find work. Evidence suggests that work requirements (along with
other policy changes that accompanied TANF’s implementation) contributed to a modest increase
in employment during the late 1990s, but that work often was not steady, a pattern reflected in
recent studies as well.
Research on adults who lost TANF due to sanctions for failure to meet a work requirement found
that these individuals have trouble finding employment after their exit. The personal, family, or
community barriers that kept them from finding a job while on TANF also prevent these parents
from finding work after TANF. Findings from TANF suggest that even if the NPRM intends to
impose work requirements only on “work able” individuals, substantial numbers of SNAP recipients
who face personal or family challenges would likely fall through the cracks and have their benefits
311

Melissa Ford Shah et al., “Predicting Homelessness among Low-Income Parents on TANF,” Washington State
Department of Social & Health Services (August 2015).
312

Robert Jay Dilger et al., “WV WORKS 2003: Perspectives of Former Recipients Who Have Exhausted Their 60
Months of Program Eligibility,” West Virginia University Interdisciplinary Task Force on Welfare Reform (Summer 2004), pp. 124.
313

Sandra Butler, “TANF Time Limits and Maine Families: Consequences of Withdrawing the Safety Net,” Maine Equal
Justice Partners (April 2013), pp. 1-16, http://www.mejp.org/sites/default/files/TANF-Study-SButler-Feb2013.pdf.

131

reduced or taken away. Evidence from studies that show that employment rates tend to be lower
for these populations include the following:
•A

2004 longitudinal study of TANF recipients in Illinois found that those respondents who
were sanctioned during the study period were more likely to be unemployed after the study
period than those who were not sanctioned and had, on average, significantly lower levels of
earnings post-sanction than those who were not sanctioned. More specifically, respondents
who received sanctions that reduced cash grants between January 1999 and March 2001 were
44 percent less likely to be engaged in formal work during the period April 2001 through
September 2001 than the respondents who did not receive sanctions during the preceding
period, even after controlling for previous work experience and other characteristics
associated with employment. And, TANF families receiving sanctions that were carried out
through grant reductions between January 1999 and March 2001 had lower average earnings
during the next six months than those respondents who were not sanctioned.314
The study also helps explain why those sanctioned had worse outcomes and were less likely to
be working. Working-age adults who have their grants reduced due to sanctions had higher
barriers to employment than those who were not sanctioned. The group had higher levels of
engagement in job training and other work activities and had a higher incidence of
participation in informal work such as babysitting and odd jobs in the final period of the study
than those with no grant reductions due to sanctions. This indicates the lower levels of formal
employment among sanctioned respondents is not easily attributed to an unwillingness to
work, since this population engages more heavily than those who did not experience grant
reductions from sanctions in job training and informal work. A stronger explanation is that
those who are sanctioned have more significant barriers to formal employment than those
who are not. 315

•A

2011 study of Maryland TANF recipients who were sanctioned found that these recipients
had consistently lower post-exit employment rates relative to those who left TANF for
reasons other than work sanctions, throughout the nine-year post-exit period that the study
covered. Similarly, the average earnings for the group that left due to work sanctions was
lower than the average earnings of those who left for other reasons.316

•A

2018 study of state-collected data on the employment of Kansas parents leaving TANF
cash assistance due to work-related sanctions and time limits between October 2011 and
March 2015 indicates that a lower share of these parents worked in the year after their exit
compared to families exiting TANF for other reasons. They also found it more difficult to
find steady work compared to the other families exiting the program. In the average quarter

314

Bong Joo Lee, Kristen S. Slack, and Dan A. Lewis, “Are Welfare Sanctions Working as Intended? Welfare Receipt,
Work Activity, and Material Hardship among TANF-Recipient Families,” Social Service Review (September 2004), pp. 370403.
315

Bong Joo Lee, Kristen S. Slack, and Dan A. Lewis, “Are Welfare Sanctions Working as Intended? Welfare Receipt,
Work Activity, and Material Hardship among TANF-Recipient Families,” Social Service Review (September 2004), pp. 370403.
316

Sarah Williamson, “Full-Family Sanctions & Economic Recession,” University of Maryland Family Research and
Training Group (January 2011), https://familywelfare.umaryland.edu/reports1/sanctionsbrief.pdf.

132

of the year after exiting, only 49 percent and 47 percent of the sanctioned and time-limited
families, respectively, were working, compared to 72 percent of families exiting due to the
income limit and 53 percent for all other reasons. Moreover, only about a quarter of the
sanctioned and time-limited families worked between seven and nine quarters in the year
before and after their exit, compared to a third of families overall.317
•A

2001 study of people who frequented a food pantry in upstate New York found that
employment rates were lower for people who had lost TANF benefits due to sanctions.
Among those surveyed in 1997, 13 percent of those sanctioned reported having had wage
earnings in the previous six months, while 22 percent of unsanctioned respondents reported
earnings over the same period. Moreover, because of the effects of sanctions, sanctioned
respondents were far more likely to report being disconnected both from work and from
TANF benefits. A full quarter of the sanctioned sample in 1997 reported no work and no
TANF benefits, compared with 3 percent reporting the same in the unsanctioned sample.318

•A

three-year study of TANF recipients in two California counties provides evidence that
employment rates are lower for those with significant barriers to employment. The study
excluded individuals who received disability benefits and focused on mental health issues such
as major depression, generalized anxiety disorder, panic attacks, social phobia, or
posttraumatic stress disorder. Those who reported functional impairment over the previous
month were far more likely not to have worked over the previous year. In the first follow-up
year, 54.2 percent of functionally impaired respondents had worked in the previous year,
compared to 75.2 percent of those without such difficulties. In the second follow-up year,
58.5 percent of those with functional impairments worked in the previous year, compared to
79.2 percent of those without such difficulties. The researchers found a statistically significant
association between having a mental health issue and having no earned income over the
previous year. The most reasonable interpretation of this result is not that the respondents
who were not working had some other source of support, but that they were unable to secure
work due to their significant barrier to employment. 319

•A

California study of CALWORKS recipients found that recipients who had been sanctioned
were much less likely to report obtaining full-time employment over three years (38 percent of
respondents) compared to non-sanctioned clients (60 percent of respondents). Researchers
measured employment history by asking respondents about previous full- or part-time

317

Tazra Mitchell, LaDonna Pavetti, and Yixuan Huang, “Life After TANF in Kansas: For Most, Unsteady Work and
Earnings Below Half the Poverty Line,” Center on Budget and Policy Priorities (February 2018),
https://www.cbpp.org/research/family-income-support/life-after-tanf-in-kansas-for-most-unsteady-work-and-earningsbelow.
318

Jean Oggins and Amy Fleming, “Welfare Reform Sanctions and Financial Strain in a Food-Pantry Sample,” The
Journal of Sociology & Social Welfare (June 2001), pp. 101-123,
https://scholarworks.wmich.edu/cgi/viewcontent.cgi?article=2725&context=jssw.
319

Daniel Chandler et al., “Mental Health, Employment, and Welfare Tenure,” Journal of Community Psychology (September
2005), pp. 578-609.

133

employment, and whether they were without work throughout the three years prior to the
survey.320
• Studies

consistently find lower employment rates among TANF leavers whose cases were
closed due to a work-oriented sanction than among families that left TANF for other reasons.
For example, in Arizona, 40 percent of sanctioned leavers were working in the first quarter
after exit, compared to 55 percent of non-sanctioned leavers.321

• In Maryland,

6 months after exit, 38 percent of sanctioned leavers were employed, compared
to 58 percent of non-sanctioned leavers.322

•A

study of TANF recipients nationwide using the Census Bureau’s Survey of Income and
Program Participation indicates those who are disconnected from both work and cash
assistance are more likely to have a significant barrier to employment than those who either
work or receive cash assistance. Those who had both no household earned income or any
cash assistance during the survey month were about twice as likely to report having a physical
or mental health condition that limits one’s ability to work compared to those who either
worked or received cash assistance. Results were consistent across geographic regions. In
southern states, 15.8 percent of disconnected respondents reported a physical or mental health
work-limiting condition, while only 7.4 percent of non-disconnected respondents had such a
condition; in non-southern states, 24.3 percent of disconnected respondents had a worklimiting condition, while among non-disconnected respondents, 10.4 percent reported having
such a condition. It should be noted that these respondents with physical and mental health
conditions were not receiving support from SSI; SSI recipients (and those who reported
school as their major activity) were excluded from the sample.323
Most TANF Recipients Who Lose Benefits Due to Time Limits
Do Not Find Steady Employment

Cutting off families because they have reached some arbitrary time limit ignores whether they can
actually support themselves or if the job market is welcoming to them. Several studies have found
that parents cut off of TANF due to the time limits have trouble finding employment. The health,
familial, and behavioral circumstances that kept them from finding a job while on TANF also
prevent these parents from finding work after TANF. Black families are not only the most likely to
be cut off by time limits, but also very likely to be discriminated against in the job market, the
evidence shows. In some instances, parents who can find work may be working inconsistently and
thus still fall short of a stable income.
320

Yeheskel Hasenfeld, Toorjo Ghose, and Kandyce Larson, “The Logic of Sanctioning Welfare Recipients: An
Empirical Assessment,” Social Science Review (June 2004), Vol. 78, No. 2, pp. 304-319,
http://repository.upenn.edu/spp_papers/88/.
321

Karen L. Westra and John Routley, “Arizona Cash Assistance Exit Study,” Arizona Department of Economic
Security Office of Evaluation (January 2000).
322

Catherine Born, Pamela Caudill, and Melinda Cordero, “Life After Welfare: A Look at Sanctioned Families,”
University of Maryland School or Social Work (November 1999).
323

Andrea Hetling, “The Importance of Region and State Welfare Rules for Disconnected Single Mothers,” University of
Kentucky Center for Poverty Research Discussion Paper Series (September 2011).

134

• In Washington State,
324

leaving TANF.

time-limited parents were less likely to be employed in the year before

• Researchers

from the University of Maryland’s School of Social Work found that compared to
other people leaving TANF, those leaving because they reached the time limit had less
employment history while on TANF and worked fewer quarters in the year after leaving
assistance.325

• Other

researchers of Maryland’s TANF program found that recipients who reported having a
criminal record were more likely to reach the time limit than those who did not report having
a criminal background. While recipients with a criminal conviction are as likely to be
employed as other recipients, their employment is more unstable. These women are often
more likely to have other barriers as well, such as human capital deficits and situational
barriers.326

• An analysis

of the Building Wealth and Health Network pilot program found depression is
often a barrier to employment among TANF recipients, and that adverse childhood
experiences (ACEs) and exposure to community violence are often associated with
depression. The study investigated how resilience affects the relationship between ACEs,
community violence, and depression. TANF families have a high prevalence of health
impediments and significant barriers to employment, such as domestic violence, food
insecurity, utility shut offs, homelessness, child hospitalizations, and child developmental
risks.327
Most TANF Recipients Who Lose Benefits Due to Sanctions
Do Not Find Steady Employment

The rationale for reducing or eliminating benefits for not meeting a work requirement is that this
will compel parents to find work. Evidence suggests that work requirements (along with other
policy changes that accompanied TANF’s implementation) contributed to a modest increase in
employment during the late 1990s, but that work often was not steady,28 a pattern reflected in
recent studies as well.
Another consequence of work requirements in TANF raises concerns about the NPRM’s goal of
subjecting more unemployed adults to SNAP’s time limit. As it has become harder for single
324

Christina McHugh and J. Taylor Danielson, “TANF Time Limit Analysis Comparing Cases Closed Due to Time
Limits with Other Case Closures,” Washington State Department of Social and Health Services (February 2019).
325

Andrea Hetling, Kathryn Patterson, and Catherine Born, “The TANF Time Limit: Comparing Long-Term and Other
Welfare Leavers,” Family Welfare Research and Training Group (February 2006) pp. 1-19,
https://familywelfare.umaryland.edu/reports1/timelimitleavers.pdf.
326

Valerie Head, Catherine Born, and Pamela Ovwigho, “Criminal History as an Employment Barrier for TANF
Recipients,” Family Welfare Research Training Group (March 2009), pp. 139. https://pdfs.semanticscholar.org/2c12/a0663bd8249586c000a82122cc2e6c796e23.pdf?_ga=2.3188080.1987654092.
1552314136-1225174621.1551994502.
327

Seth L. Welles, Falguni Patel, and Mariana Chilton, “Does Employment-Related Resilience Affect the Relationship
between Childhood Adversity, Community Violence, and Depression?” Journal of Urban Health (2017), Vol. 94, pg. 233243, https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5391326/.

135

mothers to get direct financial assistance when they cannot find work, the number with neither jobs
nor TANF has grown substantially over time. In 1995, the number of families receiving cash
assistance in an average month exceeded the number of jobless single mothers by about a million.
By 2016, the number of families receiving cash assistance in an average month was roughly 2
million below the number of jobless single mothers. As a result, work requirements in TANF fueled
an increase in deep poverty (measured as income at or below half of the poverty line).

J. Evidence from Medicaid Work Requirements Shows Beneficiaries Lose
Benefits But Don’t Gain Employment
Additional evidence that taking benefits away from individuals who are unable to meet strict work
requirements does not lead to increased work rates comes from the recent state waivers to apply
such policies in Medicaid. In June 2018, Arkansas became the first state to condition receipt of
Medicaid benefits on meeting a work requirement. Certain beneficiaries must participate in and
report 80 hours of work or work-like activity each month.328 Those that fail to meet the requirement
for three months in a calendar year lose coverage.329
In 2018, over 18,000 Arkansas Medicaid beneficiaries ― nearly 25 percent of the total population
the state identified as potentially subject to the work requirement ― lost coverage for failing to meet
the requirement.330 This far exceeds the population that the state’s articulated policy intended to
target with the requirement: beneficiaries who were able to work but were not working. Many
individuals who qualified for an exemption for being unable to work or who were working are
among those who lost coverage. Many beneficiaries were unaware of the requirement, did not
understand what they had to do to meet the requirement, or were unable to navigate the reporting
requirement.331
Moreover, there is no evidence that the work requirement has led to increased employment.
Although the state has cited data from the New Hire Database as evidence that beneficiaries are
starting new jobs, it has not provided any evidence that the work requirement caused these new
hires; low-income workers frequently begin new jobs or change jobs. Further, the data source
includes individuals who worked for only a few hours or one day, doesn’t show if the job is

328

In 2018, non-disabled adult beneficiaries ages 30 through 49 who do not have minor children in their homes were
subject to the work requirement in Arkansas. The state has begun phasing in the requirement for beneficiaries 19
through 29 in 2019.
329

Jennifer Wagner, “Commentary: As Predicted, Eligible Arkansas Medicaid Beneficiaries Struggling to Meet Rigid
Work Requirements,” Center on Budget and Policy Priorities, July 30, 2018,
https://www.cbpp.org/health/commentary-as-predicted-eligible-arkansas-medicaid-beneficiaries-struggling-to-meetrigid.
330

Jennifer Wagner, “Medicaid Coverage Losses Mounting in Arkansas From Work Requirement,” Center on Budget
and Policy Priorities, January 17, 2019, https://www.cbpp.org/blog/medicaid-coverage-losses-mounting-in-arkansasfrom-work-requirement.
331

Jennifer Wagner, “4,109 More Arkansans Lost Medicaid in October for not Meeting Rigid Work Requirement,”
Center on Budget and Policy Priorities, October 16, 2018, https://www.cbpp.org/blog/4109-more-arkansans-lostmedicaid-in-october-for-not-meeting-rigid-work-requirements.

136

temporary (such as seasonal work around the holidays), and doesn’t indicate if the employee had
been previously unemployed as opposed to just recently changed jobs. 332
In fact, other evidence from state Medicaid administrative data indicates that at most a few
hundred people may have found jobs due to the federal waiver. Most Medicaid beneficiaries don’t
face monthly reporting requirements, mainly because they’re already working or qualified for
exemptions. Only the remaining group, which has to report hours of work each month, faces any
new work incentive due to the new policy. And of that group, only a few hundred each month have
met the requirement by reporting some work hours, the state reports. What’s more, many of them
likely would have found jobs anyway.333
These data are consistent with focus group interviews showing that the work requirement isn’t
changing Medicaid beneficiaries’ behavior. Beneficiaries already had enough reasons to work: they
need to pay their bills. But they often struggle with unstable work hours, live in rural areas with few
jobs, or face other barriers to employment — and the state hasn’t invested any new money in job
training programs, services to address barriers, or supports like transportation to help beneficiaries
connect to jobs.334
Meanwhile, the work requirement has even proved counterproductive for some. News reports
describe working beneficiaries who struggled with the reporting requirement and lost Medicaid
coverage. Consequently, some have gone without needed medication, worsening their health and in
some cases costing them their jobs. Moreover, any small increase in employment must be viewed in
light of the 18,000 beneficiaries who lost coverage. 335
Numerous other states are in the process of implementing Medicaid work requirements, and
estimates show similar coverage losses are likely. For example, the waiver in Michigan may lead up
to 27 percent of the state’s Medicaid expansion population to lose coverage the first year. 336
Kentucky’s own projections say its work requirement, which has been challenged in court, would
cause 95,000 enrollees to lose coverage in five years.337 A group of health care providers and
advocates filed an amicus brief in the Kentucky litigation pointing out that its work requirement will
worsen health and won’t promote work.338
332

Jennifer Wagner, “Fact Checking Arkansas Governor’s Claims About Jobs and Medicaid Waiver,” Center on Budget
and Policy Priorities, January 28, 2019, https://www.cbpp.org/blog/fact-checking-arkansas-governors-claims-aboutjobs-and-medicaid-waiver.
333

Ibid.

334

Ibid.

335

Ibid.

336

Cindy Mann and April Grady, “Potential Enrollment Impacts of Michigan’s Medicaid Work Requirement,” Manatt
Health, February 6, 2019, https://www.manatt.com/Manatt/media/Media/Images/White%20Papers/Manatt_MIWork-Req-Estimates_20190206-Final.pdf.
337

Phil Galewitz, “Judge Blocks Kentucky Medicaid Work Requirement,” Kaiser Health News, June 29, 2018,
https://khn.org/news/judge-blocks-kentucky-medicaid-work-requirement/.
338

Steward v. Azar amicus brief, January 24, 2019,
https://www.psychiatry.org/File%20Library/Psychiatrists/Directories/Library-and-Archive/amicus-briefs/amicus2019-Stewart-v-Azar-DC-No18-152.pdf.

137

The experience thus far with Medicaid work requirements demonstrates that such policies take
coverage away from large portions of beneficiaries, including those who are working or qualify for
an exemption but cannot navigate the red tape of the requirement. At the same time, they fail to
lead to increased work activity or employment.

138

Chapter 7: Proposed Rule’s Requirement That State
Waiver Requests Have the Governor’s “Endorsement”
Violates Congressional Intent
The proposed rule would require that state requests to waive the time limit in areas with
insufficient jobs “be endorsed by the State’s Governor”339 (emphasis added). This change is in direct
violation of Congressional intent, as clearly expressed less than two months before publication of
the proposed rule. If FNS proceeds to publish a final rule it must reject this change and conform to
the intent of Congress.
Current regulations regarding state requests to waive the three-month time limit say simply that
such requests are made to FNS “on the request of the State agency.”340 The presumption is that
state agencies will be acting under the direction of their political leadership, including the Governor.
CBPP has worked with states on their waiver requests for more than 20 years. We cannot
remember ever working with a state agency that knowingly sought a waiver against the wishes of the
Governor. It is true, however, that Governors are not typically aware of every detailed policy option
and choice that their cabinet Secretaries adopt for SNAP. Similarly, Governors do not typically sign
waivers, review nutrition education plans, or even personally review large scale procurements.
Governors serve as chief executives rather than detailed policy implementers.
The House-passed 2018 farm bill sought to require the “approval of the chief executive officer of
the State”341 for waiver requests (emphasis added). The final conference agreement on the 2018
farm bill rejected the House approach, and instead requires “the support of the chief executive officer
of the state” (emphasis added).342 The 2018 farm bill, The Agriculture Improvement Act of 2018,
passed the Congress in mid-December and was signed by the President on December 20, 2018.
The conferees in the conference report that accompanied the bill were very clear about their
intent in making this change:
The Managers intend to maintain the practice that bestows authority on the State agency
responsible for administering SNAP to determine when and how waiver requests for
ABAWDs are submitted. In response to concerns that have been raised by some Members
that State agencies have not fully communicated to the chief executive their intent to request a
waiver under section 6(o), the Managers have included a provision to encourage
communication between the State agency and the chief executive officer of the State. The
Managers agree that State agencies should have the support of these officials in their application
for waiver, ensuring maximum State coordination. It is not the Managers’ intent that USDA
339

NPRM, p. 992.

340

7 C.F.R. § 273.24(f).

341

Section 4015 of House-passed H.R. 2, https://www.congress.gov/115/bills/hr2/BILLS-115hr2eh.pdf.

342

Section 4005 of H.R. 2 as enacted, https://www.congress.gov/115/bills/hr2/BILLS-115hr2enr.pdf.

139

undertake any new rulemaking in order to facilitate support for requests from State agencies, nor
should the language result in any additional paperwork or administrative steps under the waiver
process.343 (Emphasis added.)
Thus, the conferees were clear that they did not intend for FNS to engage in new rule-making
based on the change and did not want to introduce any new “paperwork or administrative steps.”
State Administrators are left on their own to ensure that they have the support of their Governor.
The change in statute simply clarifies this practice for those who were unduly concerned that state
agencies were acting against the wishes of the Governor.
By requiring the “endorsement” of the state’s governor in the proposed rule, FNS ignored this
expressed intent of Congress and went too far. The only explanation FNS gives in the NPRM is a
short sentence in the preamble:
The Department proposes clarifying that any State agency’s waiver request must have the
Governor’s endorsement to ensure that such a critical request is supported at the highest
levels of State government.344
The Merriam Webster’s Collegiate Dictionary definition of “endorsement” suggests that the term
implies a signature, which would necessarily require additional paperwork. Such a step would
directly contradict Congressional intent. From other aspects of the NPRM it is clear that FNS was
aware of the passage of the 2018 farm bill.345 So the only reasonable conclusion is that FNS chose
to ignore Congressional intent and intends to add paperwork burden and steps to the process.

Conference Report to accompany H.R. 2, pp. 616-617, https://www.congress.gov/115/crpt/hrpt1072/CRPT115hrpt1072.pdf.
343

344

NPRM, p. 983.

See, for example, p. 987 of the NPRM: “The proposed rule would end the unlimited carryover and accumulation of
ABAWD percentage exemptions, previously referred to as 15 percent exemptions before the enactment of the
Agriculture Improvement Act of 2018. Upon enactment, Section 6(o)(6) of the Act provides that each State agency be
allotted exemptions equal to an estimated 12 percent of ‘‘covered individuals…’’
345

140

Chapter 8: Proposed Rule Would Make Implementing
The Time Limit Harder by Removing Provisions That
Give States Certainty Around Approval
The proposed rule would eliminate the ability of states to implement a waiver at the time a request
is submitted, requiring FNS approval prior to any waiver implementation. The proposed rule would
also remove language that identifies waivers that meet certain standards as “readily approvable.”
Currently, these two provisions give states certainty of approval that enables them to better plan for
waiver implementation while waiting for approval. Given that FNS can substantially delay approval
(and recently has done so), this proposal would put an undue burden on states preparing for the
complex and error-prone process of implementing the time limit. FNS also failed to articulate a need
for these changes, making it difficult for commenters to weigh in on any potential benefit. We
therefore urge the Department to keep current regulations at 7 C.F.R. § 273.24(f)(3) and 7 C.F.R. §
273.24(f)(4), which establish the “readily approvable” standard and allow states to implement
waivers upon submission of the waiver request in some instances.
Current regulations have two provisions that give states more certainty in the waiver approval
process. These provisions allow them time to prepare for implementation of the time limit while
waiting for FNS approval. The first provision, at 7 C.F.R. § 273.24(f)(3), establishes that waivers that
meet certain standards are “readily approvable.” A readily approvable waiver includes data from the
Bureau of Labor Statistics showing a 12-month unemployment rate of 10 percent, a 24-month
unemployment rate 20 percent above the national average, or designation as a Labor Surplus Area
(LSA) by the Department of Labor’s Employment and Training Agency. The final rule, published in
2001, stated that the Department decided to designate that it would approve those waivers “to
facilitate the waiver process.”346 The second provision, at 7 C.F.R. § 273.24(f)(4), allows states to
implement waivers based on having either a 12-month unemployment rate of 10 percent or LSA
designation for the current fiscal year upon waiver submission, rather than waiting for FNS
approval. With those two provisions, states can plan on implementing the waiver submitted under
the first provision while awaiting FNS approval, and can actually implement prior to approval if it is
one of the waivers specified in 7 C.F.R. § 273.24(f)(4). (FNS can contact the state to modify the
waiver if needed.)

A. Certainty About Waiver Approval Process Is Crucial Due to Lengthy
State Implementation Process
With a reasonable amount of certainty about FNS waiver approval, states can begin to plan earlier
than if they had to wait for FNS to process waiver approval, which can take months and
substantially delay planning. Having time to plan for implementation is crucial for states because of
the demands of thoroughly implementing the time limit. As several documents from USDA —
including memos, guidance, and a report from USDA’s Office of the Inspector General (OIG) —
make clear, before a state can implement the time limit in a new area, states must:
346

66 Fed. Reg., No. 11, 4438 (January 17, 2001).

141

• Identify

individuals subject to the time limit: as one FNS memo explains, “Prior to waiver
expiration, states must review case file information to identify individual ABAWDs and
determine whether or not the ABAWD is subject to the time limit.”347

• Inform

individuals subject to the time limit: state agencies have minimum requirements for
notifying people who are subject or potentially subject to the time limit (such as an individual
a state has identified as likely subject to the time limit based on age and other characteristics,
but who may be eligible for an exemption). As one memo explains, states must “inform
ABAWD and potential ABAWD households of the time limit, exemption criteria (including
exemptions from the general work requirements), and how to fulfill the ABAWD work
requirement,” as well as the requirement to report when work hours fall below 20 hours per
week.348 The law requires caseworkers to explain these rules during the individual’s eligibility
interview, but given the complexity of the policy, FNS recommends providing written notice
to clients at least 30 days before the waiver ends. FNS encourages states to write notices in
clear, understandable language, develop public information materials for websites and waiting
rooms, and leverage partnerships in the community such as service providers.349 To properly
implement the time limit, states must therefore train staff to ensure they can effectively
explain the requirements to individuals subject to the time limit, develop written notifications,
and use other resources such as community partnerships — all well before a waiver ends.

• Develop policies:

States must develop policies for many aspects of the time limit, such as
whether they will use a fixed or rolling clock, what procedures they will use to screen
individuals for exemptions and what verifications are required, whether they will count unpaid
or volunteer work towards the requirement, and how they will use 15 percent exemptions,
among many others.350 States must also communicate these policies to caseworkers and other
relevant staff, and ensure that computer systems reflect their policy choices. While some of
these policy decisions may not change depending on the waiver outcome if the state has
developed these policies for areas that already have the time limit, states or counties preparing

347

U.S. Department of Agriculture, Food and Nutrition Service, “Supplemental Nutrition Assistance Program –
Expiration of Statewide ABAWD Time Limit Waivers,” March 4, 2015, https://fnsprod.azureedge.net/sites/default/files/snap/SNAP-Expiration-of-Statewide-ABAWD-Time-Limit-Waivers.pdf
348

U.S. Department of Agriculture, Food and Nutrition Service, “SNAP – Requirements for Informing Households of
ABAWD Rules,” April 17, 2017, https://fnsprod.azureedge.net/sites/default/files/snap/Requirements_for_Informing_Households_of_ABAWD_Rules.pdf
349

U.S. Department of Agriculture, Food and Nutrition Service, “SNAP - Best Practices and Resources for Informing
Households of ABAWD Rules, May 25, 2018, https://fnsprod.azureedge.net/sites/default/files/snap/BestPracticesforInformingABAWDS.pdf.
350

USDA Office of Inspector General, FNS Controls Over SNAP Benefits For Able-Bodied Adults Without Dependents,
September 2016, https://www.usda.gov/oig/webdocs/27601-0002-31.pdf; U.S. Department of Agriculture, Food and
Nutrition Service, “Supplemental Nutrition Assistance Program – Expiration of Statewide ABAWD Time Limit
Waivers,” March 4, 2015, https://fns-prod.azureedge.net/sites/default/files/snap/SNAP-Expiration-of-StatewideABAWD-Time-Limit-Waivers.pdf; https://fns-prod.azureedge.net/sites/default/files/snap/SNAP-Expiration-ofStatewide-ABAWD-Time-Limit-Waivers.pdf; U.S. Department of Agriculture, Food and Nutrition Service, “Guide to
Serving ABAWDs Subject to Time-limited Participation, 2015 https://fnsprod.azureedge.net/sites/default/files/Guide_to_Serving_ABAWDs_Subject_to_Time_Limit.pdf; U.S. Department of
Agriculture, Food and Nutrition Service, “Supplemental Nutrition Assistance Program - ABAWD Time Limit Policy
and Program Access,” November 19, 2015, https://fns-prod.azureedge.net/sites/default/files/snap/ABAWD-TimeLimit-Policy-and-Program-Access-Memo-Nov2015.pdf.

142

to implement the time limit for the first time will need time to ensure that policies are ready
for implementation prior to the expiration of a waiver.
• Ready

computer systems for tracking: as the OIG report explains, “Each month, the States are
responsible for tracking an ABAWD’s status; countable months; fulfillment of the work
requirement; exemption status with respect to age, pregnancy, and mental or physical capacity
to perform work; 15 percent exemption status; and good cause for not meeting the work
requirement.”351 Setting up computer systems to accurately perform this complex monthly
tracking, which may require states to work with contractors to re-program systems and test for
errors, can be a time-consuming process.

• Train caseworkers:

states must build in adequate time to ensure that eligibility workers
thoroughly understand and can implement related policies, which may take months. For
example, workers must be prepared to follow procedures to assess individuals’ fitness for
work in order to screen for exemptions from the time limit352 and must be prepared to explain
the requirements to those individuals during the eligibility interview,353 among other tasks
instrumental to implementing the time limit. The OIG report states, “FNS national officials
informed us that the ABAWD provisions were very complex and that it takes months of
extensive training for new staff to fully understand the ABAWD requirements. A State official
said the ABAWD laws and regulations are the ‘most complicated SNAP policy in existence’
and are ‘fraught with the potential for case errors.’354

• Identify

providers for qualifying work activities: Most states are not required to provide
individuals subject to the time limit with spots in work programs that can fulfill the 20 hour a
week requirement (called “qualifying activities”). The exceptions are “pledge states,” which
receive additional funding for employment and training (E & T) programs if they commit to
providing a work training spot to individuals subject to the time limit in their last month of
SNAP benefits. FNS has in the past encouraged states to provide qualifying activities to
individuals subject to the time limit.355 States that do wish to provide these services must

351

USDA Office of Inspector General, FNS Controls Over SNAP Benefits For Able-Bodied Adults Without Dependents,
September 2016, https://www.usda.gov/oig/webdocs/27601-0002-31.pdf.
352

USDA Office of Inspector General, Supplemental Nutrition Assistance Program – Able-Bodied Adults without
Dependents (ABAWD) Questions and Answers – June, 2015, https://fnsprod.azureedge.net/sites/default/files/snap/ABAWD-Questions-and-Answers-June%202015.pdf.
353

U.S. Department of Agriculture, Food and Nutrition Service, “SNAP – Requirements for Informing Households of
ABAWD Rules,” April 17, 2017, https://fnsprod.azureedge.net/sites/default/files/snap/Requirements_for_Informing_Households_of_ABAWD_Rules.pdf.
354

USDA Office of Inspector General, FNS Controls Over SNAP Benefits For Able-Bodied Adults Without Dependents,
September 2016, https://www.usda.gov/oig/webdocs/27601-0002-31.pdf.
355

U.S. Department of Agriculture, Food and Nutrition Service, “Supplemental Nutrition Assistance Program ABAWD Time Limit Policy and Program Access,” November 19, 2015, https://fnsprod.azureedge.net/sites/default/files/snap/ABAWD-Time-Limit-Policy-and-Program-Access-Memo-Nov2015.pdf.

143

identify current E & T providers that can offer work placements for participants, and/or
develop new relationships with providers to offer placements.356
Preparing for implementation is therefore a lengthy and difficulty process, given that states must
identify and notify individuals subject to the time limit, develop policies and guidance to support
implementation, train workers, ready computer systems, and (if they choose) develop slots in work
programs. Ensuring that local offices are ready to implement when a waiver changes or when a state
or county implements the time limit for the first time is important not only to ensure that needy
individuals don’t mistakenly lose access to food assistance, but also to prevent incorrect
implementation of the time limit from causing case errors.
Unclear if Proposed Rule’s Core Standards
Are Different From “Readily Approvable” Standard
The proposed rule would weaken both provisions that currently provide states with more
certainty of FNS approval. First, the NPRM would remove the language establishing that waivers
with certain criteria are “readily approvable.” The preamble to the NPRM explains that the waivers
requested under the “core standards” are likely to be approved, stating: “These revisions would
include the establishment of core standards that would allow a State to reasonably anticipate whether
it would receive approval from the Department.”357 While the preamble therefore suggests that
states may understand that waivers requested under the “core standards” can be reasonably be
expected to be approved (provided they include the correct data and are calculated accurately), the
actual rule lacks the specificity of the “readily approvable” language in current regulations at 7 C.F.R.
§ 273.24(f)(3). The proposed rule states: “(2) Core standards. FNS will approve waiver requests
under (1)(i) and (ii) that are supported by any one of the following.” If these core standards are
indeed “readily approvable,” then clarifying that USDA will approve waiver based on those
standards would enable states to continue to plan for implementation.
The proposed rule would also eliminate the current provision at 7 C.F.R. § 273.24(f)(4) that allows
states to implement the waiver upon submission. The preamble states:
The proposed rule would bar States from implementing a waiver prior to its approval. Though
rarely used, current regulations allow a State to implement an ABAWD waiver as soon as the
State submits the waiver request based on certain criteria. By removing the current pertinent
text in 273.24(f)(4), the proposed rule would require States to request and receive approval
before implementing a waiver. This would allow the Department to have a more accurate
understanding of the status of existing waivers and would provide better oversight in the
waiver process. It would also prevent waivers from being implemented until the Department
explicitly reviewed and approved the waiver.358

356

U.S. Department of Agriculture, Food and Nutrition Service, “Guide to Serving ABAWDs Subject to Time-limited
Participation, 2015, https://fnsprod.azureedge.net/sites/default/files/Guide_to_Serving_ABAWDs_Subject_to_Time_Limit.pdf.
357

NPRM, p. 983.

358

NPRM, p. 987.

