Вы находитесь на странице: 1из 21

This article was downloaded by: [71.192.165.

4]
On: 22 April 2012, At: 15:14
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Hunger & Environmental


Nutrition
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/when20

Food Insecurity After Leaving SNAP


a a
Mark Nord & Alisha Coleman-Jensen
a
USDA Economic Research Service, Washington, DC, USA

Available online: 08 Dec 2010

To cite this article: Mark Nord & Alisha Coleman-Jensen (2010): Food Insecurity After Leaving SNAP ,
Journal of Hunger & Environmental Nutrition, 5:4, 434-453

To link to this article: http://dx.doi.org/10.1080/19320248.2010.527277

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation
that the contents will be complete or accurate or up to date. The accuracy of any
instructions, formulae, and drug doses should be independently verified with primary
sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand, or costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
Journal of Hunger & Environmental Nutrition, 5:434–453, 2010
ISSN: 1932-0248 print/1932-0256 online
DOI: 10.1080/19320248.2010.527277

Food Insecurity After Leaving SNAP∗

MARK NORD and ALISHA COLEMAN-JENSEN


USDA Economic Research Service, Washington, DC, USA

Households that have recently left the Supplemental Nutrition


Assistance Program (SNAP) are more likely to be food insecure
than those who remained on the program. We analyze data from
the Current Population Survey Food Security Supplement to find
Downloaded by [71.192.165.4] at 15:14 22 April 2012

reasons for this seeming anomaly. Analysis of a 2-year panel sam-


ple of households interviewed in 2 successive years found different
conditions in 2 distinct groups. One third of households that left
SNAP in one year returned in the following year and were more
likely to have had very low food security in both years. Households
that left SNAP in 1 year and remained off through the following
year were also more likely than current recipients to have had very
low food security in the first year, but by the end of the follow-
ing year, the prevalence of very low food security was lower than
among those still on the program. Logistic regression analysis indi-
cates that, for households with very low food security, leaving SNAP
was strongly associated with higher income and full-time employ-
ment, suggesting that most such households left SNAP because their
income had increased so that they were no longer eligible.

KEYWORDS food security, food insecurity, SNAP (Food Stamp


Program)

INTRODUCTION

Very low food security—a severe range of food insecurity in which eat-
ing patterns of some household members are disrupted and their food

This article not subject to U.S. copyright law.


The views expressed here are those of the authors and cannot be attributed to the
Economic Research Service or the US Department of Agriculture.

SNAP, the Supplemental Nutrition Assistance Program, was formerly called the Food
Stamp Program.
Address correspondence to Mark Nord, USDA Economic Research Service, 1800 M
St. NW, Room N2180, Washington, DC 20036. E-mail: marknord@ers.usda.gov

434
Food Insecurity After Leaving SNAP 435

intake reduced below levels they consider appropriate—is more prevalent


in households that have recently left the Supplemental Nutrition Assistance
Program (SNAP) than among households still receiving SNAP benefits.1–6
For example, among households in the December 2008 Current Population
Survey Food Security Supplement (CPS-FSS) that received SNAP benefits at
some time during the year but not in the 30 days prior to the survey, 24.5%
had very low food security during that 30-day period, compared with 14.4%
of households still receiving SNAP and 8.9% of households with similarly
low incomes that did not receive SNAP benefits at any time during the year.6
The question motivating this study is why very low food security is
more prevalent among recent SNAP leavers than among current SNAP recip-
ients. We seek to answer the question by addressing the obverse question
of why households exit SNAP when they are unable to meet their food
needs without assistance. A number of studies have examined the reasons
households exit SNAP, but none has looked specifically at why households
Downloaded by [71.192.165.4] at 15:14 22 April 2012

with continuing unmet food needs exit SNAP or at how the food security
of households that leave SNAP may differ depending on why they leave the
program. Answers to these questions are important because different policy
and program options to support the food security of households as they exit
may be appropriate depending on the factors causing them to exit SNAP or
causing their food insecurity after doing so.
Using data from the CPS-FSS for 2002–2006, we examined the food
security status of households 1 to 11 months after leaving SNAP and again a
year later. We conducted separate analyses of households that remained off
SNAP in Year 2 and those that left SNAP in Year 1 but returned to SNAP in
Year 2. We then used logistic regression to assess whether the higher food
insecurity of recent SNAP leavers could be accounted for by less favorable
income, employment, and other household circumstances and to examine
the demographic and economic characteristics of households that left SNAP
in spite of having very low food security.

PRIOR RESEARCH ON SNAP LEAVERS

A series of state-level studies in the late 1990s examined the unexpected fall
in SNAP caseloads after implementation of the Personal Responsibility and
Work Opportunity Reconciliation Act of 1996.7–11 Findings from these studies
included

● The most common reason for leaving SNAP, reported by one half to two
thirds of leavers, was improved income and employment.
● Households also commonly left SNAP because they failed to recertify or to
complete necessary paperwork, but often those households also had had
an increase in income or other resources.
436 M. Nord and A. Coleman-Jensen

● Between one fourth and one third of SNAP leavers rejoined the program
within a year or two of leaving.
● The prevalence of very low food security was relatively high (23% to 26%)
among households 1 to 2 years after leaving SNAP. In some studies, very low
food security was somewhat less prevalent among leavers than stayers, but
in one study,10 food hardship prior to leaving SNAP, assessed retrospectively
after exit, was considerably more prevalent than after leaving SNAP.A
Findings from national surveys are consistent with the findings from state
leaver studies: A substantial majority of households that leave SNAP do so
because of a new job or increased earnings.12–14 One third to one half of
SNAP leavers rejoin the program within 1 or 2 years.14,15
A longitudinal study of low-income families in 3 large cities found that
food hardships were more prevalent among households receiving SNAP in
only 1 or 2 waves of the study than in those receiving SNAP in all 3 waves
Downloaded by [71.192.165.4] at 15:14 22 April 2012

of the study.16 This may suggest that recent SNAP leavers were more likely
to be food insecure than longer-term SNAP recipients.
Studies of indebtedness and arrearages of households leaving cash wel-
fare programs provide inferential evidence of what may be expected in
the case of SNAP leavers. Temporary Assistance to Needy Families (TANF)
leavers in North Carolina were found to have incurred significant debt and
to have accumulated further debt as they sought and entered employment.17
Leavers were also more likely than current recipients to have been contacted
by bill collectors and to have had credit counseling due to credit problems.
The earlier research leaves several important gaps. They confirmed the
relatively high rates of food insecurity among recent SNAP leavers but did
not systematically compare food security among leavers who remained off
SNAP, leavers who rejoined SNAP, and households that remained on SNAP
throughout the observation period. Previous studies consistently found that
most households left SNAP because of improved income or employment, but
none looked specifically at why households with very low food security left
the program. The present study addresses these gaps to provide information
on whether SNAP could be enhanced to better meet the food needs of
households as they exit the program.

Why Might Households Exit SNAP Even Though They Are Unable to
Meet Their Food Needs Without Assistance?
Several processes may provide partial explanations for the higher food
insecurity among recent SNAP leavers. Some may leave SNAP, even though
their income is still within the eligibility range, because

A
The methodology relied on retrospective reporting of 4 of the 6 questions in the 6-item short form
of the Household Food Security Survey Module. As a result, prevalence rates of the standard categories
of food insecurity could not be calculated for the earlier period. Instead, responses to individual food
security questions were compared between the 2 time periods.
Food Insecurity After Leaving SNAP 437

● they become ineligible due to criteria other than income (such as failing
to comply with work requirements).
● they fail to recertify, either deliberately or through negligence, even
though they are still eligible.

Other households may leave SNAP because their income increases beyond
the eligibility range, yet remain food insecure because

● they have accumulated debts and overdue bills that reduce resources
available for food and other needs.
● they have ongoing expenses that reduce resources available for food and
other needs.
● the cost of living in their geographic area is high, and income higher than
the SNAP income eligibility limit is needed to ensure adequate resources
for food.
Downloaded by [71.192.165.4] at 15:14 22 April 2012

● their income fluctuated; a temporary increase made them ineligible, but


their income then declined.

The data used in this study cannot differentiate among all of these processes,
but associations of SNAP exit with various household characteristics can
provide inferential evidence of why households left SNAP.B

DATA AND METHODS

Data were from the CPS-FSS conducted in December of each year from
2002 to 2007. The CPS-FSS is an annual supplement to the monthly Current
Population Survey and is nationally representative of the civilian, nonin-
stitutionalized population. It is the data source for the US Department of
Agriculture’s (USDA) annual report on household food security in the United
States.6 The CPS-FSS collects information on how much households spend
for food, their use of federal and community food and nutrition assistance
programs such as SNAP, and their food security during the 12 months and
30 days prior to the survey. The CPS core includes data on household com-
position, demographics, and income and data on the employment and labor
force status of persons age 15 and older.
The samples used for specific analyses are described after the food
security and SNAP status variables used to select the samples.