144

The Department’s rationale for eliminating this provision is unclear given that the proposed rule
also establishes “core standards” and current regulations require states to submit a detailed waiver
request before implementing. The Department claims that eliminating the provision would allow the
Department to “have a more accurate understanding of the status of existing waivers,” but the
Department does not explain why it lacks this clarity under current rules (given that states must
submit waiver requests with the proposed waiver date of implementation) and cannot instead clarify
requirements around informing FNS about implementation, rather than limit states’ ability to
implement a waiver while waiting for FNS approval. The Department also states that removing this
provision would allow the Department to “provide better oversight in the waiver process,” but again
does not explain what current issue this proposal would address. If states can only submit waivers
based on very clear criteria with clear methods, and FNS has the ability to modify the waiver, why
does the Department suggest it currently lacks oversight in this process? The provision does not
remove the ability of FNS to review and approve waivers, but instead moves up the timeline to give
states the ability to more effectively implement waivers they know will be approved. FNS does not
explain what need or deficit this proposal seeks to remedy, which makes it very difficult for
comments to respond.
Department Does Not Address Impact
of Its New, Lengthy Approval Process on State Implementation
The most problematic aspect of the Department’s proposal to remove the ability of states to
implement waivers prior to approval is that the Department does not make any proposal that will
ensure that the Department approves waivers in a timely enough fashion to give states the certainty
they need to properly implement the time limit. Current regulations and guidelines require waivers to
be based on recent economic data, which by definition narrows the window of time between a
state’s waiver submission and the implementation date.
For example, as recent guidance explains, for waivers based on a 12-month unemployment rate of
10 percent, the data must include at least one month in the year prior to implementation; therefore
the “furthest a State could look back in requesting a waiver for January 1, 2018, implementation
would be the 12-month period of February 2016 through January 2017.”359 Local unemployment
data is generally available with a lag of about two months, so January 2017 data would be available in
early March 2017 or so.360 In addition, annual revisions from BLS that typically occur in April can
substantially change recent estimates of unemployment and thus substantially alter waiver eligibility,
so states often have to confirm waiver requests submitted before the April revision to ensure that
the waiver request reflects the most up-to-date data.361 The state would therefore have at most ten
months total (or nine months if waiting for the BLS update) to: analyze the data, prepare a waiver
request, receive approval through the state’s internal process (a process that the deeply flawed
proposal to require the Governor’s endorsement could substantially lengthen), submit the waiver
359

U.S. Department of Agriculture, Food and Nutrition Service, “Guide to Supporting Requests to Waive the Time
Limit for Able-Bodied Adults without Dependents (ABAWD),” December 2, 2016, https://fnsprod.azureedge.net/sites/default/files/snap/SNAP-Guide-to-Supporting-Requests-to-Waive-the-Time-Limit-forABAWDs.pdf
360

For reference, on March 15, 2019, the BLS website stated it would release February 2019 state data on March 22 and
local data (such as counties and metropolitan areas) on April 3, 2019, a typical lag of around two months for local
unemployment data. https://www.bls.gov/lau/, accessed March 15, 2019.
361

BLS updates sub-state data annually, typically in April. https://www.bls.gov/lau/launews1.htm

145

request to FNS (including through the regional office, which must review and then forward it to the
national office), receive approval from FNS after its review, and prepare for implementation of the
waiver by taken the steps outlined above, such as identifying and notifying participants,
programming computer systems, and training caseworkers.
Given that the waiver preparation and internal review process may take at least a month or two
within a state, if FNS review extends into several months, that can leave states with very little time
for preparation given the complexities of implementation outlined above. The certainty that they can
implement the requested waiver allows states to plan more effectively, while also allowing FNS time
to review and issue an approval, knowing that its review does not hinder the state from preparing
for implementation. Given that FNS seeks to eliminate the provisions that currently allow states this
certainty without committing to approval within a certain timeframe, this proposed rule will instead
make it harder for states to plan effectively.
One recent example of why this ability helps states is California’s experience with its 2018 waiver,
which the state began to implement prior to approval when waiting for an extensive and lengthy
FNS review process. As mentioned in Chapter 2, any uncertainty has arisen due to the most recent
Administration substantially delaying the waiver review process. In September 2017, California
submitted a waiver request for areas with unemployment rates 20 percent above the national average
(one of the categories of “readily approvable” areas), with an implementation date of September
2018. Given that California was transitioning off a statewide waiver and implementing the time limit
for the first time in several years in some counties, and given the complexities with a large, countyadministered state, the state needed at least six months to prepare for waiver implementation. By
February 2018, about five months after the state had submitted the request, FNS still had neither
approved nor denied the request. California wrote FNS that it would prepare to implement based
on its waiver request, given that the request was based on data that fit the “readily approvable”
standard, and requested that FNS advise the state by March 2018 if it wished to modify the waiver.
Though FNS approval took over five months in this instance, the “readily approvable” standard
enabled the state to properly plan for implementation. Unless FNS plans to impose deadlines on its
own review and approval process that will ensure a timely response to states via the regulation,
taking away these provisions will result in substantially less certainty for states as they await FNS
approval.
The Department claims that the NPRM would improve consistency in the waiver approval
process, but eliminating these provisions would introduce more uncertainty and inconsistency. The
Department several times makes clear that one of the motivating factors for the NPRM is to
improve consistency, such as stating, when introducing “core standards,” that “The Department
proposes updating criteria for ABAWD time limit waivers to improve consistency across States.”362
Reducing the ability of States to predict approvals and await FNS approval, therefore cutting into
implementation planning time, would result in states’ planning becoming more contingent on the
length of time that various steps in the waiver preparation and approval process take. Factors such
as the length of the approval process within the state and the length of FNS approval would have
even more weight on the length of time states have to implement. A state waiting six months for
approval would have significantly less effective implementation time than a state waiting three
months. It is not clear if the Department considered the effect of the elimination of these provisions
362

NPRM, p. 983.

146

on the consistency of time limit implementation outcomes. If so, the Department did not explain
how eliminating these provisions could affect implementation and how it weighed those costs
against what it perceived to be the benefits of improved oversight, for which it did not articulate a
need.
FNS proposes substantially limiting the ability of states to plan for implementation while waiting
for waiver review. In proposing this change, FNS makes statements about the intended effect of the
proposal to increase oversight without explaining why this change is necessary or acknowledging the
substantial burden it could impose on states and clients subject to the time limit. We recommend
that FNS keep the current language in regulation that gives states more certainty around approval,
which lets states better plan for waivers.

B. Proposed Implementation Date Would Cause Severe Burden for States
The Department also proposes that the rule take effect in October 2019, only six months after the
end of the comment period for the NPRM — an extremely short period following the final rule’s
publication. The preamble states:
The Department proposes that the rule, once finalized, would go into effect on October 1,
2019, which is the beginning of federal fiscal year 2020. All waivers in effect on October 1,
2019, or thereafter, would need to be approvable according to the new rule at that time. Any
approved waiver that does not meet the criteria established in the new rule would be
terminated on October 1, 2019. States would be able to request new waivers if the State’s
waiver is expected to be terminated.363
The Department clearly is not considering the length of time states need to prepare a waiver
request, the time it takes states and FNS to review and approve waivers, or the substantial time it
takes states to ensure that they can prepare for implementation and properly notify individuals
subject to the time limit. Given that the comment period ends in April 2019, it is not plausible that
there would be anywhere near enough time for any one step of this process, let alone all of them. In
the past, when many states have waivers due at the same time, this has substantially delayed FNS
review; with this rule implementation, at least 30 states would likely attempt to submit a waiver
request at the same time. FNS does not acknowledge the additional resources it would need to
designate to review these requests, which would need to happen within a very short time frame.
Nor does FNS acknowledge the burden this proposal would place on states, which would need to
devote resources to quickly analyzing the data to put forward new requests and to implementing the
time limit in new areas, and on the participants who would be harmed by a likely chaotic
implementation in many states.
In addition, FNS does not put forward any need that would justify this short timeline. As we have
explained, the current regulations changed little since the 1996 guidelines; in practice, then, states
have been operating under current waiver criteria for more than 20 years. FNS proposes to make
significant changes to longstanding policy without articulating a need for this change, but also
proposes an extremely short timeline that it does not attempt to justify. The lack of explanation for
the proposed implementation date suggests that in proposing this rule, the Department did not fully
grapple with the realities of implementing these changes. Without any explanation of why such a
363

NPRM, p. 983.

147

drastic change would be necessary under such a short timeline, and without any consideration of the
downside of forcing states to implement the time limit in new areas with very little preparation time,
the Department leaves us with little opportunity to address the unstated need motivating this
change. Withdrawing the proposed rule would be the best solution to avoid a rushed
implementation of an ill-considered and harmful policy.

C. Limiting the Duration of Certain Waivers to the Fiscal Year in Which
They Are Implemented is Unnecessarily Restrictive
The Department proposes that waivers based on the 20 percent standard outlined in paragraph
(f)(2)(ii) would not be approved beyond the fiscal year in which the waiver is implemented. Since
most waivers are currently and likely would continue to be requested under the criteria specified in 7
C.F.R. § 273.24(f)(2)(ii), it’s likely that this shift would mandate that most waivers shift to a fiscal
year cycle.
As of March 2019, 36 states (including Guam and the Virgin Islands) have ABAWD time limit
waivers. Nine states are on the federal fiscal year 2019 cycle, 19 states are on the calendar year 2019
cycle, and eight states are covering parts of both fiscal year 2018 and 2019. This grouping of waivers
around calendar year and fiscal year is a relatively new phenomenon that is an outgrowth of two
pieces of legislation that motivated states to pursue waivers along those time cycles:
• The

American Recovery and Reinvestment Act (ARRA) of 2009. That legislation suspended
the time limit, with a state opt out, through FY2010.

• The

Emergency Unemployment Compensation program (EUC) that operated through
December 2015. Many states sought statewide waivers through their eligibility for Extended
Benefits (EB) under EUC.

As the statutory suspension of the time limit (set under a fiscal year cycle) expired and as state
eligibility for waivers under EUC phased out in 2016 (set under a calendar year cycle), states sought
to renew their waivers using alternative criteria but according to the new time cycles. Prior to the
passage of ARRA and EUC, states waiver cycles were spread throughout the year with many
running from April to May.
In the NPRM, FNS claims that the proposed rule would prioritize recent data by preventing states
from requesting to implement waivers late in the federal fiscal year. This proposal would actually
have a different outcome because states would have fewer recent periods of data available to use
under this criterion. Under the proposed rule waivers beginning in fiscal year 2020 can use
unemployment data starting no earlier than January 2017, so approximately five 24-month time
periods would be available to states. In contrast, for a waiver starting in January 2020, states would
have eight 24-month time periods of unemployment data to use (including three more recent than
under the fiscal year calendar scenario). Shifting states to a fiscal year waiver calendar removes the
current option that states have to avail themselves of the most recent data.
For example, states typically submit waiver requests three to six months prior to the waiver
implementation date to give FNS sufficient time to process waiver requests. For a waiver to begin
on October 1, states are recommended to submit the waiver in June, when approximately five 24148

month time periods would be available. For a waiver to begin on January 1, states would have about
three additional and more recent time periods to use.
The proposed rule would also force states to have short waivers under some circumstances.
States are permitted to submit a waiver at any point during the year. This is an important feature in
times of rising unemployment when states may wish to submit new waivers for newly eligible areas.
If a state wants to request a new waiver or modify a waiver after October 1 based on more recent
unemployment data, a waiver would need to be approved for less than a year under the proposed
rule. This would impose addition paperwork for waiver renewals on states during an economic
downturn because states that submit new waivers during the fiscal year would not get a 12-month
approval regardless of how distressed their local labor market is.
Also, this limitation does not give states sufficient time to plan and implement waivers. As noted
above, states typically submit waiver requests well in advance of their start date to allow for needed
implementation planning as well as FNS’ slow processing. States do not know their eligible areas
until late April when BLS revises historical estimates for substate areas from the Local Area
Unemployment Statistics (LAUS) program. These revisions reflect new population estimates from
the Census Bureau, updated input data, and estimation.364 If states need to submit a waiver in June
(for an October 1 start), they would only have one to two months to plan and request the
subsequent waivers. This would likely be particularly challenging for states pursuing a thoughtful and
thorough implementation of the new waiver.
Consider states that have to shift from a statewide waiver using the EB criteria (which would not
be limited to a fiscal year cycle under the proposed rule) to a 20 percent above the national average
criterion (which would operate on a fiscal year cycle only under the proposed rule). For example,
consider a state that has a statewide waiver based on the EB criteria running from September to
August. For the next waiver (which would likely be based on the criteria listed in in paragraph
(f)(2)(ii)), the waiver could only be for one month because the fiscal year runs through September. If
the state wanted continuous waiver coverage, it would have to request a 1 one-month waiver for the
month of September, then a one-year waiver starting in October. These multiple requests would
create additional administrative work for states and FNS.
Moreover, FNS does not have enough capacity to process waivers if they are all on the fiscal year
cycle. And, as we comment elsewhere, the Department is not imposing a timely review on itself
which has resulted in delayed approvals. These delays would only grow worse if virtually all waivers
were on the same cycle for review.
This proposal is flawed and should not be included in the final rule.

D. Limiting Waivers to One Year Would Impose an Unnecessary
Administrative Burden on states
The proposed rule would limit the duration of waiver approvals to one year. We believe this
would impose an additional administrative burden on states that is unjustified and unnecessary. In
the NPRM, the Department asserts that limiting waivers to one year would ensure that the waiver
364

Bureau of Labor Statistics, Annual Revisions, https://www.bls.gov/lau/launews1.htm

149

request reflects current economic conditions, but it provides no evidence or discussion to support
this assertion. This makes it difficult to comment on the proposed change and its potential impact
on both the alignment of waivers with current labor market conditions and state agencies. This
section provides an overview of existing requirements for two-year waivers and explains why the
proposed change is unnecessary.
Existing Requirements for Two-Year Waivers Are Already Restrictive
The Department generally approves waivers for one year. Existing regulations state that the
Department reserves the right to approve a waiver for a longer period if the reasons are
compelling.365 In previous guidance, the Department acknowledged the role of two-year waivers in
minimizing administrative burdens on states from preparing annual requests for waivers covering
areas with chronic high unemployment. 366 Areas that qualify for two-year waivers are those that have
had chronic high unemployment and are likely to continue to experience high unemployment. Twoyear waivers have also been used to cover states and sub-state areas hit hard by the Great Recession
of 2007 to 2009.
The data requirements to support a request for a two-year waiver are much more restrictive than
those required for a one-year waiver. The area must satisfy at least one of the following:
• Have

an unemployment rate above 10 percent for the two-year period immediately prior to
request;

• Be

designated as a Labor Surplus Area for at least two consecutive years; or

• Have

an unemployment rate more than 20 percent above the national average for a 36-month
period ending no earlier than three months prior to the request.

The data requirements are more restrictive in several ways. First, an area eligible for a two-year
waiver must have evidence of high unemployment sustained over a significantly longer period of
time than that required for a one-year waiver. For instance, under the third criterion above, the area
must have elevated unemployment over a period that is 50 percent longer than that required to
support a one-year waiver (36 months compared to 24 months). These are areas with persistent,
chronic high unemployment and are likely to continue to experience adverse labor market
conditions. As we saw during the Great Recession, areas eligible for two-year waivers included those
that experienced a rapid rise in unemployment rates before the rest of the country or experienced
slower recovery.
Second, a request for a two-year waiver must be supported by very recent data. To be eligible
under the third criterion above, the 36-month period must end no earlier than three months prior to
the request. Given that there already is a time lag of one to two months before BLS Local Area
Unemployment Statistics becomes available at the sub-state level, this is a very restrictive
requirement.
365

7 C.F.R. § 273.24(f)(5).

366

USDA, “2-Year Approvals of Waivers of the Work Requirements for ABAWDs under 7 CFR 273.24,” February 3,
2006. USDA, “Guidance on Requesting ABAWD Waivers,” August 2006. USDA, “Guide to Supporting Requests to
Waive the Time Limit for Able-Bodied Adults without Dependents (ABAWD),” December 2, 2016.

150

To support a one-year waiver requested on October 2018, for example, a state could submit data
for the period January 2016 to December 2017. This corresponds with the time period used to
compile the LSA list for FY 2019. To request a two-year waiver, the state would have to submit data
no older than June 2018 and the 35 previous months, a period that starts July 2015. Data supporting
a two-year waiver incorporates data both earlier and later than what is required for a one-year
waiver. To qualify, the area would have to have chronic high unemployment and be likely to
continue having it within the time frame of a two-year waiver.
Use of Two-year Waivers Has Been Very Limited
Under existing rules and guidance, waivers longer than one year in duration have only been
requested and approved under limited circumstances, reflecting their more restrictive and extensive
data requirements. Over the two decades of waiver approvals, FNS has approved approximately 900
waiver requests.367 Of those approved requests, only about 6 percent (50 waiver requests) were based
on the 36-month unemployment rate criteria.368 Nearly half of the 50 waiver approvals were in effect
during Federal Fiscal Years 2007 to 2009, helping states that were hit hard by the Great Recession,
like Alaska, Mississippi, Oregon, and South Carolina, weather the economic downturn.369
States have also requested two-year waivers to cover jurisdictions and Native American tribal
areas that have had chronic high unemployment. For example, Nebraska prepared, applied for, and
received two-year waivers for tribal areas in FFY 2002 (waiver in effect May 2002 to April 2004),
FFY 2004 (waiver in effect May 2004 to April 2006), FFY 2006 (waiver in effect May 2006 to April
2008), and FFY 2008 (waiver in effect May 2008 to April 2010). Had the proposed rule been in
effect, Nebraska would have had to apply for one-year waivers for these tribal areas every year. The
Nebraska state agency only had to apply four times instead of eight times to cover these areas with
chronic high unemployment from May 2002 to April 2010.370
The Department has also approved waivers longer than a year on a case-by-case basis to
accommodate states facing unusual administrative constraints. For example, it approved a 17-month
waiver (from May 2007 to September 2008) for Utah to ease administrative burdens while the state
was transitioning to a new eligibility system.

367

A waiver request includes one or more jurisdictions in a state and may even cover the entire state. Each approved
waiver request is given a distinct waiver serial number by FNS. States may not implement waivers in all areas approved
by FNS.
368

A waiver request includes one or more jurisdictions in a state and may even cover the entire state. Each approved
waiver request is given a distinct waiver serial number by FNS. Based on CBPP internal records and summary of twoyear waivers from SNAP three-month time limit. Prepared in March 2019.
369

A table with the states that have had 2-year waivers from the time limit is included in Appendix B as, “Center on
Budget and Policy Priorities. Summary of 2-Year Waivers from SNAP Three-Month Time Limit.”
370

The American Recovery and Reinvestment Act went into effect April 1, 2009, suspending the time limit in all states
through September 30, 2010 unless state agencies chose to impose specific work requirements.

151

Limiting Waiver Duration to One Year Is Inefficient
Given the more restrictive data requirements, areas eligible for a two-year waiver are experiencing
chronic high unemployment and would likely be eligible for one-year waivers in two or more
consecutive years. By prohibiting waivers longer than a year, the Department would be requiring
states to prepare and submit waiver requests twice over the course of a two-year period, instead of
submitting a request once. Our analysis finds that most areas approved for two-year waivers in
FFY16-17 would have qualified for the second year,371 so requiring the state to submit — and FNS
to review — the information would have been inefficient and burdensome.
The existing data requirements for a two-year waiver capture high unemployment using data that
is very current. The Department did not substantiate its assertion that a one-year time frame would
ensure that waiver requests reflect current economic conditions. Nor did it discuss why the
proposed change is warranted given that it would add administrative burdens both to state agencies
preparing waiver requests and the Department itself. The option to request a two-year waiver is
already very restrictive and limited in use. We therefore recommend that the Department abandon
its proposal to limit waivers to one year and keep the existing rules allowing two-year waivers as they
are.

371

The Department approved two-year waivers covering 19 jurisdictions (7 states, 1 island, and 11 Indian reservations)
in Federal Fiscal Years 2016 and 2017. Of the 19 jurisdictions, 17 would have been eligible for back-to-back one-year
waivers.

152

Chapter 9. Eliminating the Carryover of Unused
Individual Exemptions Would Cause Hardship and
Exceeds Agency Authority
In addition to significantly restricting the ability of states to request waivers of the three-month
time limit, the NPRM proposes to eliminate the accrual of unused individual exemptions for more
than one fiscal year. As a result, some individuals who might otherwise be exempted from the time
limit would lose SNAP benefits and the program’s integrity would be undermined as states would be
less able to judiciously exempt particularly vulnerable individuals. The NPRM fails to define a
problem it is addressing with this proposal, incorrectly reads the intent of Congress, and proposes a
less effective alternative.
Under current law, states can exempt a limited number of individuals who are, or would be,
subject to the time limit. Each year, FNS is required to estimate the number of exemptions available
to each state, based on a percentage (currently 12 percent as revised from 15 percent in the 2018
Agricultural Improvement Act) of “covered individuals.” These “covered individuals” are SNAP
participants subject to the time limit during the fiscal year or individuals denied eligibility in SNAP
because of the time limit.
It is disconcerting to note that the NPRM incorrectly describes the way in which exemptions are
calculated. The preamble describes “covered individuals” as “the ABAWDs who are subject to the
ABAWD time limit in the State in Fiscal Year 2020 and each subsequent Fiscal Year.” But this is not
a correct description of “covered individuals.” Section 6(o)(6)(A)(ii) of the Food and Nutrition Act
(7 U.S.C. § 2015(o)(6)(A)(ii)) defines a “covered individual” as “a member of a household that
receives [SNAP], or an individual denied eligibility for [SNAP] benefits solely due to paragraph (2)” (emphasis
added), with several additional clarifications. As we discuss in our comments on the Regulatory
Impact Analysis, the imprecise use of “ABAWD” makes it unclear whether the NPRM is accurately
describing the group of SNAP participants who form part of the pool that is used to determine the
number of exemptions, but the NPRM also fails to include individuals denied eligibility due to
failure to meet the time limit requirements. As most “ABAWDs” subject to the rule lose benefits
over time, this can be a significant number of individuals.

A. There Is No Statutory or Legislative Support for the Claim That Unused
Exemptions Cannot Be Kept By States
The NPRM suggests that Congress did not explicitly intend for states to maintain and accrue
unused exemptions, but this is not supported by the record. The NPRM describes the accrual of
unused exemptions as an “unintended outcome of the current regulations.”372 It further expresses
concern that “such an outcome is inconsistent with Congressional intent to limit the number of
exemptions available to States each year.” The NPRM does not provide any evidence supporting
this claim of Congressional intent. We are unable to find any record of Congressional intent to limit
the carryover of unused exemptions. The historical evidence and recent actions by Congress show
the opposite.
372

NPRM, p. 987.

153

Congressional history shows that exemptions were enacted in legislation approximately one year
after the time limit was enacted precisely due to concerns that the policy was too harsh and states
did not have enough tools to mitigate the impact of the time limit for vulnerable individuals living
outside of waived areas. Adding this resource gave states an additional way to protect vulnerable
residents not specifically identified in the exemptions from the time limit provided under 7 U.S.C. §
2015(o)(3), based on the priorities and concerns of the state or local agency.
The Balanced Budget Act of 1997 contained two major changes in SNAP to ameliorate the
impact of the three-month time limit. One was an increase in funding for ABAWD training slots in
the Employment and Training (E&T) program. The other was providing states with the authority to
exempt a limited number of individuals from the time limit.373 Commonly referred to as hardship
exemptions, these gave states the ability to continue to provide SNAP to individuals subject to the
time limit who could not find jobs or training slots after three months of participation. Just after
passage of this change, FNS clarified in an October 1997 guidance to states that unused exemptions
could be carried over or saved for future use.374
The current individual exemption policy has been in place for over 20 years. Congress did recently
intend to limit exemptions, but not in the way proposed in the NPRM. Instead Congress reduced
the percentage of exemptions created each year, but explicitly left the longstanding accrual policy in
place. In the 2018 farm bill, Congress reduced the annual percentage of exemptions from 15 percent
to 12 percent, but notably did not propose ending the practice of accruing unused exemptions. In
fact, the Conference Report to accompany H.R. 2, the Agricultural Improvement Act, clarified that
“States will maintain the ability to exempt up to 12% of their SNAP population subject to ABAWD
work requirements, down from 15%, and continue to accrue exemptions and retain any carryover exemptions
from previous years, consistent with current law.”375 (emphasis added). Congressional intent as recently as
several months ago shows a deliberate expectation that states can carryover an unlimited number of
unused exemptions.
The Statute Clearly Allows States to Accrue Unused Exemptions
By drastically reducing the way in which states that choose not to use exemptions in the year in
which they are issued are able to accrue these exemptions, the NPRM suggests that the current
policy is an interpretation of the intent of the underlying statute. However, the statute is less
confusing than it appears. It authorizes states to exempt up to 12 percent of the caseload (formerly
15 percent) but does not mandate that states use the exemptions over any particular time period. It
then, separately, authorizes the Secretary to adjust the number of exemptions based on the state’s
use of exemptions in the prior fiscal year. Under the provision, if a state does not use all exemptions,
the Secretary increases the number of exemptions available in the current year. If the state overuses
exemptions, then the Secretary reduces the number of exemptions available in the current year. The
statute reads:
373

Section 1001 of P.L. 105-33.

374

U.S. Department of Agriculture, Food and Nutrition Service, “Implementation of the Provisions of the Balanced
Budget Act of 1997 Relating to Exemptions for Able-Bodied Adults without Dependents (ABAWDs),” October 27,
1997.
375

House of Representatives, Conference Report to Accompany H.R. 2, December 10, 2018, p. 616,
https://www.agriculture.senate.gov/imo/media/doc/CRPT-115hrpt1072.pdf.

154

. . . the Secretary shall increase or decrease the number of individuals who may be
granted an exemption by a State agency under this paragraph to the extent that the
average monthly number of exemptions in effect in the State for the preceding fiscal
year under this paragraph is lesser or greater than the average monthly number of
exemptions estimated for the State agency for such preceding fiscal year under this
paragraph.
The language sets out that the Secretary adjusts one way for one circumstance (too many
exemptions used), and in another way for the other condition (fewer exemptions used than issued).
The Secretary shall increase the number of individual exemptions to the extent that the average
monthly number used in the previous year is less than then number estimated for that year.
Similarly, the Secretary shall decrease the number of individual exemptions to the extent that the
average monthly number used in the previous year is more than the number estimated for that year.
Note that the Secretary is required to adjust the number of exemptions, but that the use of
exemptions remains a state option (“individuals who may be granted an exemption”). And, if the
state uses fewer exemptions than allotted in the previous fiscal year, the Secretary should increase
the number of exemptions in the following year. The provision requires the Secretary to “increase or
decrease” exemptions depending on whether the state’s use of exemptions is “lesser or greater than”
the allotment for the previous year. So, it’s an increase if the state uses fewer exemptions and a
decrease if the state uses more exemptions than allotted. This makes sense. By decreasing the
allotment to a state that overuses the exemptions, the statute ensures that states cannot routinely use
more than the yearly allotted amount. But that means that a state does increase its allotment each
year that it does not use that year’s amount. States that repeatedly underuse allotments will accrue a
bank of exemptions. This approach, codified in the current regulations, is a straightforward and fair
reading of the statute’s directive.
The proposed rule, in contrast, makes several unsupported assertions. First, it claims without
support, that the intent was not to accrue exemptions for more than one year. Second, by
eliminating the existing supply of unused exemptions, it treats them as having no value to the state
even though many states have accessed these accrued exemptions for a variety of allowable and
sensible reasons. Third, it fails to explain why the current procedure to adjust exemptions each year
is a flawed reading of the underlying statute.
Legislative History Demonstrates That Congress Fully Understood
and Approved of the Uncapped Accrual of Exemptions
The guidelines explaining the calculation and use of individual exemptions were first promulgated
in the September 3, 1999 interim rule implementing two SNAP provisions in the Balanced Budget
Act of 1997.376 In that interim rule, the Department outlined how it would comply with the statutory
requirement that the Secretary adjust the number of individuals who may be granted an exemption
to account for any difference between the average of exemption used and the number estimated by
the Agency for the preceding fiscal year. If a state uses more exemptions than estimated, the state’s
376

U.S. Department of Agriculture, Food and Nutrition Service, “Food Stamp Program: Food Stamp Provisions of the
Balanced Budget Act of 1997,” Interim rule, Federal Register Vol. 64, No. 171, Sept. 3, 1999, p.48246,
https://www.govinfo.gov/content/pkg/FR-1999-09-03/pdf/99-23017.pdf.

155

subsequent allocation is reduced. Likewise, if a state uses fewer exemptions than estimated in the
previous year, the state’s subsequent allocation is increased by the amount not used. As the
Department explained “if this level of exemptions is not used by the end of the fiscal year, the State
may carry over the balance.”377
This longstanding implementation of the statutory directive is clear, reasonable, and fair to states.
It addresses the reasonable concern that an annual allotment of exemptions could be either overused
or underused. The continual overuse of individual exemptions has an impact on overall program
integrity because individuals not eligible under an exemption are issued benefits, which is an
overissuance and error. To address this, the regulation treats this issue in a sensible way, by reducing
future exemptions. The continual underuse of individual exemptions does not create the same
problem, and the regulation’s treatment is similarly reasonable.
The statute does not direct the Secretary to make adjustments beyond a one-year period. In other
words, the statute does not give the Secretary the authority to adjust the number of exemptions
issued more than one year prior ago. Combining the requirement that the Secretary adjust
exemptions from the previous fiscal year with the limitation on looking further back to adjust
exemptions based on use means that a state can accrue unused exemptions in multiple years, and
these exemptions can accrue over multiple years.
States Have Relied on Current Policy:
USDA Has Never Emphasized the Need to Use Exemptions Each Year
The Department has not, in the past, suggested that unused exemptions would not accrue.
Developing a reasonable exemption policy is difficult — states must identify the circumstances
when an individual exemption should be used, the procedures for identifying when that
circumstance has occurred, and a tracking mechanism to ensure that the usage does not exceed the
allotment. This implementation challenge has discouraged states from experimenting with ways of
using the exemptions. But it does not indicate that states have no need for them. Instead of
eliminating earned-but-unused exemptions, the Department could provide guidance to states on
effective ways to use them. The Department could take steps to understand states’ concerns or
problems with using exemptions. Such a response would be much more in keeping with
Congressional intent and the law. Instead, via the NPRM, the Department has taken sweeping
measures to curtail a state resource counter to the law.
States Have Compelling Reasons to Accrue Individual Exemptions
The recent statutory change from 15 percent to 12 percent makes the banked or unused
exemptions more important for some states. While not every state uses its annual allotment of
exemptions, some states do or come close to doing so. Many states use the ability to rollover
exemptions to build a “bank” that gives states options that would be unavailable if exemptions
expired.

377

U.S. Department of Agriculture, Food and Nutrition Service, “Food Stamp Program: Food Stamp Provisions of the
Balanced Budget Act of 1997,” Interim rule, Federal Register Vol. 64, No. 171, Sept. 3, 1999, p.48249,
https://www.govinfo.gov/content/pkg/FR-1999-09-03/pdf/99-23017.pdf.

156

States use individual exemptions for a variety of purposes, as the original provision intended.
Some identify certain vulnerable populations, such as victims of domestic violence, veterans, young
adults aging out of foster care, or those with acute barriers to employment like a lack of education or
limited proficiency in English. States have also used exemptions to allow individuals in limited areas
to remain eligible for SNAP, often because of circumstances that are not reflected in a way that
qualifies the area for a waiver or due to administrative demands. In all cases, states must estimate the
number of individuals who would receive an exemption and for how long in order to ensure that the
state does not exceed the number of available exemptions. Building up some unused exemptions
gives states important flexibility and confidence to implement these targeted approaches without
running afoul of overissuing exemptions. The buffer provided by accrued exemptions is critical in
that process.
Because of the recent change in the percentage of exemptions made available to states, these
states that are using exemptions would be at risk of exceeding their allotment and being subject to
error determinations and overpayments under the NPRM. Between 2014 and 2017, 28 states used
more exemptions than they had been issued for the fiscal year, meaning they used at least some of
the exemptions they had accrued in previous years. During that time period, some states did so for
more than one year. Several of the states used all of their multi-year exemptions which demonstrates
the importance of accruing exemptions over several years.
For example, Washington used 28,886 exemptions in fiscal year 2016.378 It had earned no
exemptions in the prior year (because it had a statewide waiver in 2015). It had accrued 11,530
exemptions in previous years. It did not overuse exemptions because it was allocated 26,784
exemptions that year (meaning that it started 2017 with over 9,000 unused exemptions). The state
relied on exemptions that year because it was transitioning off of the statewide waiver and was
developing training programs and operational procedures for childless adults subject to the rule.
Other states used banked exemptions in a similar way. For example, in 2016, Maryland earned no
exemptions for the year (based on having a statewide waiver in 2015) but issued 18,871 exemptions
to aid in its transition to the time limit. It could do so only because it had a “bank” of unused
exemptions from prior years of 18,915. Under the proposed rule, neither state would have had been
able to take this approach.
The three-month time limit is complex and difficult to administer, as is documented in the USDA
Inspector General’s report.379 A majority of states have used individual exemptions to ensure that
particularly vulnerable individuals are not inappropriately terminated from the program. Allowing
states to keep unused exemptions enables states to plan in advance and prepare for major events
affecting the unemployed childless adult population on SNAP (such as an area transitioning from
waived to unwaived status).

378

U.S. Department of Agriculture, Food and Nutrition Service, “SNAP – FY 2017 Allocations of 15 Percent
Exemptions for ABAWDs – Totals Adjusted for Carryover,” March 15, 2017, https://fnsprod.azureedge.net/sites/default/files/snap/FY2017-ABAWD-15%25-Exemption-Totals.pdf.
379

U.S. Department of Agriculture, Office of Inspector General, “FNS Controls Over SNAP Benefits for Able-Bodied
Adults Without Dependents,” September 2016.

157

Congress Knows How to Limit Carryover
and Has Repeatedly Declined to Do So for Unused Exemptions
Congress has the authority and ability to limit the carryover of allocated resources in the
legislation it crafts. This authority is exercised frequently, in order to prevent unused funds or
resources from accruing. In fact, the Food and Nutrition Act demonstrates that Congress, when it
deems it appropriate, can limit or reallocate resources, though it has not done so for individual
exemptions. For example, in allocating funding for SNAP Employment and Training program,
Section 16(h) (7 U.S.C. § 2025(h)) reads:
(C) Reallocation—
(i) In General.—If a State agency will not expend all of the funds allocated to
the State agency for a fiscal year under subparagraph (B), the Secretary shall
reallocate the unexpended funds to other States (during the fiscal year or the
subsequent fiscal year) as the Secretary considers appropriate and equitable.
Here, Congress not only directs the Secretary to reallocate unspent funds but indicates when such
reallocation occurs. Further subsections provide more detail on the mechanics of the reallocation.
There is no similar provision in the Food and Nutrition Act indicating that Congress intended to
limit the accrual of unused exemptions, or indeed, any directive for the states once exemptions are
provided.380

B. The Proposed Rule Change Fails to Provide a Legitimate Reason for the
Change
The NPRM states that the change would result in administering the program more efficiently and
to further the Department’s goal to promote self-sufficiency. However, the NPRM provides no
explanation or information on how the proposed change would achieve either goal. The current
exemption policy has worked well for 20 years and FNS has never identified issues with the
efficiency of the policy. Nothing in the proposed replacement policy would make it easier for state
to administer. Indeed, because the safety valve of a bank of exemptions is eliminated, states will find
it more difficult to fine-tune policies that authorize the use of exemptions.
For example, a state may decide to provide an exemption to any individual who is working, but
not enough hours to meet the 20-hour-per-week requirement for those subject to the time limit. It
can estimate the number of individuals, and hence the number of exemptions. But inaccuracies in
the calculation of the estimate or changes in the composition of the group can significantly change
the number of exemptions needed. The existing policy provides states with a pool of unused
exemptions a way to adjust; the proposed rule almost completely eliminates this ability to adjust. As
a result, every state would be more at risk of exceeding its annual allocation of individual
exemptions.

380

Section 6(o) (7 U.S.C. § 2015(o)) of the Act does direct the Secretary to make limited adjustments to exemptions each
year, but these are limited to changes in caseload and not based on whether or not a state used the exemptions issued.
The E&T funding, by contrast, is adjusted based on state decisions to spend the allocation.

158

In the NPRM, the Department references the September 2016 report from the Office of
Inspector General to support the proposed change in the accrual of unused exemptions. While it is
true that the report notes the large number of accrued exemptions, the report very explicitly declines
to recommend any change in current policy. The report states “OIG generally agrees that FNS has
the discretion to interpret and implement the exemption provisions as it has done, so we do not
have a recommendation for FNS with respect to exemptions.” 381
The Proposed Method of Calculating Exemptions and Adjusting From Prior Years Will
Discourage States From Using Them and Increase the Potential for Errors
The proposed adjustment procedure in the NPRM is needlessly confusing, will discourage the use
of exemptions and is likely to increase errors. The varied exemption use example provided in the
NPRM (Example 2 on page 988) shows how this proposed approach would discourage the use of
allocated exemptions and contradicts the statutory requirements. In 2021, the state uses eight
exemptions.382 In 2022, use plummets to two. In 2023, use quadruples to eight, and in 2024, it drops
again to two. While the math works out to meet the proposed rule, the implementation of a policy
that varies this widely in scope is hard to conceive. The state must be able to estimate the number of
exemptions that would be used each year, design a policy and the procedures to implement to meet
that target number, correctly train staff, and actually implement and track. Then, the following year,
the state must design a policy that uses four times fewer (or greater) the number of exemptions,
retrain staff, and properly implement. The NPRM offers no assurance that a state could successfully
resdesign important program elements on a yearly basis. The history of state administration of
SNAP also offers no assurance.
In 2022, the state does not have its yearly allocation available, because in the prior year it tapped
into previously earned exemptions. The state is paying back exemptions despite not overusing the
total available exemptions in any year in the example. That conflicts with the statutory authority
granting states exemptions in each year in which individuals are subject to the time limit or ineligible
because of it. And, averaging over two years so that the average is equal to 12 percent of the
“ABAWDs” does not fulfill the statutory requirement that states can allocate an average of 12
percent per year.

381

USDA Office of Inspector General, FNS Controls Over SNAP Benefits For Able-Bodied Adults Without Dependents,
September 2016, p. 11.
382

NPRM, p. 988.

159

Chapter 10. The Proposed Rule Fails to Provide
Sufficient Rationale or Supporting Evidence for the
Proposed Policy Change
The NPRM proposes several significant changes to long-standing SNAP policy that would affect
an estimated 1.1 million low-income Americans, including 755,000 who would lose food assistance
and face increased financial and food insecurity. Despite the far-reaching impact of the proposed
rule, and contrary to requirements in the rulemaking process, the NPRM fails to provide a
meaningful rationale for most of the proposed changes and fails to identify or summarize any
research and data to support the rationale for such sweeping and consequential changes. Without
knowing what evidence justifies such a drastic change in long-standing policy, it is impossible to
assess the validity of the claim or the soundness of the evidence used to support it.
The goal of the Department’s proposed changes in policy appears to be to subject more people to
the time limit by shrinking the portion of the country that can request waivers from it. To achieve
this goal, the proposed rule would prohibit waivers of the time limit to those that are based on a
general unemployment rate of at least 7 percent and at least 20 percent above the national average
and would restrict the ways in which a state can define the area it seeks to waive. In addition, the
NPRM eliminates several ways in which a state can demonstrate a locale has an insufficient number
of jobs for those subject to the rule and ends the ability of states to save unused exemptions to the
time limit. In each case, the Department fails to identify a desired goal or outcome (such as a certain
percentage of the target group gaining employment) or explain how the proposed rule would lead to
the desired goal, and consistently fails to provide any empirical support for the proposal. These
failures prevent the public from understanding why the existing rule needs to be modified so
drastically.

A. The Proposed Rule Fails to Support the Justification for New
Rulemaking — That Too Many Unemployed Adults on SNAP Are Not
Subject to the Time Limit
The preamble to the proposed rule states that the Department now believes that the time limit for
unemployed adults was intended to apply to more individuals than it currently does. The
Department thus believes that too many individuals live in areas that are waived by states and not
subject to the rule. But the Department fails to make several important connections to justify the
need for new rulemaking.
The Department fails to show that the intent of Congress was to subject the Department’s
preferred number of individuals to the time limit. In fact, the NPRM fails to establish that Congress
had any interest in subjecting a target number of individuals to the rule. Rather, the goal was to
allow states to protect individuals in areas without a sufficient number of jobs, regardless of how
many individuals that would be. We are unable to identify any statutory reference to a policy goal of
protecting only a certain percentage of individuals subject to the rule.
The legislative record does not reveal congressional debate over the appropriate percentage of
individuals subject to the rule. In fact, the members of Congress introducing the proposed time
limit emphasized that adequate protections were included to ensure that individuals were not cut off
160

of SNAP if opportunities for work or workfare were not available. Representative Robert Ney, one
of the authors, stated on the floor of the House of Representatives that his amendment “provides
some safety; it provides a course of a safety net [sic], it has the ability to have waivers from the State
department of human services.”383 Representative John Kasich clarified that the key was that the
time limit applied only in areas where jobs were available to those subject to the time limit —
otherwise, the time limit would not apply. “It is only if you are able-bodied, if you are childless, and
you live in an area where you are getting food stamps, and there are jobs available, then it applies.”
The key issue, of course, is whether jobs are available for these individuals, as the statute requires.
We address the serious concerns with the way in which the NPRM incorrectly interprets the
standard that waivers are available in places with insufficient jobs for the individuals subject to the
rule in Chapter 3. Here, we simply note that the authors of the original legislation did not have a
targeted number of individuals they thought should be subject to the rule; nor did Congress
establish, or even debate, a targeted percentage of individuals to be subject to the rule. Instead, the
co-authors of the original legislation were careful to point out that there were adequate protections
for all individuals if jobs were not available. Given this clear history, it is incumbent upon the
Department to substantiate its claim that the legislative history somehow suggests that the current
regulations must be changed because too many individuals live in waived areas. Without explaining
the underlying claim, the rule leaves commenters with little ability to meaningfully respond.
We would also note that the temporary nature of setting such a coverage goal strongly suggests
such a goal is not intended or practical. As economic circumstances change, the ability of ABAWDs
subject to the time limit to find work will change, meaning that at different points in the economic
cycle, the portion of individuals subject to the rule who are able to find 20 hours of work per week
will change significantly, as will the portion of individuals living in areas eligible for waivers under
any set of criteria. And other factors, besides the existence of a waiver, affect an individual’s
participation in SNAP, such as the accessibility of the application process, other eligibility rules and
processes, and the availability of training opportunities for unemployed adults.

B. Despite Claiming That General Unemployment Rates Are the Best
Available Measure of Job Sufficiency for Low-Income Adults on SNAP,
the Proposed Rule Fails to Support the Claim with Evidence
The proposed rule asserts that low general unemployment rates indicate sufficient jobs are
available for those subject to the time limit. But it offers no reason why a general unemployment
rate of 7 percent is a good proxy measure for establishing that there are sufficient jobs for the
individuals subject to the rule. This lack of an explanation makes it difficult for interested parties to
critique the Department’s conclusion that no waivers should be permitted below 7 percent. In
contrast, as discussed in Chapter 3, there is a deep body of research that shows unemployment rates
are much higher for groups that make up SNAP’s ABAWD caseload.
Current regulations allow waiver requests that demonstrate a recent local unemployment rate
significantly above the national average but do not set a minimum unemployment rate. The NPRM
fails to explain why the reasoning behind the current rule no longer applies, and why it believes
another rulemaking process would result in a justifiable change. Indeed, for the most substantial
383

142 Congressional Record, H7904 (daily ed. July 18, 1996).

161

proposed change — to prohibit waivers for areas with unemployment below 7 percent — the
NPRM seeks input for changing the number to 6 or 10 percent but does not explain why those
thresholds are of particular importance, aside from noting that a larger or smaller group of
individuals might be protected at the different levels. It does not explicitly seek input on other
levels, such as 5 or 8 percent. The rule offers a very weak explanation, unsupported by research, for
the 7 percent that is proposed or the alternatives for which it seeks comment. Since interested
stakeholders do not have adequate information to determine why the unemployment rate floor is set
where it is, it is difficult to provide useful feedback on the appropriateness of the proposed
threshold. In the rulemaking process, an agency that promulgates a rule change needs to explain
why the original rationale is no longer sufficient when proposing to change the rule.
The Proposed Rule Makes Arbitrary Changes
to Long-Standing Regulations That Were Initially Promulgated Based on Sound Reasons
Until this proposed rule, FNS has always acknowledged that the statute requires several different
ways for states to document a lack of sufficient jobs for the individuals subject to the time limit. In
its original guidance, the Department noted, “[t]he statute recognizes that the unemployment rate
alone is an imperfect measure of the employment prospects of individuals with little work history
and diminished opportunities.”384 It then proceeds to describe the use of Labor Surplus Areas
(LSAs) as a reliable waiver criteria. However, without providing a reason or evidence that LSAs are
not a useful measure, the proposed rule eliminates LSAs as a possible way of qualifying for a waiver.
We are at a loss as to why a measure relied upon for so long by so many states is simply eliminated.
Areas designated as LSAs by the Department of Labor have been eligible for waivers because in
order to qualify as an LSA, an area must have sufficiently high unemployment (120 percent of the
national average so long as the area rate is at least 6 percent). LSAs are recognized as weak labor
markets. Federal, state, and local government use LSAs to target contracts and allocate
employment-related assistance and training. LSAs provide a reasonable indicator that there is a lack
of sufficient jobs for unemployed SNAP participants, who disproportionately struggle to overcome
barriers to employment.
To support the inclusion of LSAs as a way to demonstrate a lack of sufficient jobs for the
unemployed adults subject to the rule, the original 1999 rulemaking process established that LSAs
were a reasonable measure of labor market weakness and were based on sound and relevant data
from a trusted source (the Bureau of Labor Statistics). The original 1996 guidance explained one
reason why:
Labor surplus areas are classified on the basis of civil jurisdictions rather than on a
metropolitan area or labor market area basis. By classifying labor surplus areas in this
way, specific localities with high unemployment rather than all civil jurisdictions within
a metropolitan area, (not all of which may suffer from the same degree of

384

U.S. Department of Agriculture, Food and Nutrition Service, “Guidance for States Seeking Waivers for Food Stamp
Limits,” Dec. 3, 1996.

162

unemployment) can be identified. This feature also makes the classification potentially useful to
identify areas for which to seek waivers [emphasis added].385
The original rulemaking process emphasized the importance of relying on BLS data (much as the
NPRM does). But the original rulemaking identified LSA status as a reliable indicator of insufficient
jobs based on BLS data and as recent enough to be used to meet the waiver criteria. In fact, the
preamble to the final rule noted that an LSA designation was reliable enough to allow for
“immediate implementation of waivers for areas where the Employment and Training
Administration, U.S. Department of Labor (ETA), has designated such areas as LSAs.”386 In other
words, the Department made a reasoned decision to allow states to immediately implement (before
approval by FNS) any waiver based on an area’s designation as an LSA. The proposed rule both
eliminates the LSA criteria and the immediate implementation of certain waivers without explaining
why the current process is flawed or could be improved.
The proposed rule drops LSAs but provides no explanation for why this change is needed; nor
does it identify deficiencies in the current criteria (aside from determining that there should be a 7
percent unemployment floor). Because we do not know what faults the Department now believes
exist with the use of LSAs as credible indicator of a lack of sufficient jobs for the individuals subject
to the time limit, we are unable to assess the validity of the claim. It is unclear what information the
Department now has that invalidates its decision of more than 20 years ago — a decision the
Department has followed and subscribed to until very recently. This prevents the public from
providing relevant information that supports or refutes the reasons behind the proposed rule.
The Proposed Rule Attempts to Achieve Through Regulation
A Policy That Congress Explicitly Rejected
The Administration’s aim with this rule appears to be to do through rule-making what Congress
rejected through legislation. The Trump Administration proposed restricting waivers from the time
limit through legislation in its fiscal year 2018 budget proposal and promoted exposing more people
to the time limit throughout the 2018 Farm Bill process.387 In the budget, the Administration
proposed restricting waivers to just areas with an average unemployment rate of 10 percent. In that
proposal, the Administration described current policy as, “States can request waivers from the
ABAWD time limit that cover the entire State, or only parts of the State where unemployment is
particularly high. States decide whether or not to request a time limit waiver, and generally make this
assessment annually.” The proposed policy was described as, “This proposal limits ABAWD
waivers to counties with an unemployment rate greater than 10 percent averaged over 12 months.”
While this proposed legislative policy would be stricter than the policy in the proposed rule, the
actual near-term impact of a 7 percent floor would be similar as there are very few counties in the
country with average unemployment rates between 7 and 10 percent.
385

U.S. Department of Agriculture, Food and Nutrition Service, “Guidance for States Seeking Waivers for Food Stamp
Limits,” Dec. 3, 1996.
386

66 Fed. Reg., No. 11, 4438, “Food Stamp Program: Personal Responsibility Provision of the Personal Responsibility
and Work Opportunity Reconciliation Act of 1996,” January 17, 2001, p. 4463.
387

Food and Nutrition Service, 2018 Explanatory Notes for the FY2018 President’s Budget, pages 32-92 to 32-93
https://www.obpa.usda.gov/32fnsexnotes2018.pdf.

163

Throughout the farm bill process, the President and Secretary Perdue were quoted in the press as
saying that they were frustrated that Congress would not expose more individuals to the time limit
or “work requirement.”388 At the 2018 Farm Bill signing ceremony, the President remarked that he
wanted to implement policy counter to what Congress had decided. He said, “Therefore, I have
directed Secretary Perdue to use his authority under the law to close work requirement loopholes in
the food stamp program. Under this new rule, able-bodied adults without dependents will have to
work, or look for work, in order to receive their food stamps. Today’s action will help Americans
transition from welfare to gainful employment, strengthening families and uplifting
communities. And that was a difficult thing to get done, but the farmers wanted it done; we all
wanted it done. And I think, in the end, it’s going to make a lot of people very happy. It’s called
“work rules.” And Sonny is able, under this bill, to implement them through regulation.” 389 As
noted, the farm bill legislation which the President refers to as “the bill”, did not provide any new
authority to the Secretary to change waiver policy.
Congress expressly rejected the Administration’s proposal to substantially limit waivers in favor of
the Senate approach, demonstrating intent to keep the current interpretation of the “insufficient
jobs” criterion intact. Given that the agency did not put forward a coherent evidenced-based
argument, we are left to believe that the goal of this rule is to defy Congressional intent and the
agency’s own rulemaking to achieve a failed legislative effort.

C. The Failure to Provide a Relevant Explanation or Supporting Data to
Justify a Change in Current Regulations Occurs Repeatedly Throughout
the Proposed Rule
Under the NPRM, USDA would simply eliminate several existing criteria for requesting waivers
because the Department claims they are “rarely used, sometimes subjective and not appropriate
when more specific and robust data is available.” Under the proposed rule, waivers would not be
available for areas with low and declining employment-to-population ratios, a lack of jobs in
declining occupations or industries, or a lack of jobs as demonstrated by an academic study or other
publication. The claim that the data available is not rigorous enough to support a request is not
explained, given that states can submit a wide range of data to support a request.
Especially concerning is the elimination of the employment-to-population (E:P) ratio standard. It
is a well-established metric that has several features that make it preferable to general unemployment
rates in assessing the health of the labor market. In some ways and under some circumstances,
particularly in rural areas, the E:P ratio may be a better measure of the availability of sufficient jobs
388

Emily Birnbaum and Julie Grace Brufke, “Trump Attacks Dems on farm bill”, The Hill, September 13, 2018,
https://thehill.com/homenews/administration/406561-trump-calls-out-dems-for-opposing-farm-bill-over-workrequirements
Phillip Brasher, “Farm Bill Delayed, but Perdue signals administration support”, Agripulse, December 3, 2019,
https://www.agri-pulse.com/articles/11703-farm-bill-delayed-but-perdue-signals-administration-support
389

Remarks by President Trump at Signing of H.R. 2, the Agriculture Improvement At of 2018.
https://www.whitehouse.gov/briefings-statements/remarks-president-trump-signing-h-r-2-agriculture-improvementact-2018/

164

for low-income adults participating in SNAP. The Department fails to establish that the E:P ratio
relies on questionable or non-specific data. The proposed rule insists that sound data be used in
supporting waiver requests, emphasizing that data from BLS is the standard to be used in requesting
waivers.
The employment-to-population ratio has not been widely used, but that, by itself, is not a
sufficient reason to eliminate the option for states. As BLS itself notes, the ratio is “especially useful
for evaluating demographic employment trends.”390 In particular, it is important to rural areas,
which often have less dynamic job creation and fewer resources available for the types of training
activities that allow ABAWDs subject to the time limit to meet the 20-hour requirement. For
example, South Dakota has waived both whole counties and reservations under the employment-topopulation criteria. Even in the last few years, other states, like New York and Maryland have
waived counties. The option may not be frequently used, but it represents an important measure of
labor market weaknesses in some areas and should remain available to states. The NPRM does not
explain why frequency of request is a meaningful reason to keep or drop criteria.
Finally, the Department offers no insight into whether it considers the employment-to-population
ratio to be “sometimes subjective.” Under each of the listed concerns used to justify dropping the
employment-to-population criterion (that it is rarely used, sometimes subjective, and not appropriate
if more specific and robust data is available), the NPRM provides no explanation or information that
allows the general public to respond to the proposal.

D. The Public Input Resulting From Last Year’s Advanced Notice of
Proposed Rulemaking Does Not Appear to Inform This Proposed Rule
In March 2018, the Department issued an Advanced Notice of Proposed Rule Making (ANPRM),
seeking public input on “potential regulatory changes or other changes that might better support
states in accurately identifying ABAWDs subject to the time limit and providing meaningful
opportunities for them to move towards self-sufficiency.”391 Tens of thousands of comments were
submitted, but the agency makes only a cursory reference to a subset of the comments and does not
adequately recognize or summarize the public input. While the preamble of the NPRM contains a
brief and inconclusive summary of the submitted comments, the Department provides no
explanation for how the ANPRM informed the policy making process or whether the Department
chose to ignore input provided through the public process. Potential commenters are at a loss for
how the ANPRM informed the development of the proposed rule, what the public response to the
ANPRM was, or how to engage without any information about the comments.
The failure to respond to ANPRM raises serious concerns about the current rulemaking proposal.
We are left to wonder whether the bulk of the comments sought, or did not seek, a change in policy.
There is no summary of the reasons for supporting a change. Nor is there a summary of the input
from commentators who opposed a change in policy, or a response from the agency as to why it
concluded that these commentators were incorrect.
390

Carol Boyd Leon, The employment-population ratio: its value in labor force analysis, Bureau of Labor Statistics,
Monthly Labor Review, February 1981, pp. 36-45, https://www.bls.gov/opub/mlr/1981/02/art4full.pdf.
391

Fed. Reg., Vol. 83, No. 37, “Supplemental Nutrition Assistance Program: Requirements and Services for Able-Bodied
Adults Without Dependents; Advance Notice of Proposed Rulemaking,” February 23, 2018, p. 8013.

165

The ANPRM asked numerous questions about helping ABAWDs gain work. But the NPRM
only references the questions about waivers. This is deeply misleading as it suggests comments were
focused only on that question.
The failure to adequately respond to the ANPRM also raises concerns that the current rulemaking
process will fail to take the comments on the NPRM into account as the Department decides
whether to proceed with the current proposed rule or change or withdraw it. If the public’s input
was ignored or outright dismissed in the previous process, why should the public have confidence
that the NPRM will not yield the same result?

E. Alternatives to the Proposed Rule Are Not Discussed
Under the rulemaking process, USDA is obligated to explain why the particular policy is proposed
and why alternative approaches are inadequate. Given that the agency estimates 755,000 people will
lose benefits and provides no estimate for how many will gain employment, less harmful alternatives
exist and the Department has an obligation to consider these alternatives.
In establishing 7 percent unemployment as a floor under which no area can qualify for a waiver,
the Department claims this is “more suitable for achieving a more comprehensive application of
work requirements so that ABAWDs in areas that have sufficient number of jobs have a greater
level of engagement in work and work activities, including job training.” 392 No information is
provided as to why current policies are not comprehensive and what the current level of engagement
in work and work activities is, much less what a “greater level of engagement” would look like. This
makes it difficult for commenters to provide input on these unsupported assertions.
The claim that a 7 percent floor strengthens the work requirement is repeated throughout the
preamble. As discussed in detail [below], the NPRM fails to adequately support the proposed floor.
Without knowing what research or data the Department relied upon to conclude that 7 percent was
the appropriate floor, the public is unable to directly comment on the validity of the Department’s
action.
The Department does seek input on setting the unemployment floor — at 7, 6, or 10 percent —
but offers no explanation why 6 or 10 percent are the two alternatives rather than, say, 5 or 8
percent. Aside from a cursory mention of the natural rate of unemployment, no discussion or
information is provided to inform the public’s comments.
Other alternatives to grouping areas together also exist. For example, the proposed rule limits
waiver requests that group sub-state areas together to those based entirely on BLS Labor Market
Areas (LMAs). There are serious limitations to relying solely on LMAs as the basis of such waivers,
as discussed in more detail in Chapter 5. LMAs rely on older data, use a narrow definition of a labor
market area that does not reflect the challenges facing low-income SNAP participants, and do not
account for other factors relevant to ABAWDs subject to a three-month time limit — such as the
availability of training programs. In fact, one of the key components of the current grouping policy
– that states largely define the area of the waiver request – is largely eliminated with no explanation
392

NPRM, p. 984.

166

or evidentiary support. Other alternatives do exist, but the NPRM fails to provide any reason why
these alternatives are not appropriate and why the proposed grouping change is the best available
option.
The proposed rule is based on insufficient reasons to change current regulations, fails to provide
evidence supporting the change, and lacks any discussions of alternatives considered in developing
the proposed rule. Given this lack of supporting information, the public has an insufficient
opportunity to comment meaningfully on the proposed rule.

167

Chapter 11. The Proposed Rule’s “Regulatory Impact
Analysis” Highlights FNS’ Faulty Justification and
Includes Numerous Unclear or Flawed Assumptions
The Regulatory Impact Analysis (RIA)393 that accompanies the proposed rule contradicts the
Department’s justification for the proposed rule. The Department repeatedly asserts in the preamble
that the proposed rule would “encourage more ABAWDs to engage in work” and would “promote
self-sufficiency.” But the RIA finds instead that 755,000 individuals would be cut from SNAP in
2020 for “failure to engage meaningfully in work or work training,” 394 and it provides no evidence or
estimates that other individuals would be induced to work because of the proposed changes or
would experience any benefit from the changes.
In addition, the methodology for deriving the impact of the proposed rule ignores available
research evidence, uses imprecise terms, includes numerous unclear or inappropriate assumptions,
and excludes all together any explanations for several other key assumptions. The information that is
provided in the RIA is fundamentally flawed, imprecise, incomplete, and incoherent.
The result is that the proposed rule does not provide the analytical or conceptual information
needed to justify the policy change and to evaluate the proposed rule’s likely impacts. Because of the
deficiencies in reasoning and analysis of the RIA, the proposed rule fails to answer basic questions
related to the impact of the change and the people whom the proposed rule would affect, and so
does not contain the information and data necessary to fully evaluate the proposed rule or to
comment on key aspects on the Department’s justification for the rule.
No agency could explain every nuance and assumption, but the RIA that accompanies the
proposed changes in this NPRM is so deeply flawed that we cannot comprehend the basic reasoning
behind it. Because individuals who wish to comment on the changes cannot understand or follow
the agency’s justification, this rulemaking and comment process is compromised.

A. The RIA Does Not Provide Any Evidence to Support the Proposed Rule’s
Stated Rationale
The NPRM argues repeatedly that “the Department is confident that these changes would
encourage more ABAWDs to engage in work or work activities,”395 implying that, as a result of the
changes proposed, individuals newly subject to SNAP’s three-month time limit in areas no longer
qualifying for waivers would be likely to work more, have higher earnings, or otherwise be better off.
But the NPRM provides no evidence to support these assertions, and no estimates of any
393

The Regulatory Impact Analysis (RIA), which includes the detailed cost-benefit analysis and information about the
methodology, is included in a separate online document here: https://www.regulations.gov/document?D=FNS-20180004-6000, p. 4-7 and 18-31. The NPRM includes only a short summary of the analysis. Hereafter in citations we will
refer to the Regulatory Impact Analysis as the “RIA.”
394

NPRM, p. 989; RIA, pp. 4, 26, 27.

395

NPRM, pp. 981, 982, 987.

168

quantifiable benefits for any individuals resulting from the changes. The NPRM’s failure to justify
the stated rationale is a serious deficiency and makes it impossible for commenters to assess the
impact of the proposed rule or to comment on the Department’s justification.
The RIA, which is included as supplementary materials accompanying the NPRM, contradicts the
Department’s stated rationale. The analysis in the RIA provides that the only benefit of the rule is
budgetary savings from lower SNAP benefits resulting from 755,000 individuals “not meeting the
requirements for failure to engage meaningfully in work or work training.”396 The RIA does not
claim that any individuals would be induced to find work or have increased earnings as a result of
the proposed rule. The RIA does provide a confusing assertion that a higher share of “ABAWDs”
would be working in 2020 (34 percent) than in 2016 (26 percent), and estimates the impact under a
different scenario where that increase does not occur.397 According to the RIA, however, the
assumed increase in employment (from 26 percent with any earnings to 34 percent working at least
20 hours a week) is “based on the projected decline in the unemployment rate” in the President’s
2019 budget forecast, not on more work among low-income households because of the regulatory
change.
The loss of SNAP benefits as a result of fewer areas qualifying for waivers from the time limit is
included in the RIA as a benefit because of the reduction in federal spending, but the RIA does not
quantify any benefits to individuals from the change. There is only a small mention in the RIA of the
harm, or cost that might occur for low-income individuals who lose SNAP:
To the extent that ABAWDs newly subject to the time limit are unable to find work or
otherwise meet work requirements, and thus lose SNAP there may be increases in poverty and
food insecurity for this group. However, those ABAWDs who become employed will likely see
increased self-sufficiency and an overall improvement in their economic well-being. The
Department believes that a number of those affected by strengthened work requirements are
able to secure employment in a wide range of different industries.398 [Emphasis added.]
Thus, the analysis included in the RIA asserts that that the Department believes people are likely
to get jobs because of the rule, but provides no evidence to support that belief and quantifies only
the federal budgetary savings from the estimated reduction in SNAP benefits associated with
individuals’ “failure to engage meaningfully in work or work training.” 399 It further mentions, but
does not quantify, the secondary effects on SNAP retailers from lower SNAP redemptions,400 and
on community-based organizations (i.e., food banks and others that provide emergency assistance)
from increased demand for food and services.401

396

NPRM, p. 989.

397

RIA, pp. 4, 26.

398

RIA, p. 28.

399

NPRM, p. 989.

400

NPRM, p. 990.

401

RIA, p. 28.

169

Because the RIA and its cost-benefit analysis are lacking in internal logic or transparency, the
public cannot see clearly how the Department arrived at its conclusions about the need for the
proposed regulation or its impact.

B. Available Research Evidence Contradicts the Articulated Aims of the
Proposed Rule
The research evidence that is available on the question of the effects of policies that take food
assistance or other benefits away from individuals who don’t meet rigid work requirements is not
mentioned in the RIA. This is a serious omission and constrains the public from being able to
adequately assess and comment on the potential impacts of the proposed rule.
The loss in benefits, and the related increase in poverty and hardship that results from policies
that take away food assistance or other benefits from individuals who do not meet rigid work
requirements is well-documented in the research. The research also finds very little gain in longerterm employment as a result of such policies. In other words, the research supports the findings of
the RIA that many individuals would be cut off SNAP as a result of fewer areas qualifying for
waivers from the time limit under the proposed rule, but does not support the overall stated purpose
of the regulatory change.
The proposed rule ignores strong research evidence from independent researchers that contradicts
the stated justification for the proposed change. Below we discuss the available relevant research on
the impact of taking benefits away from individuals who are unable to meet rigid work requirements.
The NPRM Ignores Research That Finds That the Characteristics
of the Low-Wage Labor Market Contribute to Periods of Unemployment
The proposed rule implicitly assumes that taking away food assistance will cause people not
currently working to get jobs. But this assumption ignores research evidence about the realities of
the low-wage labor market that contribute to periods of unemployment and mischaracterizes the
work patterns of many people who need and receive assistance.
Features of the Labor Market Contribute to Periods of Unemployment
The basic characteristics of low-wage jobs are well-documented: low-paid jobs often don’t last;
low-wage industries that employ workers with limited education or work experience tend to expand
and shrink their workforces frequently based on demand, resulting in part-time jobs that have
unstable hours and high turnover; and low-paid workers often lack the health coverage, paid leave,
and reliable child care that can help a worker keep her job. These realities help explain why many
workers in low-wage jobs need assistance while they are working and when they are in between jobs.
The nature of low-wage jobs can make it hard for a worker to meet rigid work requirements.
Recent work by economists Kristen F. Butcher and Diane Whitmore Schanzenbach used the
Census Bureau’s Current Population Survey to show that the occupations of SNAP or Medicaid
recipients who work at least part of the year feature instability and low wages overall (not just for
SNAP or Medicaid recipients). These occupations include personal care and home health aides,
maids and housekeepers, dishwashers, food preparers, and laundry and dry cleaning workers.
Looking at all workers in the ten occupations most prevalent among SNAP recipients, the researchers
found that these workers faced more periods of joblessness and were less likely to be stably
170

employed from year to year than better-paid workers in other occupations. The researchers
conclude, “Together, these results suggest that it will be difficult for individuals who work and
participate in benefit programs to meet proposed work requirements in the private sector alone.
Although employment levels are high among many of these types of workers, employment volatility
is also quite high. Much of this volatility reflects characteristics of these types of occupations and is
not necessarily due to decisions made by the workers.”402
Another study of the jobs that are common among SNAP participants found that, “because
SNAP participants work in many industries (such as retail and hospitality) and occupations (such as
service and sales) where features such as involuntary part-time work and irregular scheduling are
common, they may participate in SNAP to supplement their low incomes due to insufficient or
fluctuating hours. Similarly, because workers often cycle in and out of these jobs, workers may
participate in SNAP during periods of unemployment or underemployment.”403
In addition, there is evidence that low-wage jobs have higher turnover and are far less likely to
have access to paid sick leave or paid family leave.
• According to a

2018 study by the Economic Policy Institute, “the monthly rate of churn into
and out of employment for low-wage workers is roughly twice as high as it is for the typical
worker in the middle of the wage distribution.”404

• Data

from the Bureau of Labor Statistics shows that low-wage workers are far less likely to
have access to paid sick leave or paid family leave.405

The Rationale of the NPRM Ignores Research on The Work Patterns of People Who Are Low-Skilled, Low-Wage
Workers Subject to The Time Limit
The NPRM states that “The application of waivers on a more limited basis would encourage more
ABAWDs to take steps towards self-sufficiency.”406 Secretary of Agriculture Sonny Perdue on
February 28, 2019 in testimony before the Senate Agriculture Committee defended the proposed

402

Kristen F. Butcher and Diane Whitmore Schanzenbach, “Most Workers in Low-Wage Labor Market Work
Substantial Hours, in Volatile Jobs,” Center on Budget and Policy Priorities, July 24, 2018,
https://www.cbpp.org/research/poverty-and-inequality/most-workers-in-low-wage-labor-market-work-substantialhours-in.
403

Brynne Keith-Jennings and Vincent Palacios, “SNAP Helps Millions of Low-Wage Workers,” Center on Budget and
Policy Priorities, May 10, 2017, https://www.cbpp.org/research/food-assistance/snap-helps-millions-of-low-wageworkers.
404

David Cooper, Lawrence Mishel, and Ben Zipperer, Economic Policy Institute, April 2018,
https://www.epi.org/publication/bold-increases-in-the-minimum-wage-should-be-evaluated-for-the-benefits-of-raisinglow-wage-workers-total-earnings-critics-who-cite-claims-of-job-loss-are-using-a-distorted-frame/.
405

Bureau of Labor Statistics, “National Compensation Survey: Employee Benefits in the United States, March 2017,” Bulletin
2787, September 2017.
406

NPRM, p. 981.

171

rule, saying that “We think the purpose is to help people move to independency…. We should help
people when they are down but that should not be interminably.”407
The belief that unemployed adults who participate in SNAP are dependent on SNAP for long
periods ignores research that finds that large numbers of recipients who are not working at a point
in time have recently worked or will work soon. A CBPP analysis of SNAP recipients shows that in
a typical month in mid-2012, some 52 percent of adult recipients not receiving disability benefits
were working, but that 74 percent worked in the year before or after that month. 408
The analysis examined adults who weren’t receiving disability benefits and who participated in
SNAP for at least a month in a period of almost 3.5 years. This allowed us to observe their work
both while they participated in SNAP and in the months when they did not, and to observe
employment among SNAP recipients over a longer period.
The adults in the analysis worked the majority of the months in the analysis, but they were more
likely to participate in SNAP in the months when they were out of work and their income was
lowest. They participated in SNAP in about 44 percent of the months that they were working and in
62 percent of the months in which they were not working. This helps explain why an analysis that
only looks at work in a single point-in-time month while people are receiving SNAP will show them
working less than they do over time: many of them are workers who temporarily receive SNAP
when they are between jobs.409
While these figures apply to all adults, not just those without children in the household, the
research finding about the difference between point-in-time employment and employment over
several years is still relevant and the NPRM does not address it. While individuals subject to the time
limit may, in any given month, not have sufficient work hours to pass a rigid work test, many will be
working (or working more hours) within a short time, with or without a work requirement.
Moreover, when SNAP participants are working, it’s often unstable work with low wages that does
not lead to self-sufficiency, contrary to the framing included in the NPRM.
Research From the TANF Program Found That Employment Impacts
Are Modest and Fade Over Time
The rigorous random-assignment evaluations of programs that imposed work requirements on
cash assistance (AFDC/TANF) recipients in the late 1990s contradicts the stated rationale for the
proposed rule, but supports the finding from the RIA that under the proposed rule one would not
expect to see increased employment or earnings. While these evaluations generally found modest,
407

“Perdue Reiterates Need to Restore Original Intent of SNAP: A Second Chance, Not a Way of Life, Food and
Nutrition Service, USDA, Press Release USDA 0025.19, February 28, 2019,
https://www.fns.usda.gov/pressrelease/2019/usda-002519.
408

For similar households with just childless adults, 46 percent were working in a typical month and 72 percent worked
within a year before or after that month. https://www.cbpp.org/unemployed-adults-without-children-who-need-helpbuying-food-only-get-snap-for-three-months
409

Brynne Keith-Jennings and Raheem Chaudhry, “Most Working-Age SNAP Participants Work, But Often in Unstable
Jobs,” Center on Budget and Policy Priorities, March 15, 2018.

172

statistically significant increases in employment early on, the effects faded over time, and people with
significant barriers to employment were not helped. In fact, many were hurt.
• In Portland,

Oregon, the site of the largest earnings impact among the evaluations, the share
of recipients with stable employment (defined as being employed in 75 percent of the calendar
quarters in years three through five after the pilot project began) rose only from 31.2 to 38.6
percent.410

• Within

five years, employment among people subject to and not subject to work requirements
was about the same in nearly all the programs evaluated.411

• Even when the

programs provided specially tailored services, the vast majority of participants
facing significant employment barriers did not find employment as a result of work
requirements.412

• The

California GAIN program, the so-called “Riverside Miracle,” which focused on getting
recipients into any job as quickly as possible, was outperformed in the long run by programs
that focused on increasing participants’ skills and building their human capital.413

One TANF expert researcher commented on the House Agriculture Committee’s work
requirement proposals from the 2014 farm bill, which would have expanded the existing approach
for SNAP for childless adults to adults with children, that, “[t]here is no credible evidence to suggest
that the specific work requirements developed by the House Agriculture Committee would ‘work.’
In fact, they are not likely to do much in the way of promoting employment and could push millions
of families/individuals deeper into poverty.”414
On balance, as we discuss in more detail in Chapter 6, this rigorous research supports the findings
of the RIA that many people lose benefits when required to comply with work requirements, but
contradicts the stated purpose of the proposed rule to increase self-sufficiency.
Strong Evidence That Many Subject to the Time Limit
Face Employment Barriers and Would Lose Needed Help
Based on the TANF experience from the 1990s, as well as the existing experience with the time
limit in SNAP and the early experience from Arkansas (the only state so far to terminate Medicaid
410

Gayle Hamilton et al., “National Evaluation of Welfare-to-Work Strategies: How Effective Are Different Welfare-toWork Approaches? Five-Year Adult and Child Impacts for Eleven Programs,” Manpower Demonstration Research
Corporation, December 2001, Appendix Table C-6.
411

Ibid., Table C.1.

412

Dan Bloom, Cynthia Miller, and Gilda Azurdia, “Results from the Personal Roads to Individual Development and
Employment (PRIDE) Program in New York City,” MDRC, July 2007.
413

V. Joseph Hotz, Guido Imbens, and Jacob Klerman, “Evaluating the Differential Effects of Alternative Welfare-toWork Training Components: A Reanalysis of the California GAIN Program,” Journal of Labor Economics, Volume 24
Number 3, 2006, pp. 521-566.
414

Peter Germanis, “Who Killed Work Requirements for SNAP in the Farm Bill? Answer: Conservative Ideologues,”
January 1, 2019, https://mlwiseman.com/wp-content/uploads/2019/01/Farmbill.120118.pdf.

173

for individuals who fail to document that they are meeting Medicaid work requirements), many
people subject to work requirements would lose benefits, and poverty and hardship would increase.
This, again, is consistent with the analysis in the RIA, but not with the justification for the proposed
rule.
• Research shows

that many of the people who would be newly subject to the time limit have
circumstances that may limit the amount or kind of work that they can do. A large share face
physical or mental health conditions or a cognitive impairment that would be difficult for state
agencies to identify or for individuals to obtain paperwork to prove. 415

• Research shows

that many TANF recipients who lost financial assistance due to work
requirements had serious barriers to employment. They were likelier than other recipients to
have physical or mental health issues, have substance use disorders, be victims of domestic
violence, have low education and skill levels, have prior criminal justice records, or lack
affordable child care.416

• The

rigorous experiments from the 1990s that required cash assistance recipients to
participate in work-related activities found that the resulting loss in benefits raised “deep
poverty” rates (the share of households with income below half the poverty line). 417 Similar
results were found with careful non-experimental analyses of leaver studies and household
survey data. Moreover, studies of TANF recipients whose assistance was taken away found
that they were likelier to experience serious hardship, such as seeing their utilities shut off,
becoming homeless, or lacking adequate food.418 In line with these findings, numerous
scholars using a variety of data and methods have concluded that cash assistance has
weakened as a guard against deep poverty under TANF,419 and that some families are worse
off as a result.420

415

Rachel Garfield, Robin Rudowitz, and Anthony Damico, “Understanding the Intersection of Medicaid and Work,”
Kaiser Family Foundation, January 5, 2018; MaryBeth Musumeci, Julia Foutz, and Rachel Garfield, “How Might
Medicaid Adults with Disabilities Be Affected By Work Requirements in Section 1115 Waiver Programs?” Kaiser Family
Foundation, January 26, 2018; Bauer, Schanzenbach, and Shambaugh; Keith-Jennings and Chaudhry.
416

LaDonna Pavetti, Michelle K. Derr, and Heather Hesketh, “Review of Sanction Policies and Research Studies: Final
Literature Review,” Mathematica Policy Research, March 10, 2003.
417

Stephen Freedman et al., “National Evaluation of Welfare-to-Work Strategies — Evaluating Alternative Welfare-toWork Approaches: Two-Year Impacts for Eleven Programs,” Manpower Demonstration Research Corporation, June
2000; Pavetti, Derr, and Hesketh; Marianne Bitler, Hilary Hoynes, and Jonah Gelbach, “What Mean Impacts Miss:
Distributional Effects of Welfare Reform Experiments,” American Economic Review, Volume 96, Number 4, 2006, pp. 9881012.
418

Ariel Kalil, Kristin Seefeldt, and Hui-chen Wang, “Sanctions and Material Hardship under TANF,” Social Service
Review, December 2002, pp. 642-662; Melissa Ford Shah et al., “Predicting Homelessness among Low-Income Parents on
TANF,” Washington State Department of Social and Health Services, August 2015,
https://www.dshs.wa.gov/sites/default/files/SESA/rda/documents/research-11-224.pdf; Andrew Cherlin et al.,
“Sanctions and Case Closings for Noncompliance: Who Is Affected and Why,” Johns Hopkins University, Policy Brief
01-1, 2001.
419

Gene Falk, “Temporary Assistance for Needy Families (TANF): Size of the Population Eligible for and Receiving
Cash Assistance,” CRS Report No. R44724, Congressional Research Service, Washington, D.C., January 3, 2017.
420Ron

Haskins, “Welfare Reform at 20: Work Still Works,” Journal of Policy and Management 35(1), 2016, 223–224; Robert
A. Moffitt, “The Deserving Poor, the Family, and the U.S. Welfare System,” Demography 52(3), 2015, 729–749; Hilary W.

174

• Research has shown that

African American TANF recipients are far likelier to have their
benefits taken away than white recipients.421 Caseworkers’ decisions about when to impose a
sanction involve some discretion; one study, using fictitious case examples, showed that
caseworkers were likelier to sanction African American recipients than white recipients.
Recipients of color may also be likelier to be sanctioned because they face greater challenges
in the labor market, including discrimination.