B
Data from the Survey of Income and Program Participation (SIPP) could provide more complete
information about eligibility and therefore support more detailed inference about reasons for leaving
SNAP. However, food security information is collected at only one point in each SIPP panel, is measured
in less detail, and references the previous 12 months, which would not support analysis of food security
in the first few months after exiting SNAP.
438 M. Nord and A. Coleman-Jensen

Measuring Food Security


Household food security is measured in the CPS-FSS by a series of questions
about food-related behaviors, experiences, and conditions that characterize
households having difficulty meeting their food needs.6,18–20 The questions
cover a wide range of severity of food access problems, and each question
specifies a lack of money or other resources to obtain food as the reason for
the condition or behavior.
This study used a measure of food security based on 7 questions about
the experiences of adults in the household over the 30-day period prior to
the survey.21 Use of the 30-day measure was essential, because the stan-
dard 12-month measure used in USDA food security reports would identify
some recent SNAP leavers as food insecure that were only food insecure
earlier in the year, prior to leaving the program. The older 30-day scale
used in this study identifies households with very low food security but
Downloaded by [71.192.165.4] at 15:14 22 April 2012

does not fully differentiate households with high, marginal, and low food
security.C This was not considered problematic, because SNAP participation
is more likely to affect food insecurity at this severe level than at less severe
levels.
On this scale, households reporting 3 or more conditions indicating
reduced food intake among adults are classified as having very low food
security. The minimum extent of food hardship reported in such households
is that an adult in the household cut the size of meals or skipped meals in
3 or more days during the previous 30 days because there was not enough
money for food and the respondent had eaten less than he or she felt that
he or she should because there was not enough money for food. A majority
also reported that they had been hungry but did not eat because there was
not enough money for food.

Measuring SNAP Participation: Identifying SNAP Leavers


Households reporting that anyone in the household had received “food
stamps or a food stamp benefit card” at any time during the year were
asked in which months benefits were received. Those reporting receipt in
November but not December were asked the date of last receipt. Households
were coded as remaining on SNAP if they received benefits either in
December or later in November than 1 month prior to the first date of

C
The 30-day scale used in this study is based on the same concepts and statistical methods as the
standard 12-month scale but uses only the subset of 30-day-referenced questions that were included in
the CPS-FSS prior to 2005.21 Beginning in 2005, the CPS-FSS has included an expanded set of 30-day-
referenced questions corresponding to the entire set of questions in the 12-month scale. An updated
30-day scale based on the full set of questions has been used as the basis for 30-day statistics in USDA
reports beginning with the 2005 data.3 The older scale was used in this study because data from earlier
years were needed to obtain a sufficiently large sample.
Food Insecurity After Leaving SNAP 439

the survey week (conducted in early to mid-December). Households that


received SNAP at some time during the year but not during the previous
30 days were classified as recent SNAP leavers.D Households were omitted
from the analysis if they had missing data on SNAP receipt during the year
(0.4% of otherwise eligible households) or during the previous 30 days (an
additional 0.5% of households).

Analysis Samples
The analyses used data for SNAP recipients and for nonrecipients with
annual incomes less than 130% of the federal poverty line. Both pooled
samples of single-year data and pooled 2-year panels were analyzed. The
pooled single-year data consisted of households in the CPS-FSS from 2002 to
2006 that received SNAP benefits at some time during the year (N = 13 881)
or had annual income below 130% of the poverty line and did not receive
Downloaded by [71.192.165.4] at 15:14 22 April 2012

SNAP during the year (N = 25 415).E Data from 2007 were omitted from
the pooled single-year samples so that the time period represented by those
analyses would be the same as that represented by Year 1 in the pooled
2-year panels.F
The pooled 2-year panels consisted of households that were inter-
viewed in 2 successive years from 2002–2003 to 2006–2007 and received
SNAP benefits in Year 1 or had annual income less than 130% of the poverty
line in at least 1 year. Households were omitted if they moved between years
or were not interviewed in either year.
Data from the December 2008 CPS-FSS were not used because of
the large changes in income, employment, and food security in 2008.
Changes in SNAP participation associated with large macro-economic shifts
may be very different from those associated with changes in the eco-
nomic and social circumstances of individual households in a stable national
economy.