• Evidence

from SNAP and Medicaid shows that administrative hurdles can lead people to lose
assistance even when they are working or may qualify for an exemption, because they do not
understand or cannot comply with the requirement or because the state agency fails to
properly process the paperwork.422

• Recent

evidence from Arkansas’s implementation of work requirements for Medicaid is
sobering. Arkansas is taking Medicaid coverage away from certain adult beneficiaries who fail
to report at least 80 hours of work or work-related activities per month for three months.
More than 18,000 Arkansans have lost coverage after just seven months of implementation,
and thousands more are at risk over the coming months. Data from the state show that a very
small share of those required to report hours of participation (many beneficiaries are exempt
from the reporting requirement) have reported their hours, with very few successfully
navigating the exemption and “good cause” processes.423

Hoynes and Diane W. Schanzenbach, “Safety Net Investments in Children,” March 8, 2018, Brookings BPEA Article,
https://www.brookings.edu/bpea-articles/safety-net-investments-in-children/; James P. Ziliak, “Temporary Assistance
for Needy Families,” 2015 NBER Working Paper 21038; Kristin S. Seefeldt and Heather Sandstrom, “When There Is
No Welfare: The Income Packaging Strategies of Mothers Without Earnings or Cash Assistance Following an Economic
Downturn,” 2015. Russell Sage Foundation Journal of the Social Sciences 1(1), 139–158; H. Luke Shaefer, Kathryn Edin, and
Elizabeth Talbert, “Understanding the Dynamics of $2-a-Day Poverty in the United States.” (2015) Russell Sage
Foundation Journal of the Social Sciences 1(1), 120–138; Christina Paxson and Jane Waldfogel, “Welfare Reforms, Family
Resources, and Child Maltreatment,” Journal of Policy Analysis and Management 22(1), 2003, 85–113.
Sanford F. Schram et al., “Deciding to Discipline: Race, Choice, and Punishment on the Frontlines of Welfare
Reform,” American Sociological Review, January 2009; Kalil et al.; Richard C. Fording, Joe Soss, and Sanford F. Schram,
“Devolution, Discretion, and the Effect of Local Political Values on TANF Sanctioning,” Social Service Review, June 2007,
pp. 285-316; Chi-Fang Wu, Maria Cancian, and Daniel R. Meyers, “Sanction Policies and Outcomes in Wisconsin,”
Focus, Volume 23, Number 1, Winter 2004, https://www.irp.wisc.edu/publications/focus/pdfs/foc231f.pdf; Pavetti
2004.
421

Rachel Garfield, Robin Rudowitz, and MaryBeth Musumeci, “Implications of a Medicaid Work Requirement:
National Estimates of Potential Coverage Losses,” Kaiser Family Foundation, June 27, 2018; Nader S. Kabbani and
Parke E. Wilde, “Short Recertification Periods in the U.S. Food Stamp Program,” Journal of Human Resources, Volume 38,
2003; David Ribar, Marilyn Edelhoch, and Qiduan Liu, “Watching the Clocks: The Role of Food Stamp Recertification
and TANF Time Limits in Caseload Dynamics,” Journal of Human Resources, Volume 43, Number 1, 2008, pp. 208-238;
Mark Edwards et al., “The Great Recession and SNAP Caseloads: A Tale of Two States,” Journal of Poverty, Volume 20
Issue 3, December 11, 2015; Colin Gray, “Why Leave Benefits on the Table? Evidence from SNAP,” Upjohn Institute
Working Paper 18-288, May 21, 2018; Food and Nutrition Service, U.S. Department of Agriculture, “Understanding the
Rates, Causes, and Costs of Churning in the Supplemental Nutrition Assistance Program (SNAP),” November 2014.
422

Jennifer Wagner, “Commentary: As Predicted, Arkansas’ Medicaid Waiver Is Taking Coverage Away From Eligible
People,” Center on Budget and Policy Priorities, updated March 12, 2019, https://www.cbpp.org/health/commentaryas-predicted-arkansas-medicaid-waiver-is-taking-coverage-away-from-eligible-people
423

175

The RIA Cites Only One Study, Which Does Not Support the Proposed Rule
As mentioned, the NPRM cites no research to support that the proposed rule would achieve its
purported goal: i.e., that taking food assistance away from certain low-income childless adults would
encourage more self-sufficiency and employment. The one study referenced in the entire RIA
document instead examines the relationship between the duration of unemployment and future
employment and earnings.424 The study finds that long-term unemployment has a negative effect on
the likelihood of future employment and that the fact that someone experiences long-term
unemployment is the main reason (as opposed to inherent characteristics of individuals who
experience long-term unemployment.) Strangely, the study offers little support for the NPRM and
raises important cautionary notes suggesting that the proposed rule would worsen, not improve,
outcomes for the targeted population.
• The

one study cited in the RIA does not support the proposed rule. The RIA suggests
that because longer unemployment spells are associated with a lower likelihood of future
employment, the proposed rule is justified. But the study does not mention SNAP and does
not address whether taking food assistance away from low-income individuals would either
decrease unemployment spells or directly increase the likelihood of future employment. In
fact, it is difficult to understand what connection could be made. The RIA estimates that
755,000 individuals would lose SNAP under the proposed rule, but provides no estimate for
increased employment. Much of the research in this area shows that those individuals will face
increased hardship and may have a more difficult time finding work.

• The

population studied is not the population subject to the SNAP policy. The study
cited looked at long-term bouts of unemployment by looking at a sample of all workers in
state unemployment insurance systems. But the childless adult population subject to the
SNAP time limit is a distinct group that includes many individuals not included in the study
group because many adults who participate in SNAP do not receive unemployment
compensation, even if they are working or had worked. A study of ABAWDs subject to the
time limit in Ohio found that nearly 80 percent had never been eligible for unemployment
benefits.425 More importantly, as discussed in detail elsewhere in these comments, other
research shows that most childless adults who receive SNAP work when they can find
employment. Based on Census Bureau SIPP data, about 75 percent of SNAP households with
a childless, working-age adults worked in the year before or after receiving SNAP. Many of
these individuals would not be in the pool of adults considered long-term unemployed in the
study, so the conclusions drawn in the cited study do not directly apply to ABAWDs as a
group and do not justify a policy change directed at them.

• Finally, the

study’s findings suggest that support for individuals to improve their skills
or participate in work programs would be a better approach. The study finds that “the
longer-term unemployed experience substantially worse employment and earnings losses than
the short term unemployed.” The methodology, “allows us to rule out the ‘bad apple’

Katharine G. Abraham et al., “The Consequences of Long-term Unemployment: Evidence from Linked Survey and
Administrative Data,” National Bureau of Economic Research, Working Paper 22665, September 2016,
http://www.nber.org/papers/w22665. The citation appears on p. 3 of the RIA.
424

Ohio Association of Food Banks, “Franklin County Comprehensive Report on Able-Bodied Adults Without
Dependents, 2014-2015,” October 14, 2015, p. 15,
http://admin.ohiofoodbanks.org/uploads/news/ABAWD_Report_2014-2015-v3.pdf.
425

176

explanation for why the long-term unemployed fare worse…and [is] consistent with duration
dependence as the explanation for their poorer outcomes.” This means that it is not the
characteristics of the individuals that cause them to be long-term unemployed that are behind
the results, but rather the fact of their long-term unemployment. So, if FNS were serious
about wanting to help improve the longer-term outcomes for individuals who experience
long-term unemployment, it would focus on helping to improve their education and skills or
providing slots in work experience programs that allow them to demonstrate their desire to
work, rather than cut their food assistance.

C. Reports That Purport to Find Positive Effects From the Time Limit Are
Deeply Flawed
The only studies that claim to find substantial positive impacts when low-income individuals are
faced with losing food assistance or other benefits if they do not meet rigid work requirements are
deeply flawed. FNS does not cite this research either, but we include here some discussion of why
FNS should not rely on these kinds of assertions in any future policy development. The faulty
results come from the researchers making causal claims without a random-assignment design (or
other analytically sound comparison-group methods), ignoring program participants’ work
experience prior to receiving assistance, and excluding the impact on households of losing
benefits.426 Below we explain how two such reports, citing data from Kansas and Maine, have
inaccurately touted the alleged success of reimposing a three-month time limit on SNAP
participation for childless adults.427
When the recession decimated the labor market and unemployment spiked, most states,
including Kansas and Maine, requested the time limit be waived statewide. Kansas reimposed the
time limit statewide beginning in October 2013 and Maine reinstated the time limit statewide in
October 2014, even though both states qualified for a statewide waiver at the time the time limit
returned.
In both states total SNAP caseloads already were declining, but they dropped significantly four
months after the time limit was put in place, as Figure 11.1 shows. Data from Kansas and Maine that
are limited to the childless adults who were potentially subject to the time limit show that SNAP
participation fell among that group by 70 to 80 percent after the time limit returned.

426

Dottie Rosenbaum and Ed Bolen, “SNAP Reports Present Misleading Findings on Impact of Three-Month Time
Limit,” Center on Budget and Policy Priorities, December 14, 2016; Tazra Mitchell, LaDonna Pavetti, and Yixuan
Huang, “Study Praising Kansas’ Harsh TANF Work Penalties Is Fundamentally Flawed,” Center on Budget and Policy
Priorities, updated February 20, 2018.
See Jonathan Ingram and Nic Horton, “The Power of Work, How Kansas’ Welfare Reform is Lifting Americans Out
of Poverty,” The Foundation for Government Accountability, February 16, 2016, https://thefga.org/wpcontent/uploads/2016/02/Kansas-study-paper.pdf; Maine Office of Policy and Management, “Preliminary analysis of
work requirement policy on the wage and employment experiences of ABAWDs in Maine,” April 19, 2016,
http://www.maine.gov/economist/opm/pub/ABAWD_analysis_final.pdf; and accompanying Maine Department of
Health and Human Services May 11, 2016 press release and other related materials.
427

177

FIGURE 11.1

178

The reports assert that, as a result of the SNAP time limit, work rates and wages have increased
dramatically and the individuals subject to the time limit are better off. The reports, however,
misrepresent or omit data and, as a result, make claims about the impact of the time limit on work
and earnings that the facts do not support. 428 The analyses also rest on faulty assumptions about why
some childless adults receiving SNAP are not working.
The reports’ three largest problems are:
• They do

not take into account that many SNAP recipients already work, or would
work soon even without the time limit. The studies attribute rising work rates and earnings
to the return of the time limit even though most, if not all, of the changes would have
happened without it. The authors fail to acknowledge that many SNAP recipients who are
subject to the time limit were working already, or would soon be working, and as a result, their
work rates and wages would likely have risen without the time limit. They make claims that
can only be identified through a rigorous evaluation that isolates the impacts of the time limit
from what would have happened without it (see box below).

• They do

not consider the potentially severe impact of the time limit on those cut off
SNAP. The studies fail to discuss the circumstances of the individuals who are subject to the
time limit and the consequences for increased hardship and food insecurity when they lose
SNAP benefits. Without addressing this side of the equation, the studies misrepresent the
effect of reinstating the time limit on the well-being of those cut off SNAP. Their figures on
the average income of those cut off SNAP are highly misleading because they do not include
the loss of SNAP benefits. They do not discuss or attempt to assess what happens to
individuals who lose their food assistance and are unable to find employment, who are a large
share of those cut off.

• They do

not adequately consider the likely explanations for why childless SNAP
participants may not work. The authors advance the theory that individuals are avoiding
work and remaining in poverty in order to qualify for modest SNAP benefits of only about $5 a
day. But research and experience in states with the time limit in effect offer evidence of
alternative explanations. Many such individuals do work when they can, but they often face
significant barriers to work, such as low education and skills or physical or mental health
issues.

A careful look at the data presented in the reports, taking these factors into account, strongly
suggests that not much changed related to work and earnings when the time limit took effect, but
the time limit did cause thousands of the states’ poorest residents to lose essential SNAP benefits.

428

Peter Germanis has also written extensively on the methodological shortcomings of these types of studies. Peter
Germanis, “How Do the Foundation for Government Accountability’s Evaluations of Welfare Reform Measure Up? A
Report Card (Hint: The FGA Fails),” June 24, 2018, https://mlwiseman.com/wpcontent/uploads/2016/05/Evaluating-Welfare-Reform.pdf.

179

Conventional Evidence-Based Research Uses a “Comparison Group”
One of the central tenets of sound, evidence-based research is the need to have a “comparison group” so
that the results can properly account for what would have happened in the absence of a change.
For example, consider researchers who are testing the efficacy of a new medicine designed to speed
recovery from the common cold. The researchers would need to know how fast people would have gotten
better without the medicine. Without a comparison group there would be no way to know what to make of
results that showed, for example, that 30 percent were better after two days and 85 percent were better
after five days. Many, perhaps all, of these people would have gotten better without the medicine.
The gold standard for comparison groups is “random assignment,” an experimental approach where people
who are otherwise the same are randomly assigned to different “treatment” groups and the effects of the
change are measured on each group so the study can isolate the effect of the “treatment.” These types of
studies are expensive, though some are underway in SNAP, funded by the 2014 farm bill.
In the case of low-income childless adults, two important factors are critical for interpreting the information
in the Kansas and Maine reports. First, many low-skill, low-wage workers do work, but they work in highturnover jobs with low job security and often experience sporadic employment. SNAP acts as a safety net,
providing assistance during periods of unemployment or when work hours are cut. It is common for SNAP
recipients to have higher employment and wages in the future. Second, both Kansas’ and Maine’s
economies were improving between 2013 and 2015, the period in which the two states implemented the
time limit and purport to measure the results. Without controlling for these factors, it is difficult to isolate the
effects of the time limit on employment.
The authors of the reports for Kansas and Maine could have established less complicated and less costly
alternative “comparison groups” by conducting the same analysis in the year before the cutoff to observe
the work rates and earnings for similar SNAP recipients during a period when the time limit was not in effect.
Such an approach would not have been perfect ⎯ it would be impossible to take the differences in the labor
market and all other factors into account ⎯ but it would have been a more informative comparison than
these reports provide.

Reports Don’t Acknowledge That Many SNAP Recipients
Subject to the Time Limit Already Work
The studies from Kansas and Maine assert that reimposition of the time limit resulted in higher
work rates and earnings for individuals who lost SNAP benefits after exhausting three months of
eligibility.
• For

Kansas, the authors claim, “These reforms immediately freed nearly 13,000 Kansans from
welfare on December 31, 2013. Nearly 60 percent of those leaving food stamps found
employment within 12 months and their incomes rose by an average of 127 percent per
year.”429

• For

Maine, the Department of Health and Human Services’ press release reported that among
the individuals whose SNAP was cut off, “Incomes rose 114 percent within a year of leaving
the program,” and “nearly half (48%) worked at least one quarter in 2015.”

429

Ingram and Horton, op. cit., p. 2.

180

These reports, however, dramatically overstate the increase in work rates and wages that resulted
from the reimposition of the time limit because many of the SNAP recipients affected were
working, or would have started working anyway, albeit mostly in low-wage jobs with high turnover.
Moreover, both states reimposed the time limit when their unemployment rates were dropping and
unemployed individuals were somewhat more likely to be able to find work or higher wages as a
result.430
Kansas Report Misrepresents Several Key Indicators
The authors of the Kansas report overstate the degree to which work rates increased after the
time limit and misrepresent the effect of the time limit on numerous outcomes for SNAP recipients
while they are receiving SNAP.
Work Rates Were Essentially Unchanged Before and After the Time Limit Returned
The Kansas authors claim that “nearly 60 percent of those leaving [SNAP after the three-month
time limit went into effect] found employment within 12 months.” (The authors’ estimate of 60
percent is the share who had ever worked in any quarter of 2014.) The claim implies that the policy
change reimposing the time limit was the reason that these people found work, but it’s misleading,
as explained below.
However, work rates before and after the time limit were very similar, as Figure 11.2 shows.
Almost 40 percent of those whose SNAP was cut off already worked in each of the last two quarters
before the time limit returned (the third and fourth quarters of 2013).431 The share working each
quarter in the year after the time limit was implemented rose slightly, to just over 40 percent. This
modest increase could be explained by two factors: (1) low-wage workers are more likely to apply for
and participate in SNAP when they lose a job or their incomes drop, so they often experience
improvements in the future as their employment situation improves; and (2) Kansas’ economy was
improving during 2014, so a slightly larger share of recipients may have been able to find jobs or
higher pay. The time limit does not appear to have dramatically affected work rates for the group
subject to it.

430

In Kansas, where the time limit went into effect in October 2013, the overall unemployment rate fell from 5.3
percent in 2013 to 4.6 percent in 2014 and 4.2 percent in 2015. In Maine, where the time limit was reimposed one year
later, the unemployment rate fell from 5.6 percent in calendar year 2014 to 4.4 percent in calendar year 2015. For SNAP
recipients, especially those with the lowest education and skills, employment opportunities are highly sensitive to
economic conditions and the availability of jobs. See Hilary Hoynes, Douglas Miller, and Jessamyn Schaller, “Who
Suffers During Recessions?” NBER Working Paper No. 17591, March 2012, http://www.nber.org/papers/w17951.pdf.
431

CBPP calculates that almost 40 percent of the people who left the program on December 31, 2013 worked in the two
quarters just before the cutoff based on data on average wages per person from table 8 of the Kansas report. (The same
calculation is used to estimate the shares working in the later quarters as well.) Instead of reporting these accurate data,
the authors misleadingly report lower rates from national data sources as though they applied to the Kansas group. See p.
6: “Currently few able-bodied adults receiving food stamps actually work. … In 2013 just one-quarter of childless adult
households receiving food stamps had any earned income. ... An analysis of food stamp recipients conducted when work
requirements first went into effect found that fewer than five percent of all able-bodied childless adults on the program
were meeting those requirements.” This latter 5 percent is extremely misleading because it excludes a large number of
individuals who were working more than 30 hours a week as “exempt” from the time limit. If the authors included the
share of non-disabled childless adults who were working, the figure would be larger.

181

FIGURE 11.2

The authors, however, reached the opposite conclusion — that work rates grew significantly after
the time limit returned. Instead of comparing the average work rates in each quarter for this
population before and after the policy change, they report the share of individuals whose SNAP was
ended after December 2013 who ever worked in a quarter over the following year. This captures
typical movement in and out of the labor force — given that this group tends to work in highturnover jobs, in any quarter some people lose jobs and some get new jobs, so the share that ever
worked increases — rather than an isolated impact of the policy change. The trends in the share
who ever worked likely followed a very similar pattern in earlier years when the time limit was not in
effect (though the authors do not present such data). As discussed below, other research about labor
force participation among childless adults who receive SNAP finds work rates over time similar to
those in the Kansas report.

182

Improvements for SNAP Recipients Reflect
SNAP Caseload Changes,
Not Improved Circumstances

FIGURE 11.3

The Kansas report presents highly misleading
information about other changes among
individuals subject to the time limit who receive
SNAP. For example:
[S]ince restoring work requirements, the
employment rate among able-bodied
adults on food stamps has doubled. As a
result their incomes have more than
doubled on average, they are spending
less time on welfare, and the need for
assistance has significantly declined.432
These claims, which the report makes across
a range of measures, are misleading because the
childless adults who remained as SNAP
participants after the time limit went into effect
were significantly different from those who
participated before because of the policy change.
The state cut off SNAP those participants who
were not working at least 20 hours a week, so
the work rates, average earnings, and other
characteristics of those who remained SNAP
participants after the return of the time limit
were better, not because those individuals became
better off, but because they were better off to
begin with and were the only ones still eligible
for and participating in SNAP.
Those who may still participate in SNAP are
more likely to have earnings and, as a result,
lower SNAP benefits and appear better off on a
range of other characteristics. In fact, the
number of childless adult SNAP recipients
working at least 20 hours a week, and thus the
only non-exempt childless adult SNAP
recipients eligible for the program, dropped

432

Ingram and Horton, op. cit., p. 8.

183

modestly in the year after the time limit took effect. 433
As an example of this misleading representation, consider the authors’ assertion that, “[p]rior to
restoring work requirements, just 21 percent of childless adults on food stamps were working at all.
Two-fifths were working less than 20 hours per week. But since work requirements have gone back
into effect, that employment rate has risen to nearly 43 percent.” 434 The change was driven by a drop
in the number of SNAP recipients who are childless adults subject to the time limit, not an increase
in the number of recipients who are working. The number of such SNAP recipients who were
working fell by more than 40 percent (from 6,300 to 3,600), as those who were working less than 20
hours a week were cut off, while the total number of non-disabled childless adults receiving SNAP
dropped by more than 70 percent (from almost 30,000 to 8,500). (See Figure 11.3.)
Maine Also Inappropriately Attributes Changes to the Policy Change
That Likely Would Have Occurred Anyway
Maine’s data on work rates and wages among individuals who lost SNAP are similar in magnitude
to Kansas, and, as in the Kansas report, the authors of the Maine report and the accompanying
materials from the state’s Department of Health and Human Services overstate the impact of the
time limit by failing to take into account the fact that changes would have occurred even without it.
As Figure 11.4 shows, before the reimposition of the time limit in October 2014, about 30 percent
of the childless adults whose SNAP was cut off were working. That proportion peaked in the
months after the time limit went back into effect at 36 percent in the third quarter (the summer, a
time when employment in Maine tends to be higher). But, though the report includes these quarterly
rates, like for Kansas, the Maine report and accompanying materials emphasize a different figure:
that 48 percent had wages some time in 2015 and 58 percent had wages at some time ever in 2014 or
2015. But again, like for Kansas, the higher numbers count any time anyone had worked in any
quarter, and thus largely reflect employment instability at a time that the state’s economy was
improving, rather than a change that could be attributed to the reimposition of the time limit.

433

The drop in the number of childless adults who worked at least 20 hours a week and received SNAP could have
occurred because those individuals who qualified for SNAP (because they were working at least 20 hours a week) had
recently been cut off SNAP (at a time when they were not working at least 20 hours a week) and did not know they
would be eligible if they reapplied.
434

Ingram and Horton, op. cit., p. 9.

184

FIGURE 11.4

No Consideration of the Well-Being of Those Cut Off or the Support SNAP Provides
The one-sided pictures in these reports fail to discuss the well-being of the individuals whose
SNAP benefits were cut off. But research suggests that many childless adults who lose SNAP as a
result of the time limit continue to struggle after losing SNAP, in contrast to the reports’ portrayals
of circumstances for recipients who lost benefits. The most comprehensive assessment of former
SNAP recipients in four states in the early 2000s suggests that their life circumstances are quite
difficult. A significant minority don’t find work, and among those who are employed after leaving
SNAP, earnings are low. Most remain poor. Many struggle to acquire enough food to meet their
needs, lack health insurance, experience housing problems, and/or have trouble paying their bills.435
In a serious omission the Kansas and Maine reports do not consider the impact of the time limit
on the large number of people who lost SNAP and are among the nation’s very poorest adults.
• In Kansas the

number of non-disabled childless adults receiving SNAP dropped by 75 percent
(from about 30,000 in late 2013 to about 7,500 in late 2015).

435

Elizabeth M. Dagata, “Assessing the Self-Sufficiency of Food Stamp Leavers,” Economic Research Service, USDA,
September 2002, https://www.ers.usda.gov/publications/pub-details/?pubid=46645, a summary of in-depth studies in
Arizona, Illinois, Iowa, and South Carolina. These studies include people who leave SNAP because of the three-month
time limit or for other reasons, for example, because they found a job or mistakenly believe they are no longer eligible.

185

• The

Maine report does not present comparable numbers, but an earlier Heritage Foundation
report cited Maine Department of Health and Human Services data showing that the number
of “able-bodied adults without dependents on food stamps” dropped by 80 percent (from
about 13,300 in late 2014 to 2,700 in March 2015).436

The individuals whose SNAP was cut off lost about $5 a day, or $150 to $170 per person per
month in SNAP benefits for purchasing food. Many of them worked in the year after losing
benefits, but for some their wages were low enough that they could have continued to qualify for
SNAP benefits, which could have helped them make ends meet. Some others with no earnings for
some or all of the subsequent year may have had virtually no resources available for food after they
were cut off SNAP.
The Kansas and Maine reports cite average income figures for the year after recipients lost SNAP,
but they fail to account for the lost SNAP benefits. To accurately compare income for a household
that used to be on SNAP to income after losing SNAP, the lost value of the SNAP benefits must be
included. When they are accounted for, total income does not increase substantially (or actually
decreases slightly).
In Kansas, the authors rest their claim that SNAP recipients were better off after the cutoff on a
point that the group’s income rose by 127 percent between before the time limit took effect and one
year later. There are three problems with this claim:
• First,

as discussed above, it implies that the time limit was responsible for the earnings
increase, when most, if not all, of it likely would have occurred anyway;

• Second,

it excludes the value of SNAP benefits from the calculation. Total resources available
to the household were higher before the time limit because the household received SNAP
benefits; and,

• Third,

the authors picked a low comparison quarter prior to the time limit returning to
exaggerate the increase — wages for the group cut off were more than 30 percent lower two
quarters before the cutoff (third quarter of 2013, the quarter used in the report) than they
were in the quarter immediately preceding the cutoff (the fourth quarter of 2013). They do not
explain why they chose this particular quarter as the baseline.

It is not possible to adjust for the first issue without a rigorous evaluation (see above box), but
even adjusting only for the other two issues makes a large difference. If $178 a month in SNAP
benefits (the average SNAP benefit among those cut off in December 2013) is included in the base
period, and if we compare the quarter immediately before the cutoff (the fourth quarter of 2013) as
the base period instead of the quarter earlier, the total resources (including earnings and SNAP
benefits) available to SNAP participants who were cut off was 3 percent lower a year after the cutoff,
rather than 127 percent higher.

436

Robert Rector, Rachel Sheffield, and Kevin D. Dayaratna, “Maine Food Stamp Work Requirement Cuts Non-Parent
Caseload by 80 percent,” The Heritage Foundation, Backgrounder No. 3091, February 8, 2016,
http://www.heritage.org/research/reports/2016/02/maine-food-stamp-work-requirement-cuts-non-parent-caseloadby-80-percent.

186

Had the SNAP recipients who remained income-eligible been able to keep receiving SNAP (rather
than being cut off by the time limit) more of them would be better off because they could have
received SNAP while working (though their SNAP benefits would be lower because income counts
in determining SNAP benefit levels).
A similar contrast applies to the Maine report. The press release that accompanied the report
claims that total income for those cut off rose by 114 percent within a year. However, if the state
had included the value of SNAP benefits in the base, the increase would be much smaller ⎯ only
about 10 to 20 percent.437
No Consideration of Factors Affecting SNAP Participants’ Ability to Work
If a better picture of the data shows that the time limit doesn’t have a strong role getting people
into work, what do we know about why this group struggles to find employment? The research that
exists on this population shows that adults who participate in SNAP work when they can, but often
in jobs with high turnover and low job security, and most struggle with multiple barriers to
employment, as discussed above.
Adults on SNAP work when they can. However, the work tends to be low wage and unstable, with
individuals cycling through periods of work and unemployment. Nearly three-quarters of nondisabled adults who participate in SNAP in a typical month work either that month or within a year
of that month. Over half of individuals who were participating in SNAP in a typical month in mid2012 were working in that month. Furthermore, 74 percent worked in the year before or after that
month. 438 (See Figure 11.5.) Limited education, lack of training, and a sporadic work history make it
difficult to compete for anything other than low-skill, low-wage jobs that do not lift them out of
poverty.

437

The Maine report does not include information about the average SNAP benefits received by childless adults who
were cut off, so the 10 to 20 percent range reflects a lower ($150 a month) and higher ($180 a month) assumption. As
with Kansas, we cannot account for the large portion of the effect that would have happened anyway and we are not
including any SNAP benefits for the workers who would have income low enough to continue participating in SNAP
were there no time limit.
438

Brynne Keith-Jennings and Raheem Chaudhry, “Most Working-Age SNAP Participants Work, But Often in Unstable
Jobs,” Center on Budget and Policy Priorities, March 15, 2018. As mentioned above, for similar households with just
childless adults, 46 percent were working in a typical month and 72 percent worked within a year before or after that
month. https://www.cbpp.org/unemployed-adults-without-children-who-need-help-buying-food-only-get-snap-forthree-months

187

FIGURE 11.5

Childless adults on SNAP face barriers. Many low-income childless adults face multiple challenges to
independence and self-sufficiency, including homelessness, physical and mental health limitations,
language barriers, unstable employment histories, and criminal records. A detailed study of childless
adults who were referred to a work experience program in Franklin County (Columbus), Ohio
found that:439
• Many

have extremely unstable living situations, illustrated by residence in short-term shelters
or with friends and family and limited telephone service.

• One-third

have a mental or physical limitation, including depression, post-traumatic stress
disorder, mental or learning disabilities, or physical injuries. Some of these disabilities, though
not severe enough to qualify for federal disability benefits, may still limit a person’s ability to
work at least 20 hours a week.

• About

a quarter have less than a high school education, and more than half have only a high
school diploma or GED.

439

See “Comprehensive Report on Able-Bodied Adults Without Dependents, Franklin County Ohio Work Experience
Program,” Ohio Association of Foodbanks, 2015,
http://admin.ohiofoodbanks.org/uploads/news/ABAWD_Report_2014-2015-v3.pdf. The Ohio Association of
Foodbanks gathered the information for the report as a result of a partnership with the county SNAP agency to help
place individuals identified as subject to the time limit in qualifying work activities after screening them.

188

• Nearly

one-quarter are non-custodial parents, and 13 percent are caregivers for a parent,
relative, or friend.

• More

than 40 percent lack access to reliable private or public transportation; 60 percent lack a
valid driver’s license.

• Fifteen percent

need supportive services like language interpretation or help with
transportation to obtain employment.

• More

than one-third have felony convictions, making it hard to find jobs and pass background
checks.

D. FNS’ Research, Policies, and Practices Show FNS Knows but Ignored
That Impact of the Proposed Rule Is Out of Line With the Stated
Rationale
FNS’ Own Research from 1998 on the Employment Prospects
of “ABAWDs” Contradicts the Stated Justification of the NPRM
In 1998 FNS published a study called, “The Effect of Welfare Reform on Able-Bodied Food
Stamp Recipients” to provide information that “[I]s critical to informing policy decisions, issuing
guidance to states, implementing new policies, as well as estimating effects of the [new provisions.]”
It concluded in the forward to the study, “the report offers a sound picture of what able-bodied
adult recipients without children look like and what will happen to them―they are an extremely poor
population with limited employment prospects and few sources of support outside the Food Stamp
Program.”440 This research directly contradicts the stated justification for the proposed rule.
The study used SNAP QC household characteristics data and the Census Bureau’s Survey of
Income and Program Participants (SIPP) to describe the characteristics of individuals subject to the
time limit, including their limited educational attainment and workplace skills, their high poverty
rates, and their patterns of SNAP participation prior to the time limit going into effect. It also
estimated how many at that time had likely hit the time limit and been cut off SNAP.
The study also included information about the research available at that time on the employment
prospects of “ABAWD” SNAP participants, which is summarized as follows:
Research indicates that the employment prospects of adults who are demographically similar
to ABAWDs are not promising, and so we can assume the same to be true for ABAWDs.
Largely for two reasons, job opportunities for less-educated job seekers are severely limited,
especially for non-whites and in urban areas, where most ABAWDs live. First, recent research
suggests that many large employers of low-skill workers have moved out of the cities to the
suburbs. Therefore, many ABAWDs will face a “spatial mismatch” between the location of
their residence and the location of low-skill jobs. Second, since employment in inner cities has
440

Michael Stavrianos and Lucia Nixon, “The Effect of Welfare Reform on Able-Bodied Food Stamp Recipients,”
prepared by Mathematica Policy Research for the USDA, Food and Nutrition Service, July 23, 1998, https://fnsprod.azureedge.net/sites/default/files/finalrep.pdf.

189

become increasingly concentrated in high-skill jobs, ABAWDs will also likely face a “skills
mismatch” between what employers require and what ABAWDs can offer. 441
As we review in Chapter 3, while the nature of spatial mismatch has changed, more recent
literature has found that there still exists mismatch between low-wage jobs and where low-wage
workers live, particularly with regards to transportation access.
The 1998 FNS study also points out that low-skilled job seekers in many places may have
difficulty finding employment even when the national unemployment rate is low:
Implicit in PRWORA’s work requirement is the assumption that there are enough
employment opportunities for ABAWDs — that is, they can find work if they seek it....
However, a relatively large body of research indicates that the labor market situation of the
low-skilled has become considerably worse in recent decades and that their current
employment prospects are limited. This suggests that even if ABAWDs are willing to work,
they may be unable to do so because there are not enough jobs for low-skilled workers.442
Despite arguing repeatedly in the preamble and RIA that the low national unemployment rate
justifies the proposed rule,443 FNS does not refute this earlier research it published in 1998, nor
address whether the landscape has changed in the intervening 20 years to justify the change in
waiver policy.
USDA Research on Individuals Who May Have Been Cut off SNAP Because of the Time
Limit Does Not Support the Assertion That the Time Limit Improves Self-Sufficiency
After the three-month time limit was enacted in 1996, USDA’s Economic Research Service (ERS)
joined with the U.S. Department of Health and Human Services to fund studies in four states
(Arizona, Illinois, Iowa, and South Carolina) that examined the well-being of people who exited
SNAP in the late 1990s after the time limit went into effect.444 The studies included people who had
left SNAP because of the three-month time limit or for other reasons, for example, because the
found a job or mistakenly believe they no longer are eligible. Even though the studies were not able
to isolate the individuals who left SNAP because of the time limit, the picture they offer of the
441

Ibid, p. xii and pp. 51-69.

442

Ibid, p. 56-57.

443

NPRM, p. 981, 982, 983; RIA, p. 2, 10.

444

Elizabeth Dagata, “Assessing the Self-Sufficiency of Food Stamp Leavers,” Economic Research Service, USDA,
September 2002, https://www.ers.usda.gov/publications/pub-details/?pubid=46645, a summary of in-depth studies in
Arizona, Illinois, Iowa, and South Carolina.
https://lib.dr.iastate.edu/cgi/viewcontent.cgi?referer=https://www.google.com/&httpsredir=1&article=1013&context
=card_staffreports – Iowa
https://www.mathematica-mpr.com/our-publications-and-findings/publications/food-stamp-leavers-in-illinois-howare-they-doing-two-years-later - Illinois
https://naldc.nal.usda.gov/download/45220/PDF - South Carolina.

190

hardship such individuals face suggest that the time limit has not spurred many to self-sufficiency, or
even resulted in their life circumstances improving modestly, and contradict the stated rationale
behind the proposed rule.
• Many

were employed but had very low earnings. In the four states, employment rates
among the individuals who were unemployed childless adults potentially subject to the time
limit who left SNAP (or “leavers”) ranged from 41 percent in Illinois to 76 percent in Iowa,
“but earnings and incomes are low and their poverty rates are high.”445

• Most

remain poor. Despite relatively high levels of work effort, between one-third and twothirds of SNAP leavers in the four states had household incomes below the poverty line —
well above the overall poverty rate of 13 percent at the time. About 40 percent of the Illinois
and Iowa SNAP leavers had income below half the poverty line.

• Many

struggled to afford adequate food. “Between 17 and 34 percent of the [SNAP
leavers in the four states] reported food insecurity with hunger, compared with 11 percent of
U.S. low-income childless households.”446

• Many

had housing problems, or had trouble paying their utility bills. About 20 to 40
percent of SNAP leavers faced housing issues, including falling behind on the rent, moving in
with relatives, or becoming homeless. Between 20 and 65 percent reported problems paying
for utilities.

Finally, the studies raise an important question about exemptions from the time limit. “In two of
the studies, the majority of nonworking [SNAP leavers] cited health problems as the reason they
were not working...it is important to know whether the standards for being categorized as ‘able
bodied’ are set appropriately.”447
FNS’ SNAP Employment and Training Best Practices
and the Department’s SNAP to Skills Initiative Do Not Promote Cutting People Off SNAP
FNS’ position that taking food assistance away from people who do not meet rigid work
requirements will lead to stable employment and self-sufficiency also conflicts with the agency’s
approach to employment and training (E&T). For more than 30 years, since the mid-1980s, SNAP
has included an employment and training (E&T) component, “for the purpose of assisting members
of households participating in the supplemental nutrition assistance program in gaining skills,
training, work or experience that will…increase the ability of the household members to obtain
regular employment.”448 States operate the program within federal rules and FNS oversight, but have
substantial flexibility about which non-exempt SNAP recipients they serve in the program, the
services they offer participants, and whether the program primarily recruits volunteers who are
interested in help finding a job or improving their education and skills, or whether the program is
445

Dagata, p. 2.

446

Dagata, p. 4. Food insecurity “with hunger” was how USDA then referred to the most severe form of food insecurity
where households had to skip or reduce the size or their meals or otherwise disrupt their eating patterns at times during
the year because they couldn’t afford sufficient food.
447

Dagata, p. 5.

448

The Food and Nutrition Act of 2008, as amended, 7 U.S.C. § 2015 (d)(4)(A)(i).

191

mandatory, meaning individuals will lose SNAP benefits (i.e., be subject to sanction) if they do not
comply with the state’s assigned E&T activity.
E&T Best Practices Research Review Does Not Include Sanctions or Cutting People Off SNAP
While over the years Congress has amended and modified the focus of SNAP E&T somewhat, it
has always included a state option for programs that are mandatory and include sanctions for noncompliance. According to a 2016 FNS-sponsored literature review of SNAP E&T best practices, the
only time FNS conducted a rigorous study of the effectiveness of SNAP E&T, was in the late 1980s
and early 1990s. At the time, SNAP E&T was primarily “a high-volume, ‘light-touch’ program to
encourage mandatory work registrants to find jobs quickly, primarily by requiring participation in job
search and job-search training components,”449 that is, it consisted primarily of components that
included taking SNAP benefits away from individuals who did not meet the program’s requirements.
The 2016 literature review described the findings from the study that was published in 1994 as
follows:
[T]here was no evidence that the SNAP E&T program—in its high participation/low
investment per-participant model—increased the likelihood of participants finding jobs. It
also found the program had no significant effects on the hourly wages, hours worked per
week, or length of job retention for those who did find employment…
The observed effect of a statistically significant decrease in the level of food stamp benefits
receipt was described as most likely the result of ‘voluntary withdrawals and administrative
sanctions’ rather than of any increase in household income or earnings…
[A]lthough some members of the treatment group did find jobs, members of the control
group were also able to obtain similar job search assistance and find employment. 450
Thus, the only rigorous evaluation FNS has ever conducted of policies that take food assistance
away from individuals who are unable to meet work requirements found no improvement in
participants finding jobs or increases in wages or hours worked, but did find a significant decrease in
food assistance benefits. The literature review quoted here was published in 2016, so was certainly
available to FNS as it formulated the proposed rule, but FNS ignored its findings. FNS promulgated
the proposed rule that would take food assistance away from people who do not comply with rigid
work requirements without ever again testing or studying the effectiveness of such an approach in a
rigorous random assignment study. Moreover, the E&T pilots from the 2014 farm bill, which do
incorporate a rigorous random assignment evaluation, include three states with mandatory SNAP
E&T approaches, but FNS promulgated this regulation without waiting for the results of those
pilots.

449

Deborah Kogan et al., “Supplemental Nutrition Assistance Program (SNAP) Employment and Training (E&T) Best
Practices Study: Final Report,” prepared by Social Policy Research Associates for the U.S. Department of Agriculture,
Food and Nutrition Service, November 2016, https://www.fns.usda.gov/snap/snap-employment-and-training-et-bestpractices-study-final-report.
450

Ibid, p. III-1.

192

Overall, the 2016 SNAP E&T best practices report had as its objective providing “Congress,
FNS, and individual States with information that can be used to shape the services provided by the
SNAP E&T program and thereby improve the employability, self-sufficiency, and well-being of
individuals receiving nutrition support from SNAP.”451 These are the very same goals as the stated
intention and justification of the proposed rule in the NPRM, and yet the SNAP E&T best practices
report does not mention taking SNAP benefits away from individuals who are unable to meet work
requirements as an effective strategy. The report includes an annotated bibliography of 160 relevant
studies from the literature review. The best practices report recommendations are summarized in the
executive summary:
The findings from the research synthesized in this report suggest that SNAP recipients will benefit
most from SNAP E&T-funded services if the services offered by State programs
• Are

based on an individualized assessment of the workforce-related strengths and weaknesses
of SNAP clients;

• Comprehensively

address an individual’s need for skills training, basic skills education, and
overcoming barriers to employment;

• Help participant
• Develop skills

earn credentials valued by employers in their chose industry or sector; and

that are closely linked to labor market demands in the local area…

In view of these findings…States that enroll a relatively large number of mandatory work
registrants in SNAP E&T services or that emphasize self-reported job search as the most
frequently prescribed program activity are less likely to see an increase in self-sufficiency
among SNAP participants.452
Thus, the most recent research available to FNS about what works for SNAP E&T for meeting
the objectives of increasing employment, self-sufficiency, and well-being does not include sanctions
or cutting people off SNAP. But FNS ignored this evidence in promulgating the regulation. We urge
FNS to consider its own best practices study.
FNS’ Most Recent E&T Efforts Also Have Not Promoted Sanctions or Cutting People Off SNAP
Moreover, in the past several years FNS has placed a new emphasis on SNAP E&T. It has created
an Office of Employment and Training with several additional staff members, brought on
consultants and collaborated with other partners, formed a “learning academy,” produced new
training materials, and provided additional technical assistance to select states. The cornerstone of
the E&T efforts has been a “SNAP to Skills” initiative that follows closely the recommendations
from the SNAP E&T best practices report cited above.
Under a “Why SNAP to Skills” section of its website, FNS makes the following case for its SNAP
to Skills approach:453

451

Ibid, p. ES-1.

452

Ibid, p. ES-4.

453

U.S. Department of Agriculture, “Why SNAP to Skills,” https://snaptoskills.fns.usda.gov/why-snap-to-skills

193

The need of Supplemental Nutrition Assistance Program (SNAP) participants to secure
the education and training required to transition to economic self-sufficiency is growing
increasingly urgent. The vast majority of jobs in the future will require at least some
education beyond high school…yet many SNAP participants have not reached this level
of educational attainment. Without the skills to meet rapidly changing labor market
demand, the chances of SNAP participants getting a good job and reducing their need
for SNAP are extremely low…
The SNAP Employment & Training (SNAP E&T) program, a skills and job training
program for SNAP participants administered by the U.S. Department of Agriculture’s
Food and Nutrition Service (FNS), is a key resource States and their partners can utilize
to help SNAP participants meet this urgent need for skills and better jobs. SNAP E&T
has historically been under-utilized, but a renewed focus on the program amid greater
urgency for job training for SNAP participants has created new momentum for states
seeking to build bigger, better, and stronger E&T programs.
There is no more opportune, or critical, time for states to build robust, job-driven SNAP
E&T programs. “Job-driven” means that programs are responsive to employer demand so
that they place ready-to-work participants in good, available jobs or provide skills training and
credentials participants require to obtain these jobs. SNAP E&T is increasingly recognized as
a critical part of each state’s skilled workforce strategy. USDA and other policy makers herald
SNAP E&T as an important part of the national conversation about the need to invest in
building a skilled workforce while addressing the nation’s growing economic inequality.454
Like the best practices report cited earlier, FNS’ signature Employment and Training initiative
does not include sanctions or cutting individuals off of SNAP as an effective strategy for increasing
employment or earnings.

E. FNS Knows That Research and Experience Shows That People Newly
Subject to the Time Limit Won’t Get a Job or Be Better Off, But It
Promulgated the Proposed Rule Nonetheless
As we have laid out in this section, there is a large body of research evidence that finds that
policies that take benefits away from individuals who do not meet rigid work requirements result in
lost benefits and increased poverty and hardship, but very little gain in longer-term employment.
The realities of the low-wage labor market, including high turnover and lack of sick time and other
benefits contribute to individuals’ turning to SNAP during temporary periods of unemployment.
Many other individuals face various employment barriers.
FNS has for more than 20 years supported the waiver policy currently in regulation, and there is
no new research that contradicts or provides new information. The RIA conspicuously lacks any
countervailing research evidence to justify that the narrowing of waivers will improve individuals’
work rates or earnings, and the impact analysis included in the RIA assumes that 755,000 individuals
will lose SNAP benefits under the change, but there will be no quantifiable increases in earnings or
work.

454

Ibid.

194

We believe that the only conclusion one can draw is that contrary to the stated rationale in the
NPRM, FNS knows that the primary effect of the regulatory change, if it were finalized, would be
that a large number of individuals would lose SNAP assistance, with no, or very little positive
impact. We strongly urge FNS to review the research summarized here and included in the
Appendix B.

F. The Proposed Rule Uses an Imprecise Definition of “ABAWDs,” and the
RIA Includes Numerous other Unsubstantiated Assumptions
In addition to the fact that the proposed rule is not supported by available research, the analysis
that is included in the RIA includes numerous imprecise, illogical, and unsubstantiated assumptions,
starting with the use of the term “ABAWDs.” Since shortly after the passage of the 1996 welfare
law, FNS has described the group of SNAP participants whose eligibility is at issue because of the
three-month time limit as “ABAWDs,” or able-bodied adults without dependents. The proposed
rule uses this term throughout, but never defines it, and seems to include in it many individuals who
are exempt from the time limit, who live in an area that is under a waiver, or who are participating in
SNAP during periods when they are eligible (for example, in their first three months or when they
are working or participating in a qualifying employment and training program.) The methodology
for assessing the impact of the proposed rule indefensibly treats everyone who is an “ABAWD” by
this broad definition as though they would be subject to the time limit under the proposed rule (i.e.,
that they live in an area that would no longer qualify for a waiver.) The rule also excludes from the
analysis individuals who are “ABAWDs” subject to the time limit, but who are no longer SNAP
recipients because they have been cut off.
The imprecise use of the undefined term “ABAWD” is confusing and makes it difficult for
readers to understand and comment on the described impacts of the proposed rule. It also appears
that the Department in its estimates of the impact of the regulation has derived percentages for this
entire population of SNAP recipients potentially subject to the time limit and then applied those
percentages to individuals who would be newly subject to the time limit under the proposed rule,
resulting in a methodology that cannot be substantiated.
This section will first explain the analytical problem in the way the RIA defines and categorizes
“ABAWDs,” and then provide additional examples of specific assumptions in the methodology that
cannot be substantiated, often because of the imprecise use of the term “ABAWD.”
Use of the Undefined Term “ABAWD” Is Confusing and Misleading
Section 6(o) (7 U.S.C. § 2015(o)) of the Food and Nutrition Act limits SNAP eligibility for certain
non-exempt adults who are aged 18-49 to three months of SNAP benefits in any 36-month period if
they are not working 20 hours a week or participating in a qualifying employment and training
program, and if the area they live in is not waived from the rule because of insufficient jobs. 455
Because of the rule’s complexity, the individuals subject to the time limit cannot be identified in
SNAP or Census Bureau data because much of the information that would need to be known is not
available in the data. Analysts often have modeled SNAP eligibility for these individuals to the best
of their ability and then described the individuals as “potentially subject to the time limit,” because
455

Individuals who have been cut off SNAP can qualify for a second three months of eligibility if they work at least 80
hours (or participate in a qualifying program) for a month and reapply for SNAP.

195

only state eligibility workers have access to the information that is needed to make a full assessment.
In a 2016 paper we explained:
About 4.7 million non-elderly, non-disabled adults aged 18-49 in childless households
participated in SNAP in fiscal year 2014. All were “subject” to the time limit in the sense that
all could, in theory, have lost benefits after three months of participation. The number
affected by the time limit in practice, however, is much smaller. 456
We also explained in the 2016 paper the various reasons why the larger figure, which we represent
as individuals potentially subject to the time limit based on existing data, applies to “the
characteristics of the larger group of childless adults, all of whom would face the time limit if their
circumstances or local labor market conditions change.”457
In estimates that FNS published in the years immediately following passage of the 1996 welfare
law, it is clear that the agency understood both the limitations in the data (“The QC database does
not contain all the information needed to determine whether an individual loses eligibility under the
able-bodied provisions of PRWORA”458) and the distinction between those potentially subject to the
time limit and those who would actually be affected (“the QC-based estimates presented in this
chapter may overstate the number of people subject to the three-month time limit.”459)
But the NPRM, RIA, and Agriculture Department materials that accompanied the December
posting of the proposed rule portray this larger group that is potentially subject to the time limit as
the number who actually would be newly subject under the proposed rule. This mis-identification is
confusing and results in unsubstantiated assumptions. It identifies as newly subject to the time limit
many individuals who are exempt, complying with the time limit, or living in an area that is under a
waiver from the time limit, and it excludes individuals who would qualify for SNAP but have been
cut off after three months in areas that are not under a waiver from the time limit.
The Department asserts that “[i]n 2016 there were 3.8 million individual ABAWDs on the SNAP
rolls.”460 We are able to reproduce this number using the public use 2016 SNAP Household

456

In the 2016 paper CBPP used this methodology and included a box explaining the difference between the larger
number potentially subject to the time limit and the smaller number affected by the time limit in practice. See, Steven
Carlson, Dorothy Rosenbaum, Brynne Keith-Jennings, “Who are the Low-Income Childless Adults Facing the Loss of
SNAP in 2016,” February 8, 2016, https://www.cbpp.org/sites/default/files/atoms/files/2-8-16fa.pdf, p. 4.
457

Ibid.

458

Mike Stavrianos, et al., “Characteristics of Childless Unemployed Adult and Legal Immigrant Food Stamp
Participants: Fiscal Year 1995,” prepared by Mathematica Policy Research for the USDA, Food and Nutrition Service,
February 13, 1997, p. 6.
459

Ibid, Stavrianos, p. 6.

460

U.S. Department of Agriculture, “Regulatory Reform at a Glance, Proposed Rule: SNAP Requirements for
ABAWDs,” December 2018, https://fns-prod.azureedge.net/sites/default/files/snap/ABAWDSFactSheet.pdf. We
believe this 3.8 million is the starting point for all of the estimates in the RIA regarding the number of “ABAWDs,” but
the RIA does not make that clear, as we discuss later in our comments.

196

Characteristics Quality Control (QC) file,461 and below recreate how we believe FNS derived the
number. (See Table 11.1)
TABLE 11.1

CBPP Understanding of FNS’ Estimate That 3.8 million SNAP Recipients Were “ABAWDs”
in 2016
Number

Percent of Total SNAP
Participants

Total SNAP Participants

43.5 million

100%

Age 18-49

15.0 million

34.4%

Not receiving disability
benefitsa

12.8 million

29.5%

No minor children in
household

3.8 million

8.8%

FNS’ 3.8 million “ABAWD” estimate

Other factors that need to be taken into account for an
individual to be subject to the time limit that were not factored
into FNS’ estimate:

Is this information available in
the QC data?

Is the individual:
• in his or her first three months of SNAP participation out of
36?
• physically or mentally unfit for employment?
• living in a waived area?
• working 20 hours a week or more?
• participating in a qualifying E&T activity?
• pregnant?
• otherwise exempt from employment and training?
• exempted by an individual exemption?
• in a second three-month period after requalifying?

Variable not reliableb
Variable not reliablec
Variable not reliableb
But may be knowable if state
was under a statewide waiver
Can be estimated using
earnings or other variables
Variable not reliabled
Not available
Variable not reliablec
Variable not reliableb
Variable not reliableb

Sources: CBPP analysis of FY 2016 USDA SNAP Household Characteristics data and Mathematica Policy Research,
“Technical Documentation for the Fiscal year 2016 [or other year] Supplemental Nutrition Assistance Program Quality
Control Database and the QC Minimodel, October 2017.
The SNAP QC data set includes a personal-level disability variable (DISi). An algorithm is used to identify individuals
with disabilities based on SSI receipt, medical expenses, age, work registration status, and other factors. The technical
a

461

The 2016 public use Quality Control Data are available at https://host76.mathematica-mpr.com/fns/. All figures we
cite based on this data are for an average month in the fiscal year.

197

TABLE 11.1

CBPP Understanding of FNS’ Estimate That 3.8 million SNAP Recipients Were “ABAWDs”
in 2016
Number

Percent of Total SNAP
Participants

documentation flags that “DISi likely underestimates the number of non-elderly individuals with disabilities” and
therefore, the 3.8 million likely overestimates the number of adults without disabilities.
The SNAP QC data includes an individual-level variable called “ABWDSTi” that is intended to collect this information,
but the technical documentation “[recommends] caution when using…due to inconsistencies.”
b

c For

the variable intended to capture exemptions for disability and other factors (“WRKREGi”) the documentation states,
“we found continued evidence…of likely miscoding of this variable.”
The variable intended to capture participation in employment and training (“EMPRGi”) is also among the variables the
documentation “recommend[s] using with caution.”
d

As Table 11.1 shows, there are many eligibility factors that the Department’s analysis did not take
into account when estimating the number of people who are subject to the time limit. As a result,
the 3.8 million individuals the Department classifies as “ABAWDs” includes many individuals who,
in fact, are not subject to the time limit. The Department’s analysis also excludes many individuals
who would be SNAP recipients except they have been cut off because of the time limit, so they do
not appear in SNAP data because they are not SNAP recipients. 462 These are individuals who do not
live in waived areas but are subject to the time limit because they did not meet any of the other
criteria in Table 11.1.
The Department’s imprecise use of the term “ABAWD” results in a lack of transparency. It is
difficult to determine what point FNS (in the NPRM and RIA), and the Administration more
broadly, in its materials about the proposed rule, are making when they call individuals “ABAWDs”
when they are not, in fact, subject to the time limit, or when they are complying with the time limit.
They are implying that far more SNAP participants are subject to the time limit and not in
compliance with it than in fact is the case, and they are not counting people who have been cut off
because of the time limit.
In addition, several of the major assumptions in the RIA’s methodology for assessing the impact
of the proposed rule rely on shares of this larger “ABAWD” group, as defined nationwide using the
2016 data, but apply those shares to individuals who would be newly subject to the time limit
because of the proposed changes in the rules for areas to qualify for waivers. For example, the
RIA’s assumption about the share of “ABAWDs” who are working is derived from the SNAP data
for 2016 including both waived areas and not-waived areas. Using shares that are derived from a
group that includes many individuals who are not subject to the time limit, and that excludes many
individuals who have been cut off SNAP in areas that were not waived in 2016 is extremely

462

Stavrianos and Nixon (1998, p. 4) and Czajka et al. (2001, p. 30) both include flow charts that makes clear FNS had
information available that made clear which “ABAWDs” would be subject to the time limit and the various reasons
individuals who might be identified in the data as “ABAWDs” might not be affected by the time limit.

198

misleading and illogical. The denominator for these percentages matters for assessing the soundness
of using certain percentages for deriving or estimating the impact of the proposed changes.
To help elucidate the problem, we conducted an analysis comparing the number and share of
SNAP participants for two different categories of states regarding waivers from the time limit in two
different fiscal years. The two types of states were those with statewide waivers in fiscal year 2013,
but no waivers in fiscal year 2017 and those with statewide waivers in both years. 463 The difference
between the two years for the two types of states can help explain how the denominator changes
when many people are cut off as a result of the time limit. We will use figures from this analysis to
help explain some of the serious methodological problems with the RIA’s analysis of the impact of
the proposed rule. It is easier to see the issue when considering these two types of states, but it is
present in other states that have had different patterns and scope of waivers.
We use fiscal 2017 instead of fiscal year 2016 for this analysis because many states reimplemented
the time limit beginning on January 1, 2016, which means that for fiscal year 2016 the time limit was
in effect for just the latter six months of the fiscal year (April through September, once the three
months of eligibility for January to March are taken into account.) By fiscal year 2017 those states
had no waiver the entire fiscal year. Under the definition of “ABAWD” that FNS appears to use, the
3.8 million national figure cited in the FNS materials and that we recreate above in Table 11.1 fell to
3.2 million in fiscal year 2017.
Table 11.2, shows the number of SNAP participants in the two types of state in each year and the
number and share that are “ABAWDs” under our understanding of how FNS is defining ABAWDs
for purposes of the RIA ― as SNAP participants aged 18-49, with no disability benefits and no
minor children in the household.

463

The states with statewide waivers from the time limit in 2013 but no waivers at all in 2017 (which represented about a
quarter of SNAP participants in 2013) were Alabama, Arkansas, Florida, Indiana, Iowa, Kansas, Maine, Mississippi,
Missouri, North Carolina, Oklahoma, South Carolina, and Wisconsin. The states with statewide waivers in both 2013
and 2017 (which represented about 20 percent of SNAP participants) included Alaska, California, District of Columbia,
Illinois, Louisiana, Nevada, New Mexico, Rhode Island, Guam, and Virgin Islands.

199

TABLE 11.2

Waivers Affect the Number and Share of SNAP “ABAWDs”
Fiscal Years 2013 and 2017 by whether the state had a statewide waiver
Total SNAP
Participants
(in 000s)

Total “ABAWDs”
(in 000s)

ABAWDs as a
Share of Total
Participants

Fiscal Year 2013
Participants residing in states
with statewide waivers in
2013 and no waivers in 2017

12,439

1,496

12%

Participants residing in states
with statewide waivers in both
2013 and 2017

8,346

953

11%

47,098

4,943

10%

10,325

609

6%

-17%

-59%

8,164

927

-2%

-3%

41,491

3,221

-12%

-35%

Total participants all statesa

Fiscal Year 2017
Participants residing in states
with statewide waivers in
2013 and no waivers in 2017
% change 2013 to 2017
Participants residing in states
with statewide waivers in both
2013 and 2017
% change 2013 to 2017
Total participants all statesa
% change 2013 to 2017

11%

8%

Source: CBPP analysis of FY 2013 and FY 2017 USDA SNAP Household Characteristics data
a Includes participants in the two types of states identified above, as well as participants residing in other states.

Table 11.2 shows:
• The

number of both SNAP recipients and “ABAWDs” declined in both types of states
between 2013 and 2017 but fell substantially more in states that had reimposed the time limit
by 2017. The number of “ABAWDs” potentially subject to the time limit fell by 59 percent in
states that reimposed the time limit, from 1.5 million in 2013 to 600,000 in 2017. The
economy may have been stronger in these states, and there may be other reasons for a larger
drop, but the fact that eligibility rules changed and many people in this group could not
participate for more than three months was likely a major factor in the larger drop.

• The

share of total SNAP participants who were “ABAWDs” in the states that reimposed the
time limit by 2017 fell from 12 percent to 6 percent but was flat at 11 percent in the states that
had a statewide waiver in both years. So, although the number and share of “ABAWDs” fell
substantially in states that reimposed the time limit by 2017, 6 percent of SNAP participants

200

still fit into the “ABAWD” category as defined by the RIA. These individuals likely meet one
of the eligibility criteria in Table 11.1 and so were not cut off.
The share of “ABAWDs” working also changes as the denominator changes when individuals are
cut off SNAP. In Table 11.3 we compare the work rates among SNAP participants who were
“ABAWDs” under the FNS’ definition in the same two types of states as above for the same two
years.
TABLE 11.3

Waivers Affect the Number and Share of SNAP “ABAWDs” Who Are Working Fiscal Years
2013 and 2017 by whether the state had a statewide waiver

Total
“ABAWDs”
(in 000s)

“ABAWDs”
working
(in 000s)

Share of
“ABAWDs”
working

“ABAWDs”
working at
least 20
hrs/wk
(in 000s)

Share of
“ABAWDs”
working at
least 20
hrs/wk

Fiscal Year 2013
States with statewide waivers
in 2013 and no waivers in
2017

1,496

331

22%

187

13%

States with statewide waivers
in both 2013 and 2017

953

165

17%

63

7%

4,943

1,087

22%

587

12%

Total all statesa

Fiscal Year 2017
States with statewide waivers
in 2013 and no waivers in
2017

609

207

34%

138

23%

States with statewide waivers
in both 2013 and 2017

926

219

24%

111

12%

3,221

897

28%

534

17%

Total all statesa

Source: CBPP analysis of FY 2013 and FY 2017 USDA SNAP Household Characteristics data.
a Includes participants in the two types of states identified above, as well as participants residing in other states.

In the states that had reimposed the time limit by 2017, although the number of ABAWDs
dropped between 2013 and 2017, the share of “ABAWDs” working went up substantially, from 22
percent in 2013 to 34 percent in 2016, and the share estimated to be working at least 20 hours a
week nearly doubled, from 13 percent to 23 percent. In part this could be a function of a stronger
economy in these states, or other factors, but the fact that many people who were not working were
cut off also contributed significantly to the change in work rate. In the states that had statewide
waivers both years, the share working went up, but by less, from 17 percent to 24 percent, and the
share estimated to be working at least 20 hours a week went from 7 percent to 12 percent.
Between 2013 and 2017 the number of “ABAWDs” in areas without waivers went down in large
part because individuals were cut off of SNAP in areas without waivers. And because individuals
201

could continue to receive SNAP if they were working more than 20 hours a week, the share of
ABAWDs working at least 20 hours a week went up in areas without waivers, in large part because
the denominator used in calculating the share (the number of ABAWDs who received SNAP) went
down. In a 2016 report responding to a misleading report that claimed the circumstances of SNAP
recipients in Kansas improved after they reinstated the time limit we explained this math as follows:
…the childless adults who remained as SNAP participants after the time limit went into effect
were significantly different from those who participated before because of the policy change.
The state cut off SNAP those participants who were not working at least 20 hours a week, so
the work rates, average earnings, and other characteristics of those who remained SNAP
participants after the return of the time limit were better, not because those individuals
became better off, but because they were better off to begin with and were the only ones still
eligible and participating in SNAP.464
The RIA methodology includes numerous imprecise, confusing, inaccurate, or misleading
assumptions, some that push in opposite directions. We cannot tell if FNS has intentionally
produced an analysis that inflates (or deflates) the results or if the individuals charged with
producing the RIA do not understand the policy and therefore were unable to produce a coherent
analysis of the population subject to the current policy and what the impact of the proposal would
be. Either way, the lack of transparency and coherency in the RIA raises serious questions about the
validity of the NPRM process.
The RIA Does Not Explain the Claim
That There Would Be 3.4 Million “ABAWDs” in 2020
The methodology evidently assumes that there would be 3.4 million “ABAWDs” in 2020 under
current law, but never explains where that figure comes from. This is a serious omission because this
is the starting point FNS uses for all of the subsequent assumptions about the number of individuals
who would be affected by the proposed changes in waiver rules. Excluding information on this
foundational point compromises all that follows.
The only time the 3.4 million figure is mentioned, on page 25 of the RIA, the document says, “As
noted previously, the Department estimated that approximately 44 percent of the projected 3.4
million ABAWDs…would live in waived areas in FY 2020 if waiver authority were unchanged.” It is
possible that the figure comes from the 3.8 million from the SNAP QC data for 2016 cited above,
adjusted to reflect the FNS’ baseline number of participants for 2020 compared to the number of
participants in 2020, but we cannot comment on this figure as the RIA provides no justification for
it whatsoever.
If the 3.4 million is the 3.8 million from 2016 adjusted only for baseline changes, then the FNS
has made no further adjustments to account for the fact that states qualified for and applied for
waivers for fewer areas in 2018 and 2019 than in 2016 and will likely qualify for still fewer waivers in
2020 even without any changes to the waiver rules.

464

Dorothy Rosenbaum and Ed Bolen, “SNAP Reports Present Misleading Findings on Impact of Three-Month Time
Limit,” Center on Budget and Policy Priorities, December 14, 2016,
https://www.cbpp.org/sites/default/files/atoms/files/12-14-16fa.pdf.

202

The RIA Assumption That 44 Percent of “ABAWDs”
Are in Waived Areas Is Based on a Proxy That Is Indefensible
As one step in its estimate of the impact of the proposed rule, the analysis in the RIA assumes
that 44 percent of the 3.4 million “ABAWDs,” or 1.5 million individuals, would live in waived areas
in FY 2020 if the regulation were unchanged. According to RIA (page 19):
The Department used State-reported data from form FNS-388A to estimate the number of
non-public assistance SNAP participants living in currently waived areas. Since the FNS 388A
does not report ABAWD participation separately, non-public assistance SNAP participants
are used as a proxy when estimating the proportion of ABAWDs living in waived areas.
The RIA does not explain what the form FNS-388A is or why it is appropriate to use it and what
its shortcomings might be. Based on a review of the July 2018 data that states submitted, it appears
that the 388A does not include all SNAP participants (Oregon and Vermont are missing) and that
there are no county-level data for several states, including all of New England, Alaska, Idaho,
Missouri, Montana, Nebraska, New York, Utah, Washington, and Wisconsin. 465
But even if the 388A included county data for all states, it does not make sense to use the number
of participants in non-public assistance households (those that do not receive TANF cash assistance,
Supplemental Security Income, or General Assistance) as a proxy for the number of ABAWDs.
ABAWDs are much more likely to be subject to the time limit and cut off SNAP than non-public
assistance households overall. ABAWDs are a small fraction of non-public assistance households
(less than 15 percent) and their distribution across counties will depend on whether the time limit is
in effect.
To show that the time limit matters for the distribution of ABAWDs across counties, we again
used the SNAP QC data for 2013 and 2017 and again compared the states that had a statewide
waiver in 2013 and not in 2017 to states that had a statewide waiver in both years. As can be seen in
Table 11.4, in 2013 when both types of states had statewide waivers, the share of non-public
assistance participants was not a bad proxy for the share of ABAWDs ― their shares differed by
only 2 to 3 percentage points. But in 2017 using the share of non-public assistance individuals as a
proxy for ABAWDs would overstate the share in areas that were not waived ― because many
ABAWDs had been cut off in 2017 ― and understate the number in waived areas. The share of
ABAWDs in states with statewide waivers in 2017 was ten percentage points higher in 2017 than the
share of non-public assistance SNAP participants (29 percent vs. 19 percent.) Thus, the
methodology in the RIA is likely to have substantially underestimated the share of ABAWDs living
in waived areas in 2016, and projected to live in waived areas in 2020.

465

U.S. Department of Agriculture, Supplemental Nutrition Assistance Program Data,
https://www.fns.usda.gov/pd/supplemental-nutrition-assistance-program-snap.

203

TABLE 11.4

Waivers Affect the Number and Share of Non-PA SNAP Participants Fiscal Years 2013
and 2017 by whether the state had a statewide waiver
Non-PA
Participants
(in millions)

Share of
National NonPA Participants

ABAWDs
(in millions)

Share of
ABAWDs

Fiscal Year 2013
States with statewide
waivers in 2013 and no
waivers in 2017

10.9

28%

1.5

30%

States with statewide
waivers in both 2013 and
2017

6.4

16%

1.0

19%

Total all states

39.0

100%

4.9

100%

Fiscal Year 2017
States with statewide
waivers in 2013 and no
waivers in 2017

9.1

26%

0.6

19%

States with statewide
waivers in both 2013 and
2017

6.6

19%

0.9

29%

Total all states

35.2

100%

3.2

100%

Source: CBPP analysis of FY 2013 and FY 2017 USDA SNAP Household Characteristics data.
Note: “Non-PA” means not pure public assistance (PA) household. A household is considered to be pure PA if each
member of the household receives Supplemental Security Income, a cash TANF benefit, or General Assistance income.

FNS’ Methodology for Determining the Share of Areas That Would Lose Eligibility for
Waivers Is Incomplete and Confusing
The RIA includes FNS’ estimate of the number and share of currently waived areas that would no
longer qualify for a waiver under the proposed rule (755 areas, representing 76 percent of areas
currently waived), and the share of the impact that is attributable to each of the major proposed
changes to waiver rules. It also provides an explanation of its methodology for deriving these
estimates. However, the explanations are incomplete, confusing, and misleading. FNS omits
fundamental information needed to assess the integrity of its analysis. For example, it bases the
analysis on inconsistent periods of time, and provides unclear explanations of its methodological
assumptions. Understanding which areas and how many areas would lose waivers under the
proposal is central to understanding the impact of the proposed changes, but the analysis FNS
included in the RIA has significant flaws and lacks sufficient explanation to allow commenters to
understand the analysis.
The proposed rule makes three major changes to the existing rules for determining waiver
eligibility:

204

1. It sets a minimum unemployment rate of 7 percent as a floor for waiver eligibility.
It makes areas with average unemployment rates below this floor ineligible for a waiver,
even if their unemployment rates are 20 percent above the national average
unemployment rate. In contrast, there is no floor under current federal regulations. (See
Chapter 3 for more.)
2. It restricts states’ flexibility to define combined areas, making federally
designated labor market areas the only geographical groups that can be eligible
for a waiver. In contrast, under current regulations states have the discretion to combine
areas into larger geographic regions that are eligible for a waiver if the regional
unemployment rates still meet the eligibility thresholds.466 (See Chapter 5 for more.)
3. It narrows the allowable criteria for states to request statewide waivers. Under
current federal regulations and FNS guidance, states can request statewide waivers based
on average state-level unemployment rates that are 20 percent above the national average
over a recent 24-month period; average statewide unemployment rates above 10 percent
for a recent three-month or 12-month period; or based on qualifying as a state for
extended unemployment benefits.467 In contrast, the proposed rule permits to states to
request statewide waivers only when they qualify for extended unemployment benefits.
(See Chapter 4 for more.)
The problems with FNS’ estimates fall into two main categories: first, the methodology is
confusing and incomplete, and second, the discussion of the relative impact of the different changes
on areas’ eligibility for waivers is misleading.
The Methodology Is Confusing and Incomplete.
Below are specific problems with the RIA’s methodology that call into question the reliability of
its estimated impact of the rule provisions on waived counties.
• FNS’

use of the term “currently” is inconsistent; as a result it is not clear what year
FNS used for the analysis. Under both current law and the proposed rule an area’s eligibility
for a waiver for a given fiscal year is based on whether the area’s unemployment rate for a
specific earlier time period exceeds a threshold that applies to that same time period.
Throughout the RIA’s discussion of the methodology for determining the impact of the
proposed changes, FNS uses the term “currently” to refer to the year on which it is basing its
estimates. For example, on page 20, where FNS discusses its estimates of the number of areas
that would lose eligibility for waivers under the proposed rule, FNS asserts that “975 counties
and county-equivalents currently have a time limit waiver” (emphasis added).468 Since FNS
issued the NPRM on February 1, 2019, the start of the fifth month of fiscal year 2019, it

466

USDA, “Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents
(ABAWD),” December 2, 2016,” page 10.
467

7 C.F.R. § 273.24 (f) (2), and FNS guidance, “Supporting Requests to Waive the Time Limit for Able-Bodied Adults
Without Dependents,” December 2016, https://fns-prod.azureedge.net/sites/default/files/snap/SNAP-Guide-toSupporting-Requests-to-Waive-the-Time-Limit-for-ABAWDs.pdf.
468

RIA, p. 20.

205

would be reasonable to assume that the term “currently” refers to fiscal year 2019. However,
nine pages later in a discussion of “uncertainties” associated with all the estimates in the RIA,
FNS notes that “these estimates are based on current waiver eligibility as of FY 2018.”469
Moreover, as discussed below, it appears that the time period FNS used for the data on local
area unemployment rates is the time period that applies to waiver eligibility for FY 2019.
FNS’ lack of clarity about the year for which it estimated the change in waiver eligibility calls
into question whether it is assessing accurately the impact of the proposed changes.
• There

are inconsistencies in FNS’ methodology to estimate of the number of counties
that would lose waivers under the proposed rule. The analysis reveals inconsistencies in
the methodology:
1. According to current and proposed waiver rules,470 the calculation to determine
whether an area is eligible for a waiver for a given fiscal year looks at the area’s
24-month average unemployment rate over a defined earlier time period and
compares it to 20 percent above the national average for the same earlier 24month period. 471 States have to compare their areas’ unemployment rates for a
24-month period to an unemployment threshold calculated over the same 24-month
period. In addition, states need to use a 24-month time period that falls within an
earlier window that is consistent with the year for which the waiver will be
implemented.472 Contrary to its own guidance, FNS fails to use consistent
periods in its analysis:
a. The 24-month period FNS says it used for the unemployment data is
inconsistent with the year FNS says it calculated the number of waived
areas. On page 21, FNS notes that it obtained data “for the 24-month
period from January 2016 to December 2017 for 3,077 counties and
county-equivalents.” As mentioned, elsewhere the RIA asserts that it
based its estimates on eligibility for waivers in 2018, but the 24-month
period used for the unemployment data is not the correct period for a
waiver implemented in 2018. The January 2016 through December 2017
period that the Department used applies to waivers implemented in 2019.
The correct period of unemployment data that applied to areas eligible
for 2018 waivers is January 2015 through December 2016. If FNS is
estimating eligibility for waivers in 2018, it should have used
unemployment data that is applicable to that year.

469

RIA, p. 29.

470

7 C.F.R. § 273.24(f)(2)(ii).

471

The proposed rule would add the additional condition that the area’s unemployment rate under this calculation be at
least 7 percent.
472

USDA, “Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents
(ABAWD), December 2, 2016,” page 7.

206

b. The Department notes on page 21 that it used unemployment data “to
identify currently-waived counties [that] did not have a 24-month
[unemployment rate] that exceeds the current waiver threshold.” This
threshold is calculated as 20 percent above the national average
unemployment rate for a 24-month period. FNS did not specify which
24-month period it used to calculate this “current” threshold. This
creates additional confusion, given that it is already unclear which year
the Department is using for waived counties and the potential
inconsistency with the period for which the unemployment data were
collected. This also matters because the thresholds are different
depending on the waiver year FNS is analyzing. For the 24-month period
of January 2016 through December 2017, which corresponds to a 2019
waiver, the threshold is 5.5 percent. For the 24-month period of January
2015 through December 2016, which corresponds to a 2018 waiver, the
threshold is higher at 6.1 percent. If FNS is comparing county
unemployment data that apply to a 2019 waiver to the threshold for a
2018 waiver, instead of the lower threshold for a 2019 waiver, then it is
potentially underestimating the number of waived counties that would
lose eligibility as a result of restricting area combinations. Although as
mentioned above, it should have been using data for a 2018 waiver if it is
evaluating eligibility for 2018.
2. FNS only collected unemployment data for a single 24-month period, but states
are allowed to use any 24-month period that is later than the 24-month period
FNS used. FNS’ estimate assumes that all states would use the same period of
data as the basis of their requests, and that this period is representative of the
other periods of data that states could use. This is unlikely to be the case as
unemployment trends change over the course of a year, and the first 24-month
period is unlikely to accurately represent unemployment conditions in other
periods that states could use for a waiver request. In addition, the decline in
unemployment rates in recent years means that the threshold for eligibility in
subsequent 24-month periods generally decreased over the course of 2018. As a
result, the restriction of area combinations would result in fewer waived counties
losing eligibility than would be estimated under a single 24-month period. The
Department’s omission of multiple periods means that it is potentially
overestimating the number of waived counties that would lose eligibility as a
result of restricting area combinations.
3. Both inconsistencies create opposite biases, the net effect of which FNS could
demonstrate if it had taken into them into account. The fact that it ignores these
factors and does not provide a rationale for doing so shows the serious lack of
rigor in its analysis.

207

• FNS fails

to adequately explain its exclusion of certain areas from its analysis.
Footnote 8 on page 21 indicates that FNS excluded five New England states (Connecticut,
Maine, Massachusetts, New Hampshire, and Vermont) when it compiled the unemployment
data from the Bureau of Labor Statistics (BLS), but FNS does not indicate if it also excluded
these states from the list of areas that it is counting as currently waived. FNS also does not
mention these states in the rest of its analysis. As a result of these exclusions, its estimate of
the number of currently waived areas is too low.
The same footnote further explains that FNS excluded these five states because New
England counties (also known as NECTAs) are conceptually dissimilar to counties in the
rest of the United States. However, Rhode Island, which contains the Providence-Warwick,
RI-MA NECTA, does not appear on the list of excluded states. The Department fails to
mention why it included Rhode Island, which shares the same New England dissimilarities
with the other states.473 It briefly notes that “some NECTAs are quite small” and “BLS data
was not consistently available for these areas,” which appears to be a reference to the BLS’
discontinuation in 2018 of unemployment data for all cities and towns with populations
below 1,000 for all New England states.474 As Rhode Island does not contain towns with
populations below 1,000, BLS data would be available for all areas. If this were the reason
for its inclusion in the FNS analysis, this would be consistent with its rationale. But FNS
provides no information to help understand its rationale.
In addition, the Department does not explain how it treats Guam and the Virgin Islands in
its analysis. Although the Bureau of Labor Statistics does not produce employment data for
these U.S. territories, these areas were also waived in 2018 but it is unclear if they are
included in the number of currently waived areas or in the number that would lose waivers
under the proposed rule.

• FNS fails

to adjust the number of waived areas for 2020, the year in which the
proposed rule would be in effect if implemented. FNS unrealistically assumes no changes
in waivers in future years. It fails to adjust for the fact that the number of areas is likely to be
lower in FY 2020, the year in which the proposed rule would first apply if implemented. As
unemployment rates have declined, states have applied and qualified for fewer areas in 2017,
2018, and 2019. It would be realistic to assume a decline in waived areas in 2020 and later
years as well.

• FNS only

examined the impact of the proposal in a year when unemployment rates
declined. FNS only examined the impact of its proposal in 2018, a year in which the
unemployment rates declined. It did not analyze the proposal’s impact during a time of rising
unemployment rates, such as prior to or during a recession. FNS did not offer any rationale
for this exclusion. Expanding its analysis to many periods with rising and decreasing
unemployment trends would have provided a fuller understanding of the impact of the
proposed rule in different economic conditions.

• FNS does not

explain its estimation of the number of areas losing eligibility due to the
narrowing of statewide waivers. FNS provided no details about its methodology for

473

U.S. Census Bureau, “New England City and Town Areas (NECTA) Maps,” https://www.census.gov/geo/mapsdata/maps/nectas.html.
474

U.S. Bureau of Labor Statistics, “Local Area Unemployment Statistics,” https://www.bls.gov/lau/laugeo.htm.

208

estimating the effect of narrowing the criteria that can be used for statewide waivers, beyond
noting that it “estimated the number of counties and county-equivalents that would lose
waiver eligibility due to the elimination of Statewide waivers.”475 It identified an additional 39
counties eligible only because of a statewide waiver and subtracted those from its total of
waived areas that would still be eligible under the rule after already eliminating the areas that
were eligible only based on states’ ability to combine data. Based on the explanations in the
analysis, it is unclear why FNS needed to subtract out these 39 counties since all of them
would already be ineligible for waivers under the rule because their data could not be
combined with the data of other areas in the state. The description of the methodology is
confusing. It is possible that FNS was estimating the impact of the narrowing of statewide
waivers separately from the impact of combining data or the 7 percent threshold. Table 3 on
page 22 of the RIA presents the results as if FNS included the narrowing of statewide waivers
as one step within the estimation. If that is not the case, then the methodology is poorly
explained.
On the other hand, if the narrowing of statewide waivers is a step in its overall estimation,
then it appears that FNS subtracted the same counties twice. The elimination of counties that
did not meet the waiver threshold (described in the second paragraph on page 21) would
already have removed counties that are ineligible based on their own unemployment rates.
This would not leave any counties that are eligible based only on being in a state with a
statewide waiver. This additional subtraction would inflate FNS’ estimate of counties that
would lose their waivers due to narrowing of statewide waivers.
• FNS ignores extended benefits

as a standard for qualifying for statewide waivers. FNS
does not mention that some states would remain eligible for statewide waivers under the
proposed rule, based on qualifying for extended unemployment benefits (EB). Under current
FNS guidance on qualifying for a waiver based on qualifying for EB, which the proposed rule
does not change, states can request statewide waivers that start no later than one year after the
date that the state qualified for EB.476 Alaska and the District of Columbia qualified for EB in
January 2018, and therefore would have been able to request statewide waivers in 2018 (and
2019). For either year these two areas count as an additional nine county-equivalents. FNS
omission of this factor also inflates the estimate of counties that would lose their waiver under
the proposed rule because these two areas would not have lost their waivers.

• FNS does not

analyze the impact of its rules on Native American reservations or on
New England towns. FNS does not provide any analysis of the impact of the proposed rule
on Native American reservations or on New England towns and cities. This is a glaring
omission, because Native American reservations tend to have high poverty rates well above
the national average, and over 200 reservations were waived or were located inside a waived
area in 2018. This is an important segment of the population that FNS analysis ignores
completely.
Similarly, 281 towns and cities in the New England states of Connecticut, Massachusetts, New
Hampshire, Vermont, and Rhode Island were waived in 2018, constituting nearly a quarter of
all towns in these states. Although the Bureau of Labor Statistics has stopped publishing

475

RIA, p. 21.

476

USDA, “Guide to Supporting Requests to Waive the Time Limit for Able-Bodied Adults without Dependents
(ABAWD),” December 2, 2016, page 3.

209

unemployment data for New England towns with populations below 1,000 people, it
continues to provide data for towns above 1,000 people. It is unclear why the Department
does not examine the rule’s impact on those towns with higher populations.
The methodological problems listed above cast serious doubt on the reliability of the FNS overall
estimate of the impact of the proposed rule on “currently waived” areas.
FNS’ Estimates of the Relative Impacts of the Rule Provisions on Waived Areas Is Misleading
In Table 3 on page 22 of the RIA, FNS indicates that the three different changes in the proposed
rule result in a cumulative 76 percent reduction in the number of waived areas. FNS then describes
the relative impact of the three different changes in the proposed rule, asserting that the change to
restrict area combinations reduces the number of areas waived by 36 percentage points, the change
to statewide waivers reduces the number by 4 percentage points, and the 7 percent unemployment
rate floor reduces the number by 37 percentage points. However, the order FNS uses for these
calculations presents misleading results.
FNS’ presentation suggests that the proposed change to restrict area combinations and the 7
percent floor have a roughly equal impact. The presentation is misleading, however, because,
although restricting the ability to combine areas would have a substantial impact by itself relative to
current rules, the 7 percent floor has a far larger effect on areas’ eligibility for waivers because 7
percent is substantially higher than the national average unemployment threshold. As a result, all of
the areas that would lose because they cannot be combined with adjacent areas would also lose
under a 7 percent floor. An example using FNS’ own numbers (despite their flaws) can be helpful to
understand why:
Under FNS’ estimates, out of 975 counties currently waived, 220 have “a 24-month
[unemployment average] of at least 7 percent and would continue to qualify for a waiver under the
proposed waiver criteria.”477 That means 755 waived counties, or 76 percent of all counties currently
waived would lose their waiver due solely to the 7 percent floor. The additional impact of restricting
area combinations would be zero at that point.
This occurs because 7 percent is higher than the 20 percent above the national 24-month average
threshold (which would be 6.1 percent for 2018 or 5.5 percent for 2019, though as discussed above,
it’s not clear which year FNS used for the analysis.) Implementing a 7 percent minimum
unemployment rate automatically eliminates any waived counties with rates below 5.5 and 6.1
percent, and the waived counties with rates above 5.5 and 6.1 percent but below 7 percent. The
impact of the floor is therefore greater, as it eliminates counties that are eligible based on their own
24-month average unemployment rates but do not meet the floor, in addition to the counties waived
through area combinations.

477

RIA, p. 22.

210

The RIA Assumes That All ABAWDs in Areas That Lose Waivers
Will Lose SNAP; An Assumption That Ignores That Many Will Be Exempt or Able to
Participate for Other Reasons
As mentioned above, in determining the impact of the change in waivers on SNAP participants,
the RIA assumes that 1.5 million ABAWDs would live in areas that would be waived under current
rules. Under the proposed changes, the RIA assumes that:
Because waived areas are estimated to be reduced by 76 percent under the revised waiver criteria,
the department assumes that 76 percent of currently-waived ABAWDs would be newly subject
to the time limit. This equals approximately 1.1 million of the estimated 1.5 million currently
waived individuals.478
Under this assumption, no individuals who are defined as “ABAWDs” (using the Department’s
definition) in the areas that lose waivers would be:
• exempted

from the time limit because of being physically or mentally unfit;

• pregnant;
• participating because

they are eligible during the first three months of participation or qualify
for a second three-month period;

• exempt

using individual “percentage” exemptions;

• participating in a
• working (or

per week.

qualifying work program; or

complying with a qualifying work program), for less than the required 20 hours

This assumption defies logic and ignores the evidence from other states that have a time limit in
effect. In every state without waivers there are “ABAWDs” who are able to continue to participate
for these reasons. As we showed above, in Tables 11.1 and 11.2, in the states that reinstated the time
limit by 2017, the number of “ABAWDs” declined substantially, from 1.5 million in 2013 to 600,000
in 2017, but even if we subtract out the number of “ABAWDs” who were working at least 20 hours
a week (138,000), that still leaves almost half a million “ABAWDs” participating in SNAP in 2017
after the time limit went back into effect.
The assumption that none would continue to participate for any of these reasons is a glaring error
in the RIA methodology. As a result, the public is left not knowing how many people who are
potentially subject to the time limit will be able to continue to participate and for which reasons, and
whether the Department knows or cares.

478

RIA, p. 25.

211

The RIA’s Assumption That One-Third (34 Percent) of ABAWDs
Subject to the Time Limit in Areas That Lose Waivers Will Remain Eligible Because They
Would Be Working 20 Hours a Week Is Flawed
Of the 1.1 million “ABAWDs” the Department estimates would be newly subject to the time limit
under the proposed rule, the RIA assumes one-third would be working and two-thirds would “lose
their eligibility for SNAP for failure to engage meaningfully in work or work training:”
Using FY 2016 QC data, approximately 26 percent of ABAWDs were working. The
Department assumes that this proportion would increase to about 34 percent in FY 2020 if
the UR [unemployment rate] declines as projected in the 2019 President’s Budget and that
these individuals will work at least 20 hours per week. Under this scenario, the Department
estimates that approximately one-third of ABAWDs newly subject to the time limit will work
and maintain their SNAP eligibility. The remaining two-thirds (755,000 individuals) would lose
their eligibility for SNAP for failure to engage meaningfully in work or work training.479
This assumption is flawed for several reasons:
• First,

the assumption being used (the 26 percent working in 2016, rising to 34 percent in 2020)
was derived from the entire SNAP population of “ABAWDs” nationally, including both areas
that currently are waived and those that are not waived. As we demonstrated above, the work
rates of SNAP participants who live in areas without waivers are higher simply because many
people who are not able to find jobs or document their work have been cut off. It is confusing
to apply a percentage derived from the entire caseload, where many states have a large share
of individuals who live in area without waivers, to areas with waivers.

• Second,

in 2017 only 12 percent of “ABAWDs” in states that had statewide waivers in 2017
were working at least 20 hours a week ― the threshold for individuals subject to the time limit
to remain eligible for SNAP. The RIA assumes a share almost triple that (34 percent) would
be able to meet the 20 hours a week threshold, with no explanation for why or how that
would occur.

The RIA’s Assumes That All Individuals Who Lose Eligibility Will Reapply Every Three Years
When estimating the effect of the change on federal spending the RIA assumes the impact will be
felt over only nine months in 2020 and 2023, presumably based on an assumption that people who
have been cut off because of the three-month time limit will reapply immediately when they become
eligible again after three years, but it provides no evidence that this occurs. To the contrary, a study
FNS published in 2001 of state implementation of the time limit found:
Many ABAWDs who left the program have not returned. ABAWDs who used up their
time-limited benefits in 1997 became eligible again in 2000, creating the potential for a sharp

479

RIA, p. 26.

212

upswing in participation, yet the trend in participation shows no such change. 480 (Bold
emphasis in original.)

G. The RIA Lacks Transparency About the Reasons Individuals May Lose
SNAP and Other Possible Impacts If Waivers Are Narrowed
As noted, the proposed rule’s fundamental rationale is that taking away (or threatening to take
away) food assistance will cause people not currently working to get jobs. The NPRM asserts that
“these changes would encourage more ABAWDs to engage in work or work activities if they wish to
continue to receive SNAP benefits,” 481 and, “[t]he application of waivers on a more limited basis
would encourage more ABAWDs to take steps towards self-sufficiency.”482
However, the only impact that FNS quantifies in the RIA comes from the estimated, “755,000
individuals…[who] would not meet the requirements for failure to engage meaningfully in work or
work training.” The rationale and the estimated impact are inconsistent. Moreover, the assessment
lacks transparency about what the impact would be and it over-simplifies the possibilities. Other
FNS materials indicate that FNS is aware that when the time limit is in effect it results in different
outcomes for different groups of people. Many people participate in SNAP for a period of time and
then leave when their circumstances change. But, when faced with a time limit there are a range of
possible outcomes and impacts. For example:
• Some

individuals may find work or additional hours and as a result their SNAP benefits may
go down as a result of income they would not have otherwise had.

• Others

may wish to comply but be unable to find a job. For these individuals there is a
question of whether a qualifying slot in an Employment and Training program would be
available. If there is no slot then that individual would likely lose SNAP. If there is, FNS and
states would potentially incur a cost for the E&T services. FNS does not contemplate any
changes in E&T that might be caused by the proposed rule.

• Other

individuals may qualify for an exemption for “unfitness for work,” or another reason,
but may lose SNAP if they don’t realize they could qualify for an exemption because it was
not properly explained to them, or if they are unable to get documentation of their health
issue because they lack medical coverage.

• Others

may actually be working, but not comply with paperwork requirements to document
their hours of work.

• And,

although unlikely, other individuals who are able to work may intentionally choose not
to comply with the time limit and lose SNAP benefits.

480

John L. Czajka et al., “Imposing a Time Limit on Food Stamp Receipt: Implementation of the Provisions and Effects
on Food Stamp Program Participation: Final Report,” prepared by Mathematica Policy Research for the USDA, Food
and Nutrition Service, September 4, 2001, https://fns-prod.azureedge.net/sites/default/files/abawd.pdf, p. xxv.
481

NPRM, p. 981.

482

NPRM, p. 981.

213

This list is not exhaustive. But FNS guidance in recent years makes clear that it is aware of both
the range of possible outcomes and the fact that the distribution of outcomes can be influenced by
state implementation choices. For example, in November 2015, FNS issued guidance that reminded
states that in addition to tracking months of participation, “States must also carefully screen for
exemption from the time limit and connect ABAWDs to the information and resources necessary to
maintain eligibility consistent with Federal requirements.”483 The guidance covered several areas
where state implementation could affect individuals’ eligibility, for example:
• “Screening for

Exemptions and Fitness for Work”;

• “Maintaining Eligibility

through Work Programs and Workfare”:

• “Maintaining Eligibility

through Unpaid or Volunteer Work”; and

• “Good

Cause for Failure to Meet the ABAWD Work Requirement.”

Another FNS guidance focused on states’ responsibilities to adequately notify individuals who are
potential ABAWDs on the details of the time limit, work requirement, exemptions, and their
responsibility to report changes in work hours.484 FNS followed this with another guidance that
outlined best practices and provided model language to “help State agencies effectively inform
Supplemental Nutrition Assistance Program (SNAP) households of the requirements for ablebodied adults without dependents (ABAWD) and to enrich training for eligibility workers.”485
The RIA oversimplifies the various impacts of the rule. Research has found, and FNS is aware,
that in practice work requirements result in individuals experiencing benefit cuts for a variety of
reasons, including when they cannot find jobs, when they should have been found exempt from the
requirement, and when they are working but fail to comply with paperwork requirements. FNS
failed to adequately reflect the various possible reasons why individuals would lose SNAP under the
proposed rule and as a result failed to adequately explain or consider its impact.

H. The RIA’s Estimate of No Impact From Eliminating the Carryover of
Exemptions Is Confusing and Misleading
The proposed rule would eliminate states’ ability to “carry over” exemptions that go unused in
one year into future years. The RIA includes two confusing and misleading assumptions about state
use of exemptions — one about current state use of exemptions and the other about how states
would use exemptions under the proposed rule.
483

U.S. Department of Agriculture, Food and Nutrition Service, “Supplemental Nutrition Assistance Program ABAWD Time Limit Policy and Program Access,” November 19, 2015, https://fnsprod.azureedge.net/sites/default/files/snap/ABAWD-Time-Limit-Policy-and-Program-Access-Memo-Nov2015.pdf.
484

U.S. Department of Agriculture, Food and Nutrition Service, “SNAP – Requirements for Informing Households of
ABAWD Rules,” April 17, 2017, https://fnsprod.azureedge.net/sites/default/files/snap/Requirements_for_Informing_Households_of_ABAWD_Rules.pdf.
485

U.S. Department of Agriculture, Food and Nutrition Service, “SNAP - Best Practices and Resources for Informing
Households of ABAWD Rules, May 25, 2018, https://fnsprod.azureedge.net/sites/default/files/snap/BestPracticesforInformingABAWDS.pdf.

214

The RIA dramatically overstates the number of exemptions states have used in recent
years. The RIA methodology includes an assumption that “states use approximately 65 percent of
their earned exemptions in an average year.” This assumption implies that many states use a large
share of their annual exemptions each year. This is a significant misrepresentation of the pattern of
state use of exemptions.
A closer look at the data FNS posts about the pattern of state use and accrual of exemptions486
shows that the actual pattern is that:
• many

states have not used any exemptions in most years;

• some

states have used a small share of the exemptions they earned for that year;

•a

few states have used the majority of the exemptions they earned for that year;

•a

few states have not earned exemptions for a year but have dipped into their accrued
exemptions; and

•a

handful of states have used more exemptions than they earned in a given year.

The last two categories result in the number of used exemptions as a share of earned exemptions
for that year exceeding 100 percent in that state. Across all states this will raise the total share of
exemptions used because the denominator for the percentage is the number of exemptions earned
for that year (as it appears to be for the FNS assumption.) See Table 11.5, below, for the distribution
of states across these categories. You can see that the small number of states in the last two
categories is playing an outsized role in raising the total share across all the states.
TABLE 11.5

State use of exemptions 2014-2018
Number of States:

2014

2015

2016

2017

2018a

Using no exemptions

43

41

21

19

N/A

Using less than 50% of
earned exemptions

4

7

15

26

N/A

Using 51-100% of earned
exemptions

1

2

0

5

N/A

Using more than 100% of
earned exemptions

2

0

2

1

N/A

Used exemptions but
earned none for that year

3

3

15

2

N/A

Average exemptions used
as a share of earned

93%

27%

149%

23%

N/A

230,000

115,000

730,000

300,000

1,300,000

Total exemptions used

486

U.S. Department of Agriculture, ABAWD 15 Percent Exemptions Data, https://www.fns.usda.gov/snap/abawd-15percent-exemptions.

215

Source: CBPP analysis of FNS-posted data on exemptions https://www.fns.usda.gov/snap/abawd-15-percentexemptions.
a According to the RIA (p. 23), “In FY 2019 state carried over approximately 6.1 million unused exemptions from the prior
year.” Since, according to FNS data, states carried 7.4 million exemptions into FY2018 they must have used about 1.3
million exemptions in FY 2018 (7.4 million – 1.3 million = 6.1 million.)

Thus, FNS’ statement that “States use approximately 65 percent of their earned exemptions in an
average year” is highly misleading. In fact, across all states and the four years between 2014 and 2017
only eight states used between 51 and 100 percent of their exemptions. The vast majority used none,
some used less than 50 percent, and a handful used more than 100 percent.
FNS’ assumption that eliminating the exemption carryover would have no impact is
indefensible. FNS estimates the number of exemptions that would be taken away from states and
concludes that the rule would eliminate 6.6 million case-months of carryover exemptions the first
year (FY 2020) and 160,000 to 180,000 a year in later years.
FNS estimates no impact from the proposed change to eliminate states’ past and future
exemptions from prior years, saying:
It is difficult to estimate the impact of such a change on transfer spending because there is no
historical record to support an estimate of if and when such a “run on the bank” may occur.
Current practice by the states suggests that elimination of the carryover will have no change
on transfers as the exemptions that will expire represent exemptions that were not distributed
to covered individuals (i.e., no transfer is occurring, so no transfer can be reduced.) However,
elimination of the carryover will give the Federal government greater predictability over
potential spending requirements because the number of exemptions subject to the sole
discretion of the states is smaller.487
The assumption of no change in federal spending from eliminating so many exemptions is highly
misleading and contrary to the experience of the last 23 years. Virtually every state has used waivers
at some point since 1996, and most states have used exemptions in some years, making clear that in
some economic situations and under some political leadership states wish to shield some SNAP
recipients who are subject to the time limit from losing SNAP.
Some states have used exemptions to suspend the time limit in areas where no E&T services are
available or to transition counties from waiver status to non-waiver status and give the area time to
establish or expand employment and training opportunities to meet the needs of individuals subject
to the time limit. Other states have used exemptions to continue to provide SNAP to certain SNAP
participants who would be cut off because of the time limit but who they determine should continue
to receive SNAP, such as individuals who are working, but less than 20 hours in a particular month,
or individuals who recently have been released from incarceration.
The information that FNS makes public about exemption usage in recent years shows that as
many areas have no longer been waived in recent years, states have increased their use of
exemptions. It seems highly likely that if the rule went into effect and states faced losing waivers for
a large share of the counties in their state with the highest unemployment rates, many would opt to
487

RIA, p. 24.

216

draw down more exemptions, and, over time, to draw down the balances of their exemptions that
they have been allowed to carry over.
Another example of when states may use exemptions is when the political leadership in a state
changes to be more sympathetic to the harshness of the time limit. In that case states might establish
a policy that begins to use exemptions and use them at a higher rate than the number that are
accrued each year. The carryover exemptions would allow such a state to sustain a larger exemption
policy for several years.
The assumption that no carried over exemptions would ever be used is indefensible, especially in
combination with the changes the proposed rule would make to the share of the United States that
could qualify for waiver.

I. The RIA Fails to Accurately Reflect the Impact of the Proposed Rule on
Medicaid and Health Coverage and Other Secondary Impacts
The Centers for Medicare and Medicaid Services of the Department of Health and Human
Services, the agency that administers the Medicaid program, has required some alignment between
SNAP and Medicaid work requirements. Specifically, states must count enrollment in SNAP as an
automatic exemption from Medicaid work requirements since individuals enrolled in SNAP are
either exempt from or complying with SNAP work requirements.488
As a result, the proposed rule’s changes to SNAP waiver and exemption policy would have a
direct ripple effect on individuals’ Medicaid eligibility and coverage. More people in states with
Medicaid work requirements would be subject to those work requirements, and a large number
would very likely lose Medicaid coverage. The per-person cost of health coverage often is higher
than the monthly SNAP benefit. The federal budget savings and the impact on individual’s health
coverage from this direct link between SNAP and Medicaid under the Administration’s policies
should have been reflected in the RIA’s cost-benefit analysis. The RIA’s failure to mention and
quantify these effects is a serious oversight that fails to accurately reflect the full impact of the
proposed rule.
Moreover, the RIA does not mention nor quantify several secondary effects that SNAP benefit
cuts could have. For example:
• SNAP

benefits are one of the fastest, most effective forms of economic stimulus when
the economy is weak. Low-income individuals generally spend all of their income meeting
daily needs such as shelter, food, and transportation, so every dollar in SNAP that a lowincome family receives enables the family to spend an additional dollar on food or other items.
Moody’s Analytics estimated that every $1 increase in SNAP benefits during 2009, when the
economy was in a recession, generated about $1.70 in economic activity.

488

State Medicaid Director Letter, January 11, 2018, https://www.medicaid.gov/federal-policyguidance/downloads/smd18002.pdf.

217

• SNAP

has been found to improve some recipients’ health outcomes. SNAP is associated
with better health and lower health care costs, according to a growing body of evidence. 489

489

Steven Carlson and Brynne Keith-Jennings, “SNAP is Linked with Improved Nutritional Outcomes and Loer Health
Care Costs,” Center on Budget and Policy Priorities, January 17, 2018, https://www.cbpp.org/research/foodassistance/snap-is-linked-with-improved-nutritional-outcomes-and-lower-health-care.

218

Chapter 12. The Proposed Rule Would
Disproportionately Impact Individuals Protected by
Civil Rights Laws, Violating the Food and Nutrition
Act’s Civil Rights Protections
According to FNS estimates, under the proposed rule some 755,000 individuals would lose
eligibility for SNAP because of a “failure to engage meaningfully in work or work training.” 490 As
described in detail in Chapter 3, evidence from the research on the impact of work requirements and
time limits, as well as the disparities in unemployment in the labor market make clear that the cuts to
SNAP eligibility from the proposed rule would fall disproportionately on African Americans,
Latinos, and people with disabilities. In addition, Native Americans also would experience a
disproportionate impact from the proposed rule because individuals who are Native American,
whether or not they reside on Indian reservations, also have poverty and unemployment rates well
above the national average, and many of the over 200 reservations that were waived or were located
inside a waived area in 2018 would likely lose their waivers under the proposed rule.
In the civil rights impact analysis included in the NPRM, FNS recognizes the disproportionate
impact, citing the rule’s “potential for disparately impacting certain protected groups due to factors
affecting rates of employment of members of these groups.”491 But the analysis finds that “the
implementation of mitigation strategies and monitoring by the Civil Rights Division of FNS will
lessen these impacts,” without providing any evidence or examples of how that mitigation could
occur. It is not clear how the Civil Rights Division of FNS could mitigate an eligibility policy that
inherently results in a disproportionate impact on certain groups. We cannot comment on the
potential effectiveness of such efforts when the NPRM does not provide any information about
what they might be and no similar interventions have occurred in the history of the program. If FNS
envisions giving the Civil Rights Division a role in determining eligibility for waivers — which the
Division apparently has not had to date — it says nothing about that in the NPRM and we cannot
readily imagine how that would work. Even if it did, without anything in the rule varying the effects
of the new standards it imposes, states would be unlikely to request the kinds of waivers that might
mitigate the rule’s disparate impact on members of protected groups.
Moreover, even if mitigation of the disparate impact were possible, the fact that the proposed rule
still would have a disproportionate impact on these protected groups directly violates Section
11(c)(2) of the Food and Nutrition Act (7 U.S.C. § 2020(c)(2)). In the 2008 farm bill Congress
reasserted its commitment to nondiscrimination and made clear that certain civil rights laws apply to
SNAP:492
(c) CIVIL RIGHTS COMPLIANCE.—

490

NPRM, p. 989.

491

NPRM, p. 990.

492

The Food, Conservation, and Energy Act of 2008 (P.L. 110-246), section 4117.

219

(1) IN GENERAL.—In the certification of applicant households for the supplemental
nutrition assistance program, there shall be no discrimination by reason of race, sex, religious
creed, national origin, or political affiliation.
(2) RELATION TO OTHER LAWS.—The administration of the program by a State
agency shall be consistent with the rights of households under the following laws (including
implementing regulations):
(A) The Age Discrimination Act of 1975 (42 U.S.C. 6101 et seq.).
(B) Section 504 of the Rehabilitation Act of 1973 (29 U.S.C. 794).
(C) The Americans with Disabilities Act of 1990 (42 U.S.C. 12101 et seq.).
(D) Title VI of the Civil Rights Act of 1964 (42 U.S.C. 2000d et seq.).493
Of particular note is that, under this amended language, the regulations implementing Title VI and
other civil rights statutes are fully applicable to SNAP. These regulations prohibit actions in federal
programs that have disparate impacts on members of protected groups as well as intentional
discriminatory acts. Therefore, the proposed rules’ disparate impact on these individuals, as
demonstrated by the research and conceded in the NPRM itself, is inconsistent with the Act. Key
members of Congress made unmistakably clear that this is what the 2008 amendments sought to
accomplish.
In his floor statement on the 2008 farm bill, Representative Joe Baca, who at the time was the
ranking member of the House Agriculture Subcommittee on Departmental Operations, Oversight,
Nutrition and Forestry, explained:
… this legislation makes clear that the [Agriculture] Department’s civil rights regulations are
among those which have the full force of law and which households have the right to enforce.
Discrimination is not acceptable in any form or at any point in the food stamp certification
process. Households should not be assisted, or not assisted, approved or denied for any
reason other than an individual assessment of their need for help or their eligibility by the
state.494
Senator Dick Durbin, a leading member of the Senate Judiciary Committee, in his floor statement
on the 2008 farm bill similarly stated that “This legislation also makes explicit that various civil rights
laws are binding in the Food Stamp Program. This is not a change — these laws and their
regulations have applied since they were written, and both have been intended to be fully
enforceable.”495

493

Section 11(c) of the Food and Nutrition Act of 2008, 7 U.S.C. § 2020(c).

Congressional Record, May 22, 2008, p. H3814, https://www.govinfo.gov/content/pkg/CREC-2008-0514/pdf/CREC-2008-05-14-pt1-PgH3801-3.pdf#page=13.
494

495

Congressional Record, May 22, 2008, p. S4747, https://www.congress.gov/crec/2008/05/22/CREC-2008-05-22pt1-PgS4743-3.pdf.

220

Given this clear expression of Congressional intent, FNS may not by regulation exacerbate
discrimination within SNAP based on race, ethnicity, or disability. Since FNS recognizes that the
proposed rule would have discriminatory effects, it must withdraw the rule.

221

Chapter 13. The Proposed Rule Fails to Adequately
Estimate the Impact on Small Entities
The Regulatory Flexibility Act (5 U.S.C. § 601-612) requires agencies to analyze the impact of a
proposed rule specifically on small businesses and entities through an initial regulatory flexibility
analysis. The Regulatory Flexibility Act specifically mandates that the analysis must contain a series
of arguments including, but not limited to: why action by the agency is being considered, what the
legal basis is for the proposed rule, and an estimate to the number of small entities to which the rule
would apply.496 The FNS failed to undertake the necessary research regarding the impact of this rule
on all small entities, with the proposed rule offering only a brief impact report with minimal analysis
that fails to accurately or adequately assess the impact of the proposed rule.
The FNS claims that aside from program participants, the proposed rule would primarily impact
state agencies. This assessment leaves out a key group of impacted stakeholders — small SNAP
retailers, who rely on SNAP purchases for consistent and dependable revenue. The Department
incorrectly assumes that after losing benefits, people would replace their monthly SNAP allotment
with cash. The individuals impacted by this NPRM are a very low-income group, as approximately
70 percent of all “ABAWD’s” are below half of the federal poverty line. 497 They do not tend to have
disposable income, and taking away their SNAP benefits would take away their ability to purchase
food. Additionally, SNAP benefits normally run out for most households before the end of the
month. 498 Many households spend their benefits rapidly because they are funds designated
specifically for food. Cash cannot be used to replace SNAP because these dollars are needed to pay
other expenses such as rent, clothing, gasoline, and many other necessities.499 The Department’s
primary and false assumption that SNAP is supplemental rather than essential lays an untrue
foundation for the argument that small retailers would not be disproportionately impacted.
Additionally, the NPRM includes an inaccurate estimate of the number of small retailers that
would be impacted. This leaves the public and stakeholder groups ill-informed about the potential
implications of the rule. A small retailer at risk of being significantly harmed by the proposed rule
would not understand the importance of the issue solely by reading the NPRM and Regulatory
Impact Analysis due to the failure to scale the estimation exclusively to the impacted areas. This
section will review which pieces of the Regulatory Flexibility Act were not adequately covered, the
impact and magnitude of the inaccurate estimation of impacted small businesses in the NPRM, and
which areas would be disproportionately or significantly impacted across the United States.

496

Regulatory Flexibility Act, 5 U.S.C. § 603, https://www.sba.gov/advocacy/regulatory-flexibility-act.

497

Center on Budget and Policy Priorities, “Unemployed adults without children who need help buying food only get
SNAP for three months,” https://www.cbpp.org/unemployed-adults-without-children-who-need-help-buying-foodonly-get-snap-for-three-months.
498

Dottie Rosenbaum, “Many SNAP Households Will Experience Long Gap Between Monthly Benefits Despite End of
Shutdown,” Center on Budget and Policy Priorities, revised February 4, 2019, https://www.cbpp.org/research/foodassistance/many-snap-households-will-experience-long-gap-between-monthly-benefits.
499

Ibid.

222

A. Inadequately Addressed Sections of the Regulatory Flexibility Act
The primary area of concern within the Regulatory Flexibility Act (R.F.A.) is §603 – Initial
regulatory flexibility analysis.500 According to the R.F.A §603, an agency publishing an NPRM is
required to do the following:501
(b) Each initial regulatory flexibility analysis required under this section shall contain —
(1) a description of the reasons why action by the agency is being considered;
(2) a succinct statement of the objectives of, and legal basis for, the proposed rule;
(3) a description of and, where feasible, an estimate of the number of small entities
to which the proposed rule will apply;
(4) a description of the projected reporting, recordkeeping and other compliance
requirements of the proposed rule, including an estimate of the classes of small
entities which will be subject to the requirement and the type of professional skills
necessary for preparation of the report or record;
(5) an identification, to the extent practicable, of all relevant Federal rules which may
duplicate, overlap or conflict with the proposed rule.
(c) Each initial regulatory flexibility analysis shall also contain a description of any significant
alternatives to the proposed rule which accomplish the stated objectives of applicable
statutes and which minimize any significant economic impact of the proposed rule on small
entities. Consistent with the stated objectives of applicable statutes, the analysis shall discuss
significant alternatives such as —
(1) the establishment of differing compliance or reporting requirements or timetables
that take into account the resources available to small entities;
(2) the clarification, consolidation, or simplification of compliance and reporting
requirements under the rule for such small entities;
(3) the use of performance rather than design standards; and
(4) an exemption from coverage of the rule, or any part thereof, for such small
entities.
The provided regulatory flexibility analysis in the NPRM fails to include a detailed description
required by §603(b)(1) and §603(b)(2), as the analysis includes no legal basis for the proposed rule or
why the agency is considering the action. Additionally, §603(b)(3) mandates an estimate of the
number of small entities impacted. The agency provides an estimate, but that estimate is flawed as
we show below. 502

500

Regulatory Flexibility Act, 5 U.S.C. § 603, https://www.sba.gov/advocacy/regulatory-flexibility-act.

501

Ibid.

502

CBPP Internal Analysis of U.S. Department of Agriculture, Food Environment Atlas Data 2018.

223

By inaccurately estimating the number of small entities that would be impacted by the proposed
rule, the Department assumed it was not required to satisfy other requirements in the Act. For
example, §609(a)(1) states that when a rule is introduced that will have a significant economic impact
on small entities; the respective agency must provide a statement or notice to the effect on small
entities.503 Providing an imprecise estimate, the Agency is able to argue that no significant impact will
be made, keeping small entities uninvolved with the rulemaking process. In addition, §603(c) of the
RFA requires a description of potential alternatives to the proposed rule. The NPRM fails to
provide any possible alternatives because the proposed rule inaccurately asserts that there is no
disproportionate impact on small entities, falsely excusing them from additional requirements.
Incorrectly estimating the number of small businesses not only represents a lack of specificity, it
more importantly exempts the Agency from providing an advanced notice to small entities, allowing
them to submit comments and address concerns in the NPRM.

B. Impact of the NPRM on Small SNAP Retailers
Perhaps the most troubling of the agency’s regulatory flexibility analysis is the inadequate estimate
of the total number of small SNAP retailers. The NPRM does accurately estimate that there are
nearly 200,000 retailers that fall under the Small Business Administration’s gross sales threshold, but
it is imprecise to assume that all of these stores would be impacted by the proposed rule.504 As a
result, the lost sales per store are too low, failing to signal to small entities the magnitude of their
losses from the NPRM. An internal analysis at the Center on Budget and Policy Priorities has shown
that a total of 639 counties across 28 states would be impacted by the proposed rule.505 Within these
counties, there are only a total of about 67,000 SNAP retailers of all sizes.506 If the same percentage
of these businesses were considered to be small entities under the Small Business Administration
gross sales threshold used in the NPRM (76 percent), then it can be estimated that a total of nearly
51,000 small entities would be impacted, significantly less than the NPRM’s estimate of 200,000.507
By making an estimate based off the total number of small SNAP retailers in the United States
(200,000) versus the number of small retailers impacted by the NPRM (51,000), FNS has artificially
lowered the average of the sales lost by four times.508 FNS has conducted a cursory analysis
regarding the impact of the proposed rule on small businesses. Using an estimate of nearly four
times the true number of stores potentially impacted minimizes the reality that small SNAP retailers
would face from the NPRM.

503

Regulatory Flexibility Act, 5 U.S.C §609(a)(1) https://www.sba.gov/advocacy/regulatory-flexibility-act.

504

NPRM, Regulatory Flexibility Act https://www.federalregister.gov/d/2018-28059/p-118.

505

CBPP Analysis of BLS Unemployment data, 2019.

U.S. Department of Agriculture, Food Environment Atlas Data 2018 https://www.ers.usda.gov/dataproducts/food-environment-atlas/data-access-and-documentation-downloads/.
506

507

CBPP Internal Analysis of U.S. Department of Agriculture, Food Environment Atlas Data 2018.

508

CBPP Internal Analysis, of U.S. Department of Agriculture, Food Environment Atlas Data 2018.

224

According to the NPRM, SNAP benefit payments are expected to be reduced by about $1.7
billion per year.509 By conducting the same calculation as FNS in the NPRM while including a more
accurate estimate of the amount of impacted small businesses ($1.7 billion x 15% redeemed at small
retailers / 51,000 stores losing waivers / 12 months), we can estimate that the loss of revenue per
small store on average each month would be $417, compared to the NPRM estimate of $106.510 The
NPRM subsequently states that the average small store redeemed $3,800 in SNAP each month in
2017, making the NPRM estimate representative of 3 percent of monthly store sales.511 In evaluating
the impact on small stores with the true loss of revenue per store ($417/month), the average small
store would realistically face an 11 percent reduction in the amount of SNAP benefits redeemed at
each store. Not only is 11 percent a significant portion of a store’s SNAP revenue, more importantly
it is nearly four times greater than the estimated impact from the initial regulatory flexibility analysis
of 3 percent in the NPRM. 512 Using the more accurate estimate would have properly signaled the
implications of the rule to small entities, allowing them the opportunity to comment on the
proposed rule.
Small Businesses Located in Rural Areas Will Be Disproportionately Impacted
The small business impact assessment in the NPRM claims that small retailers are not expected to
be disproportionately impacted by the proposed rule. This is not correct. Using 2018 as an
illustrative year, a total of 639 counties or county-equivalents would have lost access to the waiver
across 28 states.513 Of those impacted areas (which include counties, reservations, and small cities),
405 areas have a population of fewer than 50,000 people, where the Census Bureau defines the
cutoff for urbanized areas.514,515 Research has shown that rural areas with lower population levels
often rely on corner and convenience stores for food, many of which are individually owned small
businesses.516 When nearly two-thirds of the counties impacted by an arbitrary rule depend on small
businesses for SNAP purchases, the rule is certain to have a substantial impact on small business in
the impacted areas. As a result, the claim that small entities will not be substantially impacted is
incorrect.

509

NPRM, Regulatory Flexibility Act https://www.federalregister.gov/d/2018-28059/p-118.

510

Ibid.

511

Ibid.

512

Ibid.

513

CBPP Analysis of BLS Unemployment data, 2019.

U.S. Department of Agriculture, Food Environment Atlas Data 2018, https://www.ers.usda.gov/dataproducts/food-environment-atlas/data-access-and-documentation-downloads/.
514

515

Michael Ratcliffe et al., “Defining Rural at the U.S. Census Bureau,” United States Census Bureau, U.S. Department
of Commerce, Dec 2016, pp. 1-8, https://www2.census.gov/geo/pdfs/reference/ua/Defining_Rural.pdf.
516

Joseph Sharkey et al., “Association between neighborhood need and spatial access to food stores and fast food
restaurants in neighborhoods of Colonias,” International Journal of Health Geographics, 2009, pp. 1-17.

225

Individuals living in rural areas must travel longer distances to supermarkets and grocery stores
than their urban or suburban counterparts.517 As a result, they visit corner and convenience stores
frequently for their daily food and nutrition needs.518 A study conducted by USDA found that newentrant SNAP households along with SNAP households that had participated in the program for
less than six months lived, on average, four miles from a grocery store and 1.6-1.8 miles from a
convenience store. 519 While this may not appear to be a substantial difference in distance, individuals
living in rural areas are less likely to have access to public transportation and a private vehicle.520
Because lack of transportation is a common issue across low-income rural areas, residents rely on
the closest store to spend their monthly SNAP benefits.521 As mentioned, these closest stores are
typically convenience stores or small businesses, offering a limited selection of foods.
In addition to many of the impacted counties being rural and facing issues around food access, a
significant portion of the impacted counties suffer from extremely low access to food.522 Sixty-three
counties (10 percent of those impacted by the proposed rule) have five or less SNAP retailers across
the county.523 While the impact of the proposed rule on retailers across all rural counties would be
considerable, these 63 counties with such few SNAP vendors would be disproportionately impacted.
In these cases particularly, rural residents are often required to travel longer than the previously
mentioned average distances to access a supermarket or superstore.524 Table 13.1 shows some of the
impacted rural counties with few or lone SNAP retailers:

517

Joseph Sharkey, “Measuring Potential Access to Food Stores and Food-Service Places in Rural Areas in the U.S.,”
American Journal of Preventive Medicine, April 2009, pp. S151-S155.
518

Renee Walker, Christopher Keane, and Jessica Burke, “Disparities and access to healthy food in the United States: A
review of food deserts literature,” Health & Place, 2010, pp. 876-884.
519

James Mabli, “SNAP Participation, Food Security, and Geographic Access to Food,” Food and Nutrition Service,
Office of Policy Support – U.S. Department of Agriculture, March 2014, pp. 1-50.
520

Kevin Matthews et al., “Health-Related Behaviors by Urban-Rural county Classification – United States, 2013,”
Morbidity and Mortality Weekly Report, Centers for Disease Control and Prevention, February 2017, pp. 1-12.
521

C. Pinard et al., “An integrative literature review of small food store research across urban and rural communities in
the U.S.,” Preventive Medicine Reports, April 2016, pp. 324-332.
522

Lisa Powell et al., “Food store availability and neighborhood characteristics in the United States,” Preventive Medicine
(2007), pp. 189-195.
523

CBPP Internal Analysis of U.S. Department of Agriculture, Food Environment Atlas Data 2018.

524

C. Pinard et al., “An integrative literature review of small food store research across urban and rural communities in
the U.S.,” Preventive Medicine Reports, April 2016, pp. 324-332.

226

TABLE 13.1

Examples of Rural Counties Disproportionately Impacted by the NPRM
State
North Dakota
Kentucky
Virginia
South Dakota
Nevada
West Virginia

County
Eddy County
Robertson County
Charles City County
Mellette County
White Pine County
Doddridge County

SNAP Participants as of
July 2018525

Average Monthly SNAP
Retailers in 2016526

167
290
827
655
1,131
1,187

1
2
2
3
5
5

Source: CBPP Internal Analysis of U.S. Department of Agriculture, Food Environment Atlas Data, 2018

Table 13.1 shows that rural counties impacted by the NPRM with few SNAP retailers, or a single
SNAP retailer, would face significant harm if sales were lost. These are solely a few illustrative
examples of the counties impacted. These examples validate the potential of the NPRM on rural
counties with minimal SNAP retailers. A loss of sales for any of these isolated SNAP retailers would
inhibit their ability to provide for the surrounding community. A much more robust nationwide
assessment ought to have been included in the NPRM in order to meet the requirements of the law
and to allow small entities the opportunity to meaningfully engage on what the proposed rule might
mean to their sales and business. It is not representative of the Department to ignore its own
research and knowledge of the location of small businesses impacted by the rule. This failure and
lack of discussion on the impact of small businesses leaves that constituency and others unable to
comment effectively on the proposed rule.
Small Retailers in Urban Areas Will Be Significantly Impacted
While many of the counties that will lose waivers under the proposed rule are rural, a noteworthy
portion of the counties that would no longer be eligible to be waived is considered urban. Some of
the counties with the highest population in the country would be impacted, subjecting a great
number of recipients in a condensed area to the proposed rule. The following are a few of the
impacted urban counties and cities:

525

U.S. Department of Agriculture, SNAP State Issuance and Participation Estimates (FNS-388 & FNS-388A), data as
of July 2018, https://www.fns.usda.gov/sites/default/files/pd/SNAP-FNS388A.zip.
526

USDA, Food Environment Atlas, last updated on March 27, 2018, https://www.ers.usda.gov/data-products/foodenvironment-atlas/data-access-and-documentation-downloads/.

227

TABLE 13.2

Examples of Urban Locales Significantly Impacted by the
NPRM

State

County

CA

Los Angeles
County
Cook
County
Wayne
County
Clark
County
Philadelphi
a County

IL
MI
NV
PA

Primary City

SNAP
Participants
as of July
2018527

Statewide
Share of
“ABAWD”
Population

Los Angeles

1,055,314

11.3%

120,000

$19,444,000

Chicago

813,465

12.3%

100,000

$16,295,000

Detroit

416,321

11.4%

47,000

$7,685,000

Las Vegas

352,675

13.9%

49,000

$7,984,000

Philadelphia

473,269

6.6%

31,000

$5,040,590

Estimated
“ABAWD”
Population*

Estimated
Monthly
“ABAWD”
Benefits**

Source: CBPP analysis of USDA SNAP Household Characteristics data, FY 2017, where the average monthly “ABAWD”
benefit is $162.4
*Estimated “ABAWD” Population is derived from share of “ABAWDs” in each state applied to county caseload.
**Estimated Monthly “ABAWD” Benefit is calculated from average “ABAWD” benefit and ABAWD share of population.

The cities shown in Table 13.2 represent some of the largest cities impacted by the proposed rule.
Across the five cities listed, an internal analysis estimates a total of 347,000 “ABAWDs” residing
within these cities. While not all of these individuals are necessarily subject to the proposed rule due
to exemptions and other factors, it is critical to recognize that the average “ABAWD” SNAP
monthly benefit in FY2017 was $162.4, or an estimated annual contribution of $677,383,000 in
SNAP benefits from “ABAWDs” to the combined economies of these cities.528
Similar to rural residents, individuals living in low-income urban areas often depend on
convenience stores for groceries because of the lack of accessibility to full-sized supermarkets and
grocery stores.529 In 2009, a study found that “ZIP codes representing low-income areas had only
75% as many chain supermarkets available as ZIP codes representing middle-income areas.”530 With
some urban-dwelling SNAP recipients essentially being forced to redeem benefits at small retailers
because of the lack of access, the argument that urban small retailers would not be impacted by the
proposed rule is representative of the lack of consideration that FNS has presented.
527

U.S. Department of Agriculture, FNS 2018 County level SNAP data.

528

“Characteristics of Able-bodied Adults without Dependents,” Food and Nutrition Services, United States
Department of Agriculture, 2016, https://fns-prod.azureedge.net/sites/default/files/snap/nondisabled-adults.pdf.
Melissa Nelson Laska et al., “Healthy food availability in small urban food stores: a comparison of four US cities,”
Public Health Nutrition, December 2009, pp. 1031-1035.
529

530

Nicole Larson, Mary Story, and Melissa Nelson, “Neighborhood Environments: Disparities in Access to healthy
Foods in the U.S.,” American Journal of Preventive Medicine, 2009, pp. 74-81.

228

There have been additional studies conducted concerning food access in individual cities that
would be impacted by the proposed rule. For example, the food environment of Philadelphia has a
wealth of research demonstrating food access issues across the city. A study from 2014 interviewed
hundreds of low-income individuals in Philadelphia and found that residents often shopped at
convenience stores because of “easy parking, accommodation of physical disabilities or special
needs, and the integration of food shopping into other daily activities.”531 With potentially impacted
cities having published information about how their residents rely on small food retailers, it is remiss
of the literature that exists in the field to argue that small retailers would not be unduly impacted by
the proposed rule.

Conclusion
As demonstrated, the NPRM fails to adequately address the disproportionate and significant
impact on small entities across the country. First, significant portions of the Regulatory Flexibility
Analysis were either not completed or completed incorrectly, failing to signal to small entities the
importance of the NPRM. Second, the analysis in the proposed rule incorrectly calculates the
number of impacted small entities, suggesting that small entities would face an impact four times
smaller than the reality of what the NPRM would prescribe. Lastly, both the rural and urban areas
impacted by the rule would see significant losses. Rural small retailers would see disproportionate
losses in sales, while urban small retailers would experience substantial losses in sales. Until FNS
conducts further research regarding the impact of the proposed rule on small entities and the
communities that house these businesses, implementation of this rule would be unwarrantable and
detrimental to those who reside within the impacted areas.

531

Carolyn Cannuscio et al., “The social dynamics of healthy food shopping and store choice in an urban environment,”
Social Science & Medicine, October 2014, pp. 13-20.

229

Appendix A: Center on Budget and Policy Priorities’
Contributors to Our Public Comments
For more than three decades, the Center on Budget and Policy Priorities has been at the forefront
of national and state debates to protect and strengthen programs that reduce poverty and inequality
and increase opportunity for people trying to gain a foothold on the economic ladder. The Center is
a high-caliber strategic policy organization that shapes critical policies for low-income families and
individuals. To these ends, we conduct highly skilled strategic and analytic work to develop and
advance specific, actionable proposals, to achieve the maximum possible policy gains, and to ensure
their effective implementation on the ground. We also build effective partnerships and help a
diverse array of organizations and constituencies to engage more effectively in these debates.
As part of this overall mission, we work to strengthen policies and programs that reduce hunger
and poverty and improve the lives of the nation’s poorest families and individuals. Since our
founding in 1981, we have worked on federal nutrition programs, most notably the Supplemental
Nutrition Assistance Program (SNAP), formerly known as food stamps. We have extensive
experience in and knowledge of SNAP’s three-month time limit and state waiver authority within
that rule. Examples of our efforts on this front include:
• Issuing reports

and analyses on the time limit and the population impacted by the rule, as well
as policy options available to states via waiver and individual exemption policies;

• Assisting states

in assessing which areas of their states are eligible for waivers and developing
waiver requests that meet the federal criteria. The Center has supported an average of 30-40
states each year, completing or assisting with a total of over 600 waiver applications since the
late 1990s;

• Educating anti-hunger

nonprofits and other community organizations about the time limit
rule, including about the availability of waivers based on the underlying unemployment trends
in the state and individual exemptions; and

• Training state

agency officials as well as local anti-hunger and poverty advocates about
judicious practices in implementing the time limit policy. For years we have worked to help
state officials implement the time limit in a way that conforms with federal law and protects as
many individuals with low incomes as possible and avoids (to the degree possible) cutting off
individuals who are not actually subject to the limit.

Over time, the Center’s work on waivers and time limit policy has evolved into a multi-faceted
and complex approach. The Center has provided in-person trainings and direct supervision to state
officials in nearly ten states, while providing support to advocates and working on issues of
implementation for many more. The Center has also given multiple presentations to FNS staff at
several regional meetings across the United States regarding waivers, along with a series of annual
presentations to the American Association of SNAP Directors. Additional support from the Center
has included assistance regarding program integrity and payment accuracy implications, writing of
manuals for dozens of states, and constantly monitoring all 50 states’ statuses and ability to apply for
“ABAWD” waivers. The food assistance team’s experience in combination with the economic
230

expertise of other members of the Center provides CBPP with the greatest amount of experience
and knowledge concerning “ABAWD” waivers and the implications of the proposed rule.
Below is a listing of the individuals at CBPP who contributed to these comments:
Jared Bernstein joined the Center on Budget and Policy Priorities in 2011 as a Senior Fellow.
From 2009 to 2011, Bernstein was the Chief Economist and Economic Adviser to Vice President
Joe Biden, Executive Director of the White House Task Force on the Middle Class, and a member
of President Obama’s economic team. Bernstein’s areas of expertise include federal and state
economic and fiscal policies, income inequality and mobility, trends in employment and earnings,
international comparisons, and the analysis of financial and housing markets.
Prior to joining the Obama Administration, Bernstein was a senior economist and the director of
the Living Standards Program at the Economic Policy Institute in Washington, D.C. Between 1995
and 1996, he held the post of Deputy Chief Economist at the U.S. Department of Labor. Bernstein
holds a Ph.D. in Social Welfare from Columbia University.
He is the author and coauthor of numerous books for both popular and academic audiences,
including Getting Back to Full Employment: A Better Bargain for Working People, Crunch: Why Do I Feel So
Squeezed?, nine editions of The State of Working America, and his latest book, The Reconnection Agenda:
Reuniting Growth and Prosperity. Bernstein has published extensively in various venues, including the
New York Times and Washington Post. He is an on-air commentator for the cable stations CNBC and
MSNBC, contributes to the Washington Post’s PostEverything blog, and hosts On The Economy
(jaredbernsteinblog.com).
Thompson Bertschy received his Master’s in Social Work from the University of North Carolina
and began working with the Center as an intern in 2019. In North Carolina, Thompson worked with
the Carolina Farm Stewardship Association to advocate for sustainable and equitable food policy
while organizing food policy councils across the state. At CBPP, Thompson focuses on research and
analysis regarding “ABAWD” exemptions and state legislative proposals around SNAP, TANF, and
Medicaid.
Ed Bolen joined the Center in 2010 as a Senior Policy Analyst. At the Center, Ed focuses on
SNAP Employment & Training and “ABAWD” waivers. He has provided trainings to multiple
states regarding their loss of waivers, including Alabama, California, Michigan, Illinois, Alaska, and
Mississippi. Additionally, Ed has given presentations to Food and Nutrition Service staff at the
Northeast and Mid-Atlantic regional meetings. Since 2011, Ed has consistently worked with
advocates and state officials concerning implementation issues in Georgia, Illinois, Michigan,
California, and other states. He has also presented to the American Association of SNAP Directors
for four consecutive years and created a toolkit on implementing the “ABAWD” time limit
including technical writing on implementation issues. He has had many interactions with media and
has written multiple blog posts and papers concerning “ABAWDs.” Throughout his time at the
Center, Ed has written multiple pieces concerning Employment and Training for SNAP recipients.
Prior to joining the Center, Bolen was a Senior Policy Analyst at California Food Policy
Advocates. While there, he advocated for administrative and legislative improvements to food
assistance programs and provided training and technical assistance to community-based
organizations. He also has worked in public health law, most recently consulting on legal strategies
231

to combat childhood obesity with the National Policy and Legal Analysis Network. Prior to that,
Bolen was senior staff attorney at the Child Care Law Center, specializing on licensing, subsidy and
legislative issues affecting low-income families in child care and early education settings.
Kathleen Bryant is a Research Assistant in the Federal Fiscal team at the Center on Budget and
Policy Priorities. Previously, she interned for EMILY’s List, the Economic Policy Institute, and the
Center’s Legislative Affairs team. She also conducted independent research on school segregation
for her honors thesis and was selected as a fellow in the Advanced Empirical Research on Politics
for Undergraduates Program for this research by the Society for Political Methodology. Kathleen
has a B.A. in Public Policy from The College of William & Mary.
Ashley Burnside is a Research Assistant with the Center’s Family Income Support division.
Before joining the Center, she was an Emerson National Hunger Fellow with the Congressional
Hunger Center, where she advocated for anti-hunger and anti-poverty solutions, led voter
registration and community engagement efforts for food pantry clients, and conducted research on
tax credit programs. Burnside has also served as a public policy fellow at AIDS United, a national
HIV/AIDS advocacy organization. She holds a B.A. in Social Theory and Practice from the
University of Michigan.
Lexin Cai is a Research Analyst with the Center’s Food Assistance Division, where she focuses
on data analysis for nutrition assistance programs. Lexin first joined the Center in May 2015 as an
intern. Prior to joining the Center, she interned with the World Wildlife Fund, focusing on
international agricultural trade. At the Center, Lexin conducts analyses, compiles historical data and
records, and uses mapping and modeling to support states with the waiver application process.
Over the last several years, Lexin has worked with Catlin Nchako in completing waiver
applications for multiple states and providing assistance with the process to others, helping an
average of 32 states per year. She holds a Master of Science in Social Policy from the University of
Pennsylvania and a Bachelor of Management from Renmin University of China.
Steven Carlson provided background research for these comments as a consultant to CBPP. He
was a federal employee at the Food and Nutrition Service for 37 years. During his time there, he led
their Office of Analysis and Evaluation and oversaw research studies and analysis on SNAP.
Maritzelena Chirinos is a recent graduate of Meredith College and began working with The
Center as an intern in 2019 on The State Fiscal Project Team. She holds a B.A. in Criminology and
Sociology. She previously worked at The Indivisible Project and Democrats For Education Reform.
Stacy Dean is the Vice President for Food Assistance Policy at the Center on Budget and Policy
Priorities. She directs CBPP’s food assistance team, which publishes frequent reports on how
federal nutrition programs affect families and communities and develops policies to improve them.
Dean’s team also works closely with program administrators, policymakers, and non-profit
organizations to improve federal nutrition programs and provide eligible low-income families with
easier access to benefits. She brings her deep programmatic and operational knowledge along with a
strong strategic sense to help advance CBPP’s priorities.
Dean has over 20 years of experience in working in great detail with the USDA and dozens of
states concerning “ABAWD” waivers. In 1997, she began writing and distributing information to a
232

majority states regarding specific Labor Surplus Areas in the state and their ability to apply for a
waiver. She has written multiple papers on “ABAWDs” and waiver requests dating back to the early
2000s. In addition to writing the papers, Dean has worked closely with many states providing
supervision, training, and general policy support for both application and implementation.
In addition to her work on federal nutrition programs, Dean directs CBPP efforts to integrate the
delivery of health and human services programs at the state and local levels. Dean has testified
before Congress and spoken extensively to national and state non-profit groups. She has been
quoted in such publications as the New York Times, Washington Post, Wall Street Journal, and Politico, as
well as the Associated Press. Dean joined CBPP in 1997 as a Senior Policy Analyst working on
national policy issues such as the federal budget, SNAP, and benefits for immigrants.
Previously, at the Office of Management and Budget, Dean served as a budget analyst for food
stamps. In this role, she staffed the White House work on the 1996 food stamp program changes,
reviewed and cleared the ABAWD waiver guidance, and worked on policy development, regulatory
and legislative review, and budgetary process and execution for a variety of income support
programs.
Ife Floyd joined the Center in June 2011 as a Research Associate with the Family Income
Support Division, and is now Senior Policy Analyst.
Prior to joining the Center, Floyd served as an AmeriCorp VISTA with Culture Connect, Inc. and
developed a cultural competency workshop series for professional workplaces in the Atlanta area. In
addition, she also worked with the Atlanta Community Food Bank’s Prosperity Campaign as a
benefits screener to increase access to programs like SNAP, TANF, and Medicaid for low-income
families. Floyd holds a B.A. in Sociology from Northwestern University and a M.P.P. degree from
Georgia State University.
Brynne Keith-Jennings obtained her Master’s in Public Policy from the University of Southern
California and joined the Center in June 2011. Her work focuses on federal and state SNAP policies
and research. As a Senior Policy Analyst, Brynne works closely with the USDA and regularly offers
guidance with ABAWD waiver policy. She provides states with technical assistance regarding waiver
requests and has developed a suite of products around the farm bill, SNAP, and employment. Since
2013, Keith-Jennings has been one of the primary food assistance team members concerned with
“ABAWD” waivers, assisting with the process of completing or assisting with 30-40 waivers each
year.
Prior to joining the Center, she worked as an educator and policy analyst for Witness for Peace
and as a consultant for other NGOs in Nicaragua. She has also worked at the Public Welfare
Foundation supporting human rights and criminal justice reform organizations, at the Government
Accountability Office, and at the Tomas Rivera Policy Institute, where she analyzed language access
policies and other issues affecting Latino communities in Southern California.
Joseph Llobrera is a Senior Policy Analyst on the Food Assistance team. As a research associate
at the Center between 2002 and 2007, Llobrera supported the Food Assistance, State Fiscal, and
Housing Policy teams. During these years, Joseph managed the Center’s “ABAWD” waiver support
process through data analysis and extraction, assisting nearly 150 states with the waiver process. He
also interacted with and provided operational support to state advocates and agencies. Before
233

returning to the Center in 2019, he served as an Associate Director of Learning and Improvement at
Insight Policy Research, providing technical assistance and training to federal, state, and local human
service agencies that administer SNAP and the Temporary Assistance for Needy Families program.
During this time, Llobrera reviewed notices that states would send to SNAP recipients who were
subject to the time limit. He also worked as a researcher at IMPAQ International and the Urban
Institute, focusing on food assistance policy, workforce development, and health policy.
Currently at the Center, Llobrera provides operational support and technical assistance to
advocates and state agencies in an effort to streamline the SNAP and Medicaid enrollment process.
Llobrera holds a Ph.D. in Nutrition from the Friedman School of Nutrition Science and Policy at
Tufts University, a master’s degree in Geography from the University of Washington (Seattle), and a
bachelor’s degree in Mathematics and Urban Studies from Brown University.
Rachel Mbassa is an intern with the Data team. She graduated with a Master of Public Health in
Biostatistics from Loma Linda University. She then went on to work at the University of California
San Francisco Medical Center, conducting epidemiological/clinical research centered on the effects
of cumulative psychosocial stressors on cardiovascular disease outcomes. She recently moved to
Washington, D.C. to transition into a career in policy data analysis. She has a strong interest in
health economics and policy and hopes to pursue a terminal degree in this field.
Tazra Mitchell joined the Center in 2016 as a Policy Analyst in the Family Income Support
Division. Previously she worked as a State Policy Fellow and Policy Analyst with the North
Carolina Budget & Tax Center, where she conducted analysis of work and income supports as well
as fiscal and economic policies that enable low-income people and disadvantaged communities to
thrive. In addition, she worked as a Research Assistant in the non-partisan Fiscal Research Division
of the North Carolina General Assembly where she analyzed legislative proposals to determine the
fiscal impact on state government resources and worked directly with legislators to develop the state
budget.
Mitchell holds a B.A. in Political Science from North Carolina State University and an M.P.P.
from the Sanford School of Public Policy at Duke University.
Catlin Nchako joined the Food Assistance Division as a Research Associate in November 2013.
His work focuses on data analysis and research for the food stamp, school meals, and WIC
programs. Over the last several years, Nchako has worked with Lexin Cai in completing waiver
applications for multiple states and aiding with the process to others, helping an average of 32 states
per year. He has also tracked employment benefit trigger notices for all states and territories,
notifying areas when they qualify for an exemption. He has completed multiple analyses and
contributed to many of the Center’s papers concerning “ABAWDs.”
Nchako worked for the Center as a Food Assistance intern prior to joining the organization on a
full-time basis. He previously interned for the Center for Law and Social Policy. He also worked
for several years as a labor researcher for the United Food and Commercial Workers Union, where
he evaluated wage proposals during labor contract negotiations, analyzed companies’ financial
performance, and provided campaign research support. He holds a Master’s in Public Policy from
Georgetown University and a B.A. in Africana Studies from Cornell University.

234

LaDonna Pavetti is the Vice President for Family Income Support Policy at the Center on
Budget and Policy Priorities. In this capacity, she oversees the Center’s work analyzing poverty
trends and assessing the nation’s income support programs, including the Temporary Assistance for
Needy Families (TANF) program. For the last several years, she has been working on a special
initiative to identify opportunities to build executive function and self-regulation skills in TANF and
other work programs.
Before joining the Center in 2009, Dr. Pavetti spent 12 years as a researcher at Mathematica Policy
Research, Inc., where she directed numerous research projects examining various aspects of TANF
implementation and strategies to address the needs of the hard-to-employ. She has also served as a
researcher at the Urban Institute, a consultant to the U.S. Department of Health and Human
Services on welfare reform issues, and a policy analyst for the District of Columbia’s Commission
on Social Services. Across these positions, Dr. Pavetti conducted research on how states were
implementing TANF, how sanctions were implemented, and conducted a study in St. Paul,
Minnesota analyzing which populations were primarily impacted by TANF work requirements. In
recent years, she has worked with Vermont, New York, Minnesota, Colorado, California,
Connecticut, New Jersey, Pennsylvania, and Oregon.
In addition, for several years she was a social worker in Chicago and Washington, D.C. Dr. Pavetti
has an A.M. in social work from the University of Chicago and a Ph.D. in public policy from
Harvard University’s Kennedy School of Government.
Dottie Rosenbaum is a Senior Fellow who joined the Center in 2000. She has worked
extensively on federal and state issues in SNAP as well as issues that involve the coordination of
SNAP and other state-administered health and income security programs, such as Medicaid, TANF,
and child care. In addition, Rosenbaum has expertise on the federal budget and budget process.
With over more than 20 years of experience with SNAP, she written multiple papers regarding
SNAP and SNAP Employment and Training.
Before joining the Center, Rosenbaum was a budget analyst at the Congressional Budget Office.
In this role, she conducted the cost estimate of the 1996 Personal Responsibility and Work
Opportunity Act which included the time limit. She projected federal spending and provided
Congress with cost estimates for a variety of programs including: SNAP, Medicaid, the State
Children's Health Insurance Program, Child Nutrition, and Elementary and Secondary Education.
Rosenbaum holds a Master’s in Public Policy from Harvard University's Kennedy School of
Government.
Andrew Scanlan is a recent graduate of Brown University and began working with the Center as
an intern in 2019 on the Family Income Security team. At Brown, he worked as a research assistant
in the political science department and received a combined degree in philosophy and mechanical
engineering. He previously worked in the Rhode Island state government and the New York City
government.
Arloc Sherman’s work focuses on family income trends, income support policies, and the causes
and consequences of poverty. He has written extensively about the effectiveness of government
poverty-reduction policies, the influence of economic security programs on children’s healthy
development, the depth of poverty, tax policy for low-income families, welfare reform, economic
inequality, material hardship, parental employment, and the special challenges affecting rural areas.
235

He is a member of the National Academy of Sciences Committee on National Statistics Panel to
Review and Evaluate the 2014 Survey of Income and Program Participation's Content and Design.
Prior to joining the Center in 2004, Sherman worked for 14 years at the Children’s Defense Fund
and was previously at the Center for Law and Social Policy. His book Wasting America’s Future was
nominated for the 1994 Robert F. Kennedy Book Award.
Chad Stone is Chief Economist at the Center on Budget and Policy Priorities, where he
specializes in the economic analysis of budget and policy issues. Stone has conducted notable
research around measures of economic slack that are indicative of a weaker labor market, primarily
concerning labor force participation and the employment population ratio. He was the acting
executive director of the Joint Economic Committee of the Congress in 2007 and before that staff
director and chief economist for the Democratic staff of the committee from 2002 to 2006. He was
chief economist for the Senate Budget Committee in 2001-02 and a senior economist and then chief
economist at the President’s Council of Economic Advisers from 1996 to 2001.
Stone has been a senior researcher at the Urban Institute and taught for several years at
Swarthmore College. His other congressional experience includes two previous stints with the Joint
Economic Committee and a year as chief economist at the House Science Committee. He has also
worked at the Federal Trade Commission, the Federal Communications Commission, and the
Office of Management and Budget. Stone is co-author, with Isabel Sawhill, of Economic Policy in the
Reagan Years. He holds a B.A. from Swarthmore College and a Ph.D. in economics from Yale
University.
Jennifer Wagner joined the Center in 2015 as a Senior Policy Analyst with the health team. She
focuses primarily on Medicaid eligibility and implementation of the Affordable Care Act (ACA),
including analyzing opportunities to align Medicaid with other low-income support programs for
greater access and efficiency.
Before joining the Center, Wagner served for five years as an Associate Director with the Illinois
Department of Human Services. In that capacity, she oversaw SNAP and cash assistance policy as
well as the local offices throughout the state that determined eligibility for cash, SNAP, and medical
assistance. She assisted Illinois with Medicaid expansion under the ACA and improved customer
service through business process re-engineering in the local offices. Prior to that, she was a staff
attorney at the Sargent Shriver National Center on Poverty Law, where she focused on public
benefits.
Wagner received her B.S. from the University of Wisconsin – Madison and her J.D. from the
Northwestern University School of Law.

236

Appendix B: Materials Cited in Comments
Appendix B includes all of the resources and materials cited in these comments to help ensure
that the Department will have complete and simple access to the relevant research. The documents
in Appendix B are listed in alphabetical order by first author’s last name or entity where appropriate.
In the electronic submission version of our comments, Appendix B is broken into 32 documents
due to the file size restrictions on the federal register website.
This list indicates which materials (by author or entity) are included in each electronic file.
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.

Attachment B1 (1 - BLS)
Attachment B2 (BLS - BRE)
Attachment B3 (BUT - CEN)
Attachment B4 (CEN - CHE)
Attachment B5 (CHE - DRA)
Attachment B6 (ECO)
Attachment B7 (ECO)
Attachment B8 (EDI-FED)
Attachment B9 (FOG)
Attachment B10 (FOG)
Attachment B11 (FOO-GOR)
Attachment B12 (GRA-HEA)
Attachment B13 (HER)
Attachment B14 (HER)
Attachment B15 (HER-HOY)
Attachment B16 (ING-KAL)
Attachment B17 (KEI-LAS)
Attachment B18 (LEE-LOP)
Attachment B19 (MAB-MAR)
Attachment B20 (MAR-MCH)
Attachment B21 (MIL-OGG)
Attachment B22 (OHI-POW)
Attachment B23 (QUI-RHO)
Attachment B24 (RIB-ROS)
Attachment B25 (ROS-SCH)
Attachment B26 (SCH-STO)
Attachment B27 (TEN-USD)
Attachment B28 (USD-USD)
Attachment B29 (USD-VIR)
Attachment B30 (VOR-WU)
Attachment B31 (WU)
Attachment B32 (YAG-ZIL)

237

UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF COLUMBIA
DISTRICT OF COLUMBIA, STATE
OF NEW YORK, STATE OF
CALIFORNIA, STATE OF
CONNECTICUT, STATE OF
MARYLAND, COMMONWEALTH
OF MASSACHUSETTS, ATTORNEY
GENERAL DANA NESSEL ON
BEHALF OF THE PEOPLE OF
MICHIGAN, STATE OF
MINNESOTA, STATE OF NEVADA,
STATE OF NEW JERSEY, STATE
OF OREGON, COMMONWEALTH
OF PENNSYLVANIA, STATE OF
RHODE ISLAND, STATE OF
VERMONT, COMMONWEALTH OF
VIRGINIA, and CITY OF NEW
YORK,

Civ. Action No.

Plaintiffs,
v.
U.S. DEPARTMENT OF
AGRICULTURE; GEORGE ERVIN
PERDUE III, in his official capacity as
Secretary of the U.S. Department of
Agriculture, and UNITED STATES
OF AMERICA,
Defendants.
EXHIBIT B TO DECLARATION OF EDWARD BOLEN IN SUPPORT OF
PLAINTIFFS’ MOTION FOR PRELIMINARY INJUNCTION

Comparison of CBPP Comments on Proposed Rule on Time Limit Waivers to Preamble and Final Rule
Dec. 13, 2019

CBPP has done an initial review to determine if the preamble and final rule restricting waivers of the three-month time limit address the
comments CBPP made to the proposed rule. Our intent was to identify claims we make that are not adequately addressed in the final rule,
suggesting that the agency overlooked relevant public comment.
This is structured as follows. For each chapter of our comments, we identified the arguments, claims and research findings we made, noted the
page number, and then indicated where (if at all) the Final Rule discusses the comment. We characterized the agency response as either no
mention in the preamble or rule, just summarized but not directly responded to, or discussed but not sufficiently. We highlight some of the key
shortcomings of the proposed rule for most, but not all, of the chapters.
Our cites to the Final Rule are to the pre-publication, double-spaced version.

CBPP Comments Section 2: FNS Waiver Policy Has Been Consistent for the Last 22 Years
Key issues
•

•

The preamble includes at least 10 references to its “operational experience” (sometimes “20 years of operational experience”), listed below,
to defend both the rulemaking itself and specific provisions of the rule. The preamble does not give any more information to explain its
understanding of Congressional intent in PRWORA and where the Department believes interpretation of the law has veered away from that
intent. The preamble also states that the core standards will provide “consistency” but does not explain if they believe that current rules allow
for inconsistent waivers, though they no longer state the belief that the rule will improve or increase consistency.
When the Department cites this experience in the preamble, it is usually to explain how it has observed that current regulations allow states
to qualify for waivers (or use exemptions) in ways that are inconsistent with PRWORA based on subjective and unspoken criteria. For
example, they state that implementing an unemployment rate floor will prevent areas that “that do not clearly lack sufficient jobs” from
qualifying, without explaining what criteria they are using to judge whether an area “clearly” lacks sufficient jobs. Similarly, the preamble
argues that the current flexibility around grouping together geographic areas gives states “too much flexibility”, without explaining on what
basis they are judging what level of flexibility is appropriate.

Page of
CBPP said:
our
comments

Page of
USDA responded:
their
response
1

Other Notes

19-21

22-24

Proposed rule implied that waiver practice is a
departure from original intent of statute, but
waiver standards have been consistent for 22
years. Department did not give any evidence
such as legislative history to explain why they
think 1996 guidance strayed from legislative
intent, nor can they point to any major
changes since then, including in the most
recent 2016 guidance.
Current waiver policy provides consistency for
states and the Department did not explain
why they thought more consistency was
needed. (The preamble to the NPRM had
stated it would “improve consistency across
states”, but does not explain what the
Department thinks is currently inconsistent.)
Only inconsistency is from new Administration
that began denying waivers that were long
approved or delaying.

100-101

Did not respond to this argument, and instead relied on its “more than
20 years of operational experience overseeing the States’
administration of the ABAWD time limit” (p.7) as justification for the
rule itself and for specific provisions. The Department does not explain
specifically if they think any particular guidance (1996 on) departed
from Congressional intent or provide any more evidence of what they
believe Congressional intent is.

77

The preamble did not address this argument, and in fact stated that the
Department “has been committed to responding to waiver requests
prior to the State’s requested implementation date and has met this
commitment consistently” (in regards to the implementation before
approval provision, p.77). The preamble no longer makes reference to
“improving” or increasing consistency, however.

CBPP Comments Section 3: Setting a Floor for Waivers 20% Above National Unemployment is Inconsistent with Congressional
Intent and would be Harmful to Vulnerable Individuals
•

•

The preamble to the final rule acknowledges but does not meaningfully respond to comments explaining why an unemployment rate floor
(for a waiver demonstrating an area has an unemployment rate 20 percent above the national average for a 2-year period) would not provide
an accurate picture of whether there are a sufficient number of jobs in a given area to employ the individuals subject to the time limit. These
arguments discussed extensive research demonstrating that the individuals subject to the time limit face more barriers to employment than
the general population and therefore that the overall area unemployment rate is a poor indicator of available jobs for this group. The
preamble briefly summarizes most of these comments, but dismisses them all in a single paragraph that the Administration is “resolute” that
setting an unemployment rate floor will “accurately” reflect available jobs and to prevent areas with “low” unemployment rates from
qualifying, without explaining how they came to this conclusion.
The Department acknowledges arguments made against the 7 percent unemployment floor in the proposed rule and states that they took
these into account in changing the unemployment rate floor to 6 percent in the final rule, but does not acknowledge nor respond to

2

•

arguments made against the 6 percent floor. For example, we cited evidence that some subgroups face very high unemployment rates with an
overall area unemployment rate of 6 percent.
This section raises concerns about how the proposed waiver criteria would protect states during an economic recession, which the preamble
to the final rule only partially acknowledges. The preamble acknowledges these arguments in the context of the proposed 7 percent
unemployment floor, which they changed in the final rule to 6 percent. The preamble does not otherwise discuss how requiring a 2-year
unemployment rate of at least 6 percent would prove difficult for states during a time of rapidly rising unemployment, which is especially
significant given that the final rule will eliminate waivers based on qualifying for extended unemployment insurance benefits, which would
also suffer from the same problem. The final rule preamble also doesn’t address the general point made in our comments about the
importance of understanding the impact of the rule throughout the business cycle (of peaks and troughs in unemployment).

Page of
CBPP said:
our
comments
27
Congress intended for “insufficient jobs” to
be a separate criterion than the
unemployment rate
28
Current 20 percent has limitations because
unemployment rates don’t reflect jobs
available for this population; this proposed
rule would worsen problem by removing any
flexibility.

29

One of the reasons tying to the national UR
instead of unemployment rate floor is
beneficial is because of declining labor force

Page of
USDA responded:
their
response
26
Acknowledged comment, but argued that
statute gives Secretary discretion to define
“lack of sufficient jobs”
28
Acknowledged arguments but didn’t respond
in depth (“some commenters noted that
unemployment relative to the national
average is an important signal that the
economic conditions warrant waiving work
requirements….These commenters argued
that adding an unemployment rate floor to
the 20 percent standard provides less
flexibility for States to capture insufficient
jobs for the ABAWD population. The
Department appreciates this information
provided by commenters, but disagrees that a
relative unemployment rate is a sufficient
indicator of a lack of sufficient jobs in and of
itself.”)
Didn’t respond to this specific point but did in
response to arguments about the E:P ratio
criterion (“Commenters pointed to the fact

3

Other Notes

participation nationally, which makes
unemployment rate less helpful

29

30-31

32

Department is proposing to eliminate other
non-UR means of demonstrating
“insufficient jobs”, which would mean
unemployment rates are only way to show
that, and it’s taking away flexibility by
imposing a floor. Also moving farther from
congressional intent to have “insufficient
jobs” be a broad metric—Congress regularly
legislates unemployment rates for policy
purposes and chose not to do so here.
House proposed similar measures in Farm
Bill as 7 percent floor, and rejected it,
showing how this goes against Congress

26

Because ABAWDs face more barriers to
work than general population,
unemployment rate floor will overstate jobs
available

that labor market participation has not
recovered since the Great Recession, even
though unemployment rates have. “); also
summarized general arguments that a
relative rate is sufficient
Not sure they responded to this specific
point. They did discuss Congressional intent:
“Commenters also suggested that, if Congress
intended to include an unemployment rate
threshold for the ‘sufficient number of jobs’
criteria, Congress would have done so.”

21-24

18

Mentioned this criticism but didn’t respond
specifically to this point. “Some commenters
pointed to the justification provided in the
proposed rule that a 7 percent floor aligns
with a proposal in the [Farm Bill]. These
commenters argued that this rationale is
invalid because Congress ultimately did not
include that provision…”
“Many commenters opposed setting an
unemployment floor because they argued
unemployment rates fail to accurately
capture the availability of jobs specifically for
ABAWDs who face particular barriers to
employment…notwithstanding the issues
raised by these comments, the Department is
resolute that establishing an unemployment
rate floor within the 20 percent standard is
necessary to ensure that the standard is

4

Ultimately rejected 7
percent floor citing
other evidence

USDA fails to explain
its definition for “lack
of jobs,” so it is
unclear why an
unemployment floor
would “accurately”
reflect a lack of
sufficient jobs.

33-35

SNAP participants without dependents have
low levels of education and research shows
often lack skills; workers with low
attainment have higher unemployment
rates and are more likely to lose jobs during
recession

22/24

36-38

Childless adults disproportionately Black and
Latino; have higher unemployment rates
even when controlling for education;
disparities exist at local level; affected more
by recessions; lots of studies demonstrate
racial discrimination. Native Americans also
have higher rates of unemployment.

23/24

39-41

Childless adults and other SNAP participants
work in jobs with volatility and experience
spells of unemployment; unemployment
rates don’t capture those trends.

23/24

42-44

designed to accurately reflect a lack of
sufficient jobs in a given area”
They acknowledge the comment in general
but again dismiss in same paragraph on p. 24
(“notwithstanding the issues raised…”), They
don’t explain how given evidence for workers
with low levels of education, they still think
it’s appropriate to use an unemployment rate
to measure “insufficient jobs”.
Mentioned but dismissed with same
paragraph as above (“notwithstanding the
issues raised…” p.24). Again, this paragraph
didn’t address specific argument—it didn’t
explain why given this evidence they still
believe that an unemployment rate reflects
available jobs for ABAWDs. For example,
there is literature showing how full
employment narrows racial disparities in
unemployment.
“commenters suggested that ABAWDs are
more likely to have part-time work…”; again
dismissed in paragraph on p. 24 but did not
explain why they think their interpretation of
“insufficient jobs” is appropriate given this
info
They mentioned general barriers (e.g. “Many
commenters opposed setting an
unemployment floor because they argued
unemployment rates fail to accurately
capture the availability of jobs specifically for
ABAWDs who face particular barriers to
employment” P.22) but did not specifically
mention this barrier

Many ABAWDs have health conditions that
limit their ability to work which are not
reflected in unemployment rates

5

44-46

47-48

50-52

56-63

Many childless adults face geographic or
23/24
transportation limitations to work; people in
rural areas have lower employment rates;
unemployment rate may not reflect whether
individuals can get to available jobs;
research shows that increasing job
accessibility increases employment
Other barriers including housing instability
23/24
and homelessness and criminal records
make it harder for some ABAWDs to find
jobs even with low employment, meaning
unemployment rate poor indicator of
available jobs
Rule uses LSA as justification for
22
unemployment rate floor, but those are for
different purposes; waivers show lack of
jobs for specific ABAWD population and LSAs
are focused on macroeconomic conditions.
Doesn’t make sense to have a floor, if they
have it

Evidence suggests that with unemployment
rates specifically at 7 percent (or 6 or 5
percent), some groups such as people of
color face substantially higher
unemployment rates

Very limited mention (but a bit more in LMA
section); did not specifically address why they
think unemployment rate floor is still
appropriate given that workers may not have
access to available jobs
Mentioned but did not address specifically
(other than the “notwithstanding the issues
raised in these comments…”

Mentioned somewhat (“Commenters argued
that ABAWDs should not be subject to the
unemployment rate floor used in designating
LSAs because ABAWDs face labor market
disadvantages that the general public does
not.”) but they don’t address whether they
differently interpret the “insufficient jobs”
criterion to mean whether there are enough
jobs so that ABAWDs can get a job vs. the
general population.
“Some commenters asserted that jobs are not
widely available to all who may seek them
when unemployment is below 7
percent…these commenters noted that
unemployment rates for ABAWDs, as a
distinct group, would generally be higher than
the official unemployment rate because many
ABAWDs share demographic characteristics
with subpopulations that have relatively high
unemployment rates.”

18-19

6

Didn’t acknowledge
that these analyses
found similar issues
with 6 percent/5
percent

64-65

Many “distressed” communities have
unemployment rates below 7 percent

65

Underemployment rates are higher for black
and Hispanic workers than whites at 5,7, and
10% unemployment; “the rule suggests that
SNAP waivers should be disallowed in places
where about a fifth of black and Latino
workers could be un- or underemployed”
Requiring a 2-year unemployment rate of 7
percent would result in a gap between when
unemployment starts to climb and a state
qualifies for extended unemployment
benefits (EB), though the final rule
eliminates EB as a criterion for a waiver.

66-67

66

“We agree with the Department’s proposal
to continue to allow states to request
waivers when they qualify for EB, as these
are times when unemployment rates are
high and rising and individuals likely have
difficulty finding jobs.”

19

“One commenter provided data analysis
showing that many areas considered
“distressed communities” according to a
series of economic metrics would not have
met the 7 percent unemployment rate
threshold. This commenter argued that the 7
percent floor fails to capture the economic
realities of regions, and that this divergence
highlights the shortcomings of a 7 percent
unemployment rate floor.
Don’t think they addressed this specific point.

19

“Commenters also suggested that the
proposed 7 percent floor would not
adequately provide States with waiver
coverage during times of rising
unemployment because the combination of
an unemployment rate floor with the lengthy
24-month data reference period would
prevent many areas with rising
unemployment from qualifying for waivers.”

34

“Although the Department did not receive
many comments with regard to retaining the
extended unemployment benefits standard,
some commenters supported the proposal to
retain the extended unemployment benefits
standard, arguing that this standard is an
appropriate indicator that a State lacks
sufficient jobs….Although the Department

7

USDA overlooks
implications of
eliminating EB as
criterion, despite
lowering unemployment
floor to 6 percent. Not
having EB as a criterion
at all is much worse than
having a gap before it
would be triggered.

67

68

69-75

The Department’s rationale does not take
into account the fact that unemployment
rates can vary across the country and even
with low unemployment rates, individuals
subject to the time limit may lack available
jobs. Also, “the Department did not indicate
whether it considered the effect of the
proposed rule at different parts of the
business cycle, such as entering into a
recession, and how climbing unemployment
rates affect job availability.”
The Department suggests that the 7 percent
floor is “designed specifically for ABAWD
waivers” but all available evidence suggests
that if anything a floor for ABAWDs would
be lower than the floor used for LSAs

43

Department stated its goal is to minimize
waiver coverage, when statute says it should
be to approve waivers in areas with
insufficient jobs; doesn’t explain what
criteria they’re judging appropriate waiver
coverage; doesn’t explain whether this goal
applies during a recession; calculation of

appreciates these comments in support of
the criterion, the Department has decided
not to adopt the rule as proposed because
the Department is concerned that the
extended unemployment benefits criterion
would allow States to receive statewide
waivers even when there is not a lack of
sufficient jobs within certain areas of the
State.”
Final rule says they can approve waivers in
“exceptional circumstances” and they give an
example of rising unemployment, but does
not address whether they’ve considered the
importance of waivers to provide support
during recessions. Given they’re taking away
EB criteria, does not seem like they have.

26

25

“As discussed in the previous section, the
Department finds the 20 percent standard
with a 6 percent floor to be one of the most
objective and defensible ways of determining
a lack of sufficient jobs, as it aligns with a
longstanding DOL measure of job
insufficiency.”
Mentioned this criticism (“Commenters
stated that Congress did not intend for lack of
sufficient jobs criteria to be based on whether
there are too many or too few waivers that
result from the criteria – Congress did not
establish a desired level of waiver
coverage….) but reiterate they want an

8

USDA decided to align
with LSA standards and
failed to address the
argument that floor
should be lower than LSA
standards if there is one
In this preamble, USDA
does not emphasize as
much how the rule will
limit waiver coverage as
they did in the proposed
rule.

“ABAWDs living in waived area” doesn’t
make sense.

76

6 percent unemployment floor is more
consistent with LSAs, which makes their
choice of 7 percent floor even more
arbitrary since they did not give any
evidence to support it, but “would still
exclude many areas where childless adult
SNAP participants face considerably higher
unemployment or underemployment rates
and where they will not have access to jobs”

unemployment rate floor, now framed in
terms of preventing “the misapplication of
waivers to areas with unemployment rates
that do not demonstrate a lack of sufficient
jobs”.
Did not address arguments against 6 percent
floor; but did include comment about it being
more consistent with LSA (p. 15) and other
positive arguments. They also addressed
arguments about not having any floor
(“notwithstanding the issues raised in these
comments…”)

15/24

CBPP Comments Section 4: Dropping Several Key Criteria From Waiver Criteria Is Inconsistent With the Statute
USDA summarizes comments discussion the shortcomings of the general unemployment rate but argues for the use of the
measure because of the availability of standardized data at the substate level. The Department seems to be upholding
availability/consistency of data measures as its main reason for selecting criteria rather than how well the measure(s)
empirically reflect the lack of jobs for individuals subject to the time limit. USDA does not explain why the general
unemployment rate is the only measure it will accept.
Page of
CBPP said:
our
comments
81-82
Long legislative history (statute,
regulations, and guidance) acknowledging
that there is no perfect measure of an
area’s “lack of sufficient jobs”

Page of
USDA responded:
their
response
N/A
No mention.

9

Other Notes

84-85

Other measures, such as the employmentto-population ratio and U-6 measure,
provide important information on labor
market conditions

31

Cursory response.
“While these comments about alternative
unemployment measures are appreciated, the
Department also recognizes that there is not
measure available for precisely determining
the number of available jobs specifically for
SNAP ABAWDs in a given area.”

87-90

General unemployment rate may not
adequately measure weak labor demand
at the state and sub-state level
Information about declining occupations
or industries can help identify smaller
areas experiencing a lack of sufficient jobs

N/A

No mention.

39-40

Eliminating historical seasonal
unemployment rate over 10% is
inconsistent with statute

79

Cursory response.
“However, information about declining
industries or occupations, and academic
studies are not as standardized and reliable as
unemployment data…”
“…the arguments made by commenters were
not sufficiently compelling…”
Cursory response.
“Despite these comments, the Department is
maintaining the elimination of the historical
seasonal unemployment criterion as
proposed. The Department believes that the
historical seasonal unemployment criterion
was not appropriate, as an area could receive
a waiver for up to 12 months, even though it
only demonstrated a few months of high
unemployment per year.”

91

92

10

USDA dismisses
alternative
measures by
focusing on the
lack of sub-state
data for one
particular
measure (U-6).
Other measures
(employment-topopulation ratio)
can be calculated
for sub-state
areas

USDA does not
acknowledge its
guidance that
allows for
waivers of a
shorter duration
to align with
seasonal high
unemployment.

CBPP Comments Section 5: Restricting State Flexibility on Grouping Areas Is Counter to Evidence
USDA asserts its authority to narrowly define the concept of “area,” using DOL’s Labor Market Area as the only acceptable
framework for grouping substate areas. The language in the preamble suggests that the Department is adopting a strict and
restrictive definition of “area” because it is philosophically against using grouping of substate areas to reduce the number of
areas that are subject to the time limit (for example, see p. 59-60) and that states has historically “misused” their flexibility.
USDA does not explain why it is eliminating the extended unemployment benefit criterion in this final rule, even though it
had not proposed this in the proposed rule.
Page of
CBPP said:
our
comments
95-98
2 decades of legislative history of state
discretion in defining areas
99-101,
States may use grouping to align with
108-109
SNAP E&T, WIOA, and other regional
factors
102-103
104-105

106

108

Arbitrary decision to eliminate statewide
waivers, except for extended
unemployment benefits
USDA is proposing to use one specific and
narrow definition of labor market area
(based on aggregated commuting patterns
between counties)
DOL/BLS Labor Market Area definition
may be outdated

Commuting patterns are not the only
factor connecting labor market areas

Page of
USDA responded:
their
response
51-52
Summarized, but no direct response.
“The Department is within its authority…”
55
Summarized, but no direct response.
“The Department is not compelled by
arguments that E&T services or other work
program availability should be factored in…”
34-35,
Final is worse than proposed.
64
“… the Department has decided not to adopt
the rule as proposed…”
49
Summarized, but no direct response.
“The Department is not compelled by the
commenters’ suggestions described in the
preceding paragraphs…”
54
Cursory response.
“However, after assessing alternative options,
the Department has not identified any other
labor market definition that uses more recent
data…”
49
Summarized, but no direct response.

11

Other Notes

USDA does not
need to restrict
grouping to the
concept of labor
market areas

110

Requiring areas to be contiguous ignores
reality that proximity to jobs is decreasing

49

110-111

Requiring areas to be contiguous ignores
differential rise/fall in unemployment
rates in neighboring areas

N/A

“The Department is not compelled by the
commenters’ suggestions described in the
preceding paragraphs…”
Summarized, but no direct response.
“The Department is not compelled by the
commenters’ suggestions described in the
preceding paragraphs…”
No mention.

CBPP Comments Chapter 6: Taking away food benefits will not increase labor force participation, despite being the stated
goal of the proposed change
The Final Rule summarizes comments that included research supporting this claim, but generally did not respond, asserting instead that
too high a percentage of ABAWDs live in waived areas. We didn’t directly address this claim, but it, too, is misleading. ABAWDs in areas
without waivers are cut off the program so we would expect most ABAWDs on SNAP to be in waived areas regardless of how many
waived areas there are.
Page of
CBPP said:
our
comments
112
NPRM offers a belief that time limit
increases work, but no evidence that it
does
112

ABAWDs already have significant work
effort, with no evidence that the work
rate, job placement and earnings can be
increased by withholding food.
-NPRM provides no information about
existing work among ABAWDs, misleading

Page of
USDA responded:
their
response
98 and
Summarize but did not respond.
others
Commentators “very critical of the
Department’s assertion that a broader
application of time limits . . . would help
adults find work.”
99
Summarized but did not respond.

12

Other Notes

115
115

readers about their labor market
attachment.
NPRM fails to account for the barriers
facing ABAWD population
General unemployment rate does not
reflect the job prospects for this
population

99

Summarized but did not respond.

25

118-120

USDA’s own studies of people leaving
SNAP show time limit did not increase
employment.

101

120-121

Studies show loss of SNAP due to time
limit increases hardship
Time limits in other programs do not
increase employment and have a
disproportionate impact on certain
populations
Time limits and sanctions have disparate
impact on communities of color

N/A

Doesn’t directly address, but claims that the
new standard is “one of the most objective
and defensible ways” of determining a lack of
sufficient jobs.
Summarized but no direct response. Cites 20
years of implementation and that current
regulations lack “certain important
limitations” to protect against misapplication
of waivers. Department believes that who
can should work or seek work, though
“acknowledges that the rule does not in and
of itself provide ABAWDs with additional job
opportunities.”
Did not directly summarize or respond.

98

Summarized but did not respond.

108

Did not discuss but in Civil Rights Impact
Analysis section notes it did a comprehensive
analysis and that monitoring and mitigation
strategies by the FNS Civil Rights Division
“may lessen these impacts.”

122

126

CBPP Comments Section 8: Proposed Rule Would Make Implementing Time Limit Harder by Removing Provisions that Give States
Certainty Around Approval

13

•

•

The preamble to the final rule does not respond to comments that, in discussing proposed changes such as taking away the “readily
approvable” language and not allowing states to implement the waiver prior to approval, highlight the need for the Department to commit to
approving waivers in a timely manner. Our comments gave an example of a lengthy delay for approval for a California waiver, which the
Department does not acknowledge and instead states the Department has consistently approved waivers on time.
The preamble acknowledged comments critical of the proposal to shorten waivers to one year or less, but did not substantially respond to
these arguments. For example, our comments explain that current guidelines require very recent unemployment data for 2-year waiver
approval, but in explaining why the final rule limits waivers to one year or less, the Department still suggests that a shorter waiver approval is
necessary to ensure “the waiver request reflects recent economic conditions”.

Page of
our
comments
144

146

147-148

CBPP said:
While the preamble therefore suggests that
states may understand that waivers requested
under the “core standards” can reasonably be
expected to be approved (provided they
include the correct data and are calculated
accurately), the actual rule lacks the specificity
of the “readily approvable” language in
current regulations at 7 C.F.R. § 273.24(f)(3).
FNS does not make any commitment to
approving waivers in a timely fashion, but
eliminating “readily approvable” language and
ability to implement prior to approval will give
states more uncertainty. Gave example of CA
waiting over 5 months for 2018 waiver
approval and implementing prior to approval.

Proposed implementation date of Oct. 1
would present severe burden for states and
they do not justify such a short timeline

Page of
their
response

76-77

USDA responded:
Didn’t clarify whether core standards are the
same as “readily approvable”; regulation
language states “Core standards. FNS will
approve waiver requests…”

It states this is current practice and does not
include any new language to guarantee FNS
approval in a timely fashion (“Because the
Department has been committed to responding
to waiver requests prior to the State’s requested
implementation date, and has met this
commitment consistently…). Also notes waivers
don’t “waive States’ responsibility to identify
ABAWDs (screen household members for the
exceptions from the time limit at § 273.24(c)) or
to measure and track the 36-month period”
Responded to concerns about time needed for
implementation and changed date to April 1,

82-84

14

Other Notes

150-152

Existing requirements for 2-year waivers
already require states to demonstrate high,
sustained unemployment; reflect “current”
conditions because states use data that ends 3
months prior to waiver implementation; have
been used sparingly, and mostly during the
Great Recession, which helps states’ manage
administrative costs

2020, but did not explain why they included such
a short timeline to begin with
Acknowledged comments but didn’t respond to
arguments in them regarding the fact that states
must use current data and allowing 2-year
waivers reduces burden for areas with sustained
high unemployment, stating “The Department
believes that a 1-year waiver term allows
sufficient predictability for States to plan and
implement the waiver. At the same time, a 1year waiver term ensures that the waiver request
reflects recent economic conditions.”

68

CBPP Comments Chapter 9: Eliminating the Carryover of Unused Individual Exemptions Would Cause Hardship and
Exceeds Agency Authority
•
•

The statutory language directs the agency to adjust exemptions in one direction when states use too many (by reducing carryover)
and in another direction when using too few (by allowing unused exemptions to accrue). There is no further directive in statute to
make further adjustments.
Final Rule dismisses recent changes to the exemption process in the 2018 Farm Bill.

Page of
CBPP said:
our
comments
153
NPRM incorrectly describes way in which
exemptions are calculated, which makes it
hard to comment on actual impact of rule.
154
Congress recently acted to adjust
exemption policy, but not in way NPRM
proposes

Page of
USDA responded:
their
response
N/A
No mention.
86

Department views the indefinite carryover of
exemptions as an unintended result of current
regulations and inconsistent with
Congressional decision to limit the number of
exemptions in a given fiscal year, in Sec.
6(o)(6)(C), (D) and (E).

15

Other Notes

154-55
155-56

156
156

157
158

159

Statutory language clearly lays out how to
address unused exemptions – NPRM
contradicts this.
Legislative history demonstrates that
Congress approved of uncapped accrual of
exemptions

N/A

No mention.

86

Agency has never raised issues with
accruing exemptions over multiple years.
States have compelling reasons to accrue
exemptions, agency never explains why
these are not legitimate reasons.

N/A

Department views the indefinite carryover of
exemptions as an unintended result of current
regulations and inconsistent with
Congressional decision to limit the number of
exemptions in a given fiscal year, in Sec.
6(o)(6)(C), (D) and (E).
No mention.

Congress knows how to limit the carryover
of unused resources but has not done so
here.
NPRM fails to provide a legitimate reason
for the change

N/A

Proposed method of calculating
exemptions and adjusting will discourage
states from using them and increase
errors.

89

89

86

Final rule modifies the proposed carryover
limits, to one year’s worth. Department
agrees with comments that exemptions are
important but no convincing evidence that
they should be carried over indefinitely.
No mention.
Department views the indefinite carryover of
exemptions as an unintended result of current
regulations and inconsistent with
Congressional decision to limit the number of
exemptions in a given fiscal year, in Sec.
6(o)(6)(C), (D) and (E).
Modified the proposal so that states can
carryover exemptions for 12 percent of
covered individuals. Department does not
believe final rule will increase likelihood of
errors.

CBPP Comments Chapter 10: The Proposed Rule Fails to Provide Sufficient Rationale or Supporting Evidence for the
Proposed Policy Change
16

Page of
CBPP said:
our
comments
160
Department claims that waivers cover too
many individuals but provides no
justification that there is a target number
of people to be waived.
- No evidence of legislative intent
either
- Fails to acknowledge that economic
circumstances, and thus number of
ABAWDs, will change
161
Fails to provide evidence that general
unemployment rates are the best
available measure of job sufficiency for
ABAWDs
161-162
Lack of rationale provided for setting floor
at 7 percent, or why other options (6 or 10
percent) would be better or worse.
- Makes it impossible to comment on
validity of proposal.

Page of
USDA responded:
their
response
N/A
Reasserts the claim made in the proposed rule
but does not provide any legislative intent or
other information to explain why this is a valid
reason to change existing policy.

162

8

163
164

Makes arbitrary changes to long-standing
regulations for which sound reasons were
provided at the time they were
promulgated.
- Eliminates LSAs which the agency
indicated were useful in initial
guidance
Agency ignores recent Congressional
rejection of this policy
Sole reason for eliminating some criteria
(E:P, declining industries, academic study)

25

25

N/A

Doesn’t directly address, but claims that the
new standard is “one of the most objective
and defensible ways” of determining a lack of
sufficient jobs.
Brief mention that Department finds the 6
percent floor to be “one of the most objective
and defensible ways” of determining a lack of
sufficient jobs. Mentions that it aligns with
longstanding DOL measure of job insufficiency
(need to check this claim)
No explanation for why long-standing policy
must change except mentions reviewing rules
and deciding states have “taken advantage of
these weaknesses” to request waivers in areas
“where it is questionable as to whether” there
are insufficient jobs. This claim is repeated
throughout the Final Rule.
Don’t directly address, but claim that current
waiver policy has veered from original intent.
Rejects E:P because it provides “ambiguous”
information about job availability. Rejects

17

Other Notes

165

166

for waivers is that they are rarely used,
sometime subjective or not appropriate if
other data is available.
-claim that data is not rigorous is not
explained given that states can submit a
wide range of data.
Public input from ANPR does not appear
to inform this proposed rule
- Not clear if comments to ANPR justify
proposed rule change
Alternatives to the proposed rule are not
discussed.
- 7, 6 and 10 percent are offered but
not explained
- Alternatives to LMAs not discussed,
reason for dropping LSA not sufficient

academic studies as not the best available
data.

N/A

Did not mention.

30-33

Summarized comments on alternatives, then
says there’s “no measure available for
precisely determining the number of available
jobs specifically for ABAWDs”. Dismisses
setting a fluctuating floor as administratively
difficult and is not persuaded by any others.

CBPP Comments Chapter 11: Regulatory Impact Analysis (RIA) highlights faulty Justification and Includes Unclear or
Flawed Assumptions
•

We said the RIA contradicts USDA’s justification for the rule because they say their goal is to encourage work and promote selfsufficiency, but the RIA (and the preamble) provides no evidence of such effects and instead finds that 755,000 individuals would be
cut from SNAP for “failure to engage meaningfully in work or work training.” The final RIA does not explicitly acknowledge this
comment, or respond to it, or provide any additional evidence, beyond saying “a review of existing evidence…was not sufficient”
and citing KS and ME studies we had addressed as flawed in our comments.

•

There is a new issue that was not in the proposed rule because USDA did not propose eliminating the Unemployment Insurance
Extended Benefit (EB) trigger in the proposed rule.
The final rule shows that USDA does not understand the impact of eliminating the EB trigger during any future recession. On p. 29
and p. 55 USDA mentions that because of the EB trigger being eliminated some areas may not qualify for waivers as quickly during
an economic downturn. But USDA doesn’t seem to understand that many areas with high unemployment wouldn’t qualify AT ALL
during and after recessions when national unemployment is high, because an area’s unemployment rate must be both over 6% AND

18

20% above the national average. So an area with 9% unemployment wouldn’t qualify if the national average unemployment rate
was 8%. This will result in far, far fewer areas qualifying for waivers during and after recessions.
This could be an important example of something they might have learned from public comments had they proposed eliminating
the EB trigger, but since they didn’t propose it no one commented about the impacts and they completely misrepresent them in the
final rule.
•

They tie themselves to using Department of Labor criteria in terms of justifying both the 6% floor (as being the Labor Surplus Area
(LSA) criteria) and the Labor Market Area (LMA) as being the appropriate area for consideration of waivers, but they reject the EB
trigger, which is the Labor Department’s measure of insufficient jobs during the period when unemployment is rising rapidly, as
occurs during a recession.

•

They were so focused on the period we currently are in of low unemployment and ensuring that low employment areas never get
waived that they have completely disregarded Congressional intent for what should happen when unemployment is rising or
objectively high.

Page of
our
comments

CBPP said:

168-170,
p. 213-214

The RIA contradicts USDA’s justification.
They say the changes would encourage
work and promote self-sufficiency, but RIA
just says (755k) people will be cut off for
failure to comply, with no evidence of any
increased work.
- No assumption about E&T
- No discussion of state
implementation choices having an
impact or other complexities of
the impacts.

170-195

Specifically, we cite research evidence on:
- Characteristics/realities of labor
market

Page of
their
response
(in RIA)
23

USDA responded:

-

They did not mention this comment
nor provide any evidence.

-

Mentions not enough evidence on
E&T to estimate the magnitude state
response.

23 RIA,
RIA - A review of existing evidence17 on the
97+
behavior of ABAWDs following a repreamble imposition of time limits was not sufficient to

19

Other Notes

Footnote in RIA
references FGA

-

195-

p.198-201

p. 202

p. 203

Work patterns and SNAP
TANF impacts modest and fade
Barriers faced, sanction rates
One study they cite doesn’t support
rule
- FGA reports that “find” positive
effects are flawed
- FNS’s own research and FSET
contradicts the rationale
The term “ABAWD” is never defined and
USDA’s use of it is confusing and leads
them to faulty assumptions in their
methodology. They seem to include many
people who would not be affected by the
change.
p. 198 – They imply more people would be
affected than is the case.
Several assumptions in the methodology
rely on shares of this larger “ABAWD”
group inappropriately.
i.e., share working.
It is misleading to include individuals who
are not subject to the time limit because
the denominator matters. We did analysis
to explain this point in detail.
They assume there would be 3.4 million
“ABAWDs” in 2020 but never explain
where that number came from. This is the
starting point for all the estimates, so it’s
important.
They assume 44% of “ABAWDs” are in
waived areas based on an assumption that
uses all non-PA participants by county,
which is indefensible because “ABAWDs”

permit us to reasonably estimate
employment rates that are likely to result
from this rule.

10, 20

Preamble – mention various studies. “this
does not change the statutory work
requirements established by Congress…policy
changes are in line with the intent of Congress
when passing PRWORA.”
Mention. Recognize complex issue. Say it was
addressed in baseline discussion.
p. 20-21 – discussion, but still way too few
people assumed able to participate compared
to the experience in other states.
Mentioned. See below on share in waived
areas and work rates.

11, 20-21

Not sure if the mention about projections for
future years is in response to this.
p. 22 – Assume share “ABAWDs” constant at
7.3% of baseline in all years.

2, 8, 22

p. 2- New assumption is “more than half of
ABAWDs currently live in areas in which the
time limit is waived.”

20

KS study paper
and ME study.

are a small fraction of non-PA participants.
Moreover, their distribution will depend
on whether the time-limit is in effect. So it
overstates the share in areas that are not
waived and understates the share in
waived areas.

P. 203-

The methodology for determining the
share of areas losing waivers is incomplete
and confusing:
- Use term “currently” inconsistently,
so not clear what year was used for
analysis – both the time periods
examined and the waiver period.
- Inconsistencies in the 24-month
period and thresholds used (p. 206)
- Only used 1 24-month period.
- Fails to explain treatment of New
England states and excludes Guam
and VI. (p. 208)
- Fails to adjust number of waived
areas for 2020 when there likely will
be fewer waived areas. (p. 208)
- Did not analyze a period when
unemployment rates are rising. (p.
208)
- Didn’t explain estimation of areas
losing due to narrowing of statewide
waivers. May have subtracted out
twice (p. 208)

11, 43-47

p. 7-8- Defend what they did but also agree
ABAWDs are a small share of overall SNAP
caseload. For the final rule they use, “SNAP
QC data counts of ABAWDs by county,” which
I don’t really understand because there is no
county identifier in the QC and the sample
sizes would likely be too small.
Result is 53.3% of ABAWDs live in waived
areas, dropping very slightly because of
unemployment rate rising in baseline, so
fewer areas waived.
Mention that a commenter said the
methodology was confusing.
-

Currently – replace with a reference year
for clarity. See p. 44-45.

-

They do not agree. Think we did not
understand. Add clarifying details.
Not mentioned.
Added clarifying discussion. (See p. 43 fn
and p. 47.)

-

-

Acknowledge did not fully explain that
did adjust. Added details.

-

Acknowledges that impact different
during downturn and includes a
qualitative discussion.
Disagrees. Thinks p. 21 discussion
describes and includes similar.

-

21

-

Omits EB as a standard. (p. 209)
Does not analyze impact on Native
Americans on reservations or New
England towns. (p. 209)

-

p. 210

And the discussion of the relative impact
of the three changes is misleading: The
presentation suggests 7% floor and
restriction to combine areas are roughly
the same size, when the impact of the
inability to combine areas is subsumed in
the 7% floor change. All who are no longer
eligible because of not being able to
combine areas are also ineligible because
of the 7% floor.

24-43
and 4547, 55

p. 211

RIA assumes all “ABAWDs” in areas that
would lose waivers would lose SNAP,
which ignores that many will be exempt or
able to participate for other reasons.
None would be:
- exempted for being unfit
- pregnant
- participating in first (or 2nd) 3 months
- exempt because of discretionary
exemptions
- participating in E&T
- working for less than 20 hours a week

9-10, 23

Agrees not clear, but no effect.
Mention comment. Added clarifying
discussion. See p. 43 and 47 and p. 31 for
reservations.
Fn 25 on p. 46 says the impact would vary
depending on the order, which sort of
addresses this point. They eliminated EB as a
criteria and say there would be no effect in
the 2020-2024 window, which is misleading.
They mention there would be a delay in areas
qualifying for a waiver during an economic
downturn, implying that all areas with
unemployment over 6.5% would qualify,
which is misleading since they would have to
ALSO have unemployment more than 20%
above the national average, and so
many/most would NOT qualify if national
unemployment rate also had risen.
Also see p. 55 regarding changes in economy.
Mention that a commenter noted that the
proposed rule RIA did not take into account
that some ABAWDs in unwaived areas may
receive exemptions for pregnancy or
unfitness for work. Appreciates the comment
and has included assumptions regarding such
exemptions in the final rule. (The comment
included others.)
Assume 8% of ABAWDS in newly unwaived
areas will receive an exemption for pregnancy
or unfit for work based on data from 6 states.
States will use banked exemptions in 2020
and 5% of ABAWDs will be exempt under
discretionary exemptions after that.

22

Note- p. 55 they
say they expect
the “exceptional
circumstances”
provisions may
alleviate the
issue of fewer
areas qualifying
quickly during a
recession.

8% unfit or
pregnant seems
way too low.

p. 212

p. 212

p. 214

p. 217

RIA assumption that 1/3 of “ABAWDs”
subject to the time limit would be working
is flawed because it was derived from the
entire population, which includes nonwaived areas where a higher share are
working. Only 12% of “ABAWDs” in nonwaived areas were working in 2017. No
explanation for how this would triple.
RIA assumes all individuals who lose
eligibility will reapply every three years.
There is no evidence of this, and a 2001
FNS study suggests it is not true.

7, 22

Mention, combine with point about
questioning whether work requirements
increase work, and say they “found comments
regarding the ABAWD employment rate to be
compelling and has adjusted that
assumption.” Use 23.8% from current partial
or statewide waiver.

10

Treatment of discretionary exemptions
confusing and misleading.
- Assumption that “states use
approximately 65% of their earned
exemptions in an average year” is
overstated.
- Assumption that eliminating carryover has no impact is indefensible.

13, 2425, 4243, 4849.

Mention and say, “the Department does not
find this argument to be compelling.” SNAP
participation rates have increased
dramatically since 2001 for this group from
54% to 85%. So not clear that study would still
apply.
Mentioned 2nd, not first

RIA fails to reflect impact on Medicaid and
health coverage (because of work
requirements alignment) and other
secondary impacts (on economy and
health outcomes).

6

-

-

Not mentioned.

Notes provision has changed and
includes discussion. They now assume
fewer people cut off in 2020 because
many states will use banked exemptions
before they are rescinded.
Mentions “Commenters noted specific
potential costs that they did not believe were
addressed in the RIA.” Including
- “Costs to U.S. and local economies in
generating economic activity and acting
as an economic stabilizer” and
- “Healthcare costs related to increases in
food insecurity and poverty.”

23

Assumed all who
are working at all
would make it to
20+ hours a
week. Provide no
evidence.

“In response to these comments, this RIA
includes expanded qualitative and
quantitative discussion of these potential
impacts where appropriate.” Says they can’t
estimate them, though.

24