D
A sensitivity analysis was conducted in which households were considered to have remained
on SNAP if they reported SNAP receipt in either November or December. Relatively few additional
households were classified as remaining on the program by this definition, and the analytic results were
not substantively different than those presented.
E
The sample of households that did not receive SNAP at any time during the year was restricted
to those with annual incomes less than 130% of the poverty line even though some SNAP recipient
households had incomes above that level (likely because their income fluctuated during the year, the
SNAP eligibility unit did not include all household members, or they were adjunctively eligible due
to participation in a cash welfare or other assistance program). Statistics for nonrecipient households
are presented only in the descriptive statistics; the regression analyses included only households that
had received SNAP at some time during the year. The lower income limit on the nonrecipient sample
would have the effect of muting somewhat the difference between SNAP leavers and nonrecipients in
the descriptive analyses.
F
As a robustness check, the analyses were repeated with the 2007 data included, and the results
were essentially unchanged.
440 M. Nord and A. Coleman-Jensen

Two-Year Panel Construction


Two factors reduce the number of households that can be used for any
2-year panel analysis of the CPS-FSS. First, some households are interviewed
in only 1 of the 2 years or complete the supplement in only 1 of the 2
years. Second, the CPS is a sample of housing addresses; if a household
interviewed in Year 1 moves out and another moves in, the new house-
hold is interviewed in Year 2.22 (See Wilde and Nord22 for further details on
matching CPS-FSS households across years.)
Low-income households are more likely to move or to miss an interview
and so are typically underrepresented in 2-year panel samples. In this study,
the underrepresentation of economically vulnerable households is seen in
the lower prevalence of very low food security in the 2-year panel sample
compared with the full sample. For example, among current SNAP recipi-
ents, 11.8% of households in the pooled 2-year panel sample had very low
Downloaded by [71.192.165.4] at 15:14 22 April 2012

food security compared with 13.2% in the pooled single-year sample.


To assess whether the nonrandom sample loss biased results in this
study, most analyses that used the 2-year panel data were replicated using
the single-year sample. The similarity of results in the 2 samples increases
confidence that results of analyses that could only be conducted using the
2-year data were not substantially distorted by nonrandom sample loss in
the 2-year panels.

Measuring Household Characteristics Used as Covariates


Household characteristics were calculated from data elements in the CPS-
FSS and the CPS core labor force survey. Descriptions of several constructed
variables follow. The remaining covariates are readily understood.
Income is reported in ranges in the CPS. Dummy variables were calcu-
lated to represent 8 categories of income (assigned at the midpoint of the
reported range) relative to the poverty line for the household.
The employment/labor force status of a “primary earner” in each
household was assigned in a 2-step process. First the status of each adult
was assigned to one of the following categories based on the individual’s
“monthly labor force recode” (PEMLR) and “full-time/part-time work status”
(PRWKSTAT):

1. Employed full time


2. Not in the labor force—retired
3. Employed part time for noneconomic reasons (ie, do not want to work
more hours)
4. Employed part-time for economic reasons (ie, want to work more hours,
unable to find a job with more hours)
5. Unemployed, looking for work
Food Insecurity After Leaving SNAP 441

6. Not in the labor force due to disability


7. Not in the labor force for reasons other than retirement or disability

The primary earner was then identified as the adult with the lowest-
numbered status, and that status was assigned to the household, represented
by a set of dummy variables.
The size of SNAP benefit was calculated as the ratio of the amount of
the monthly benefit last received to the cost of the Thrifty Food Plan (TFP)G
for that household. The benefit amount was classified as small (up to one
third the cost of the TFP), medium (one third to two thirds the cost of the
TFP), or large (more than two thirds the cost of the TFP. TFP costs were for
December of the survey year.23

Analytic Methods
Downloaded by [71.192.165.4] at 15:14 22 April 2012

The analysis consisted of a descriptive comparison of the food security of


households by SNAP status and 4 multivariate logistic regression analyses.
For the descriptive comparison, the prevalence of very low food security
was calculated for households on SNAP, for recent SNAP leavers, and non-
SNAP households with low income. In the pooled 2-year panel sample, these
rates were calculated separately for each year (with SNAP status identified
as of Year 1) and separately for those that remained off SNAP in Year 2 and
those that rejoined SNAP in Year 2.
The first logistic regression model assessed whether the higher preva-
lence of very low food security among recent leavers could be accounted
for by less favorable income, employment, and other household characteris-
tics. The analysis sample consisted of households in the pooled single-year
sample that had received SNAP benefits at any time during the year. The
dependent variable was very low food security status in the 30 days prior to
the survey, and the key independent variable was whether the household
had exited SNAP before the last 30 days prior to the survey. Controls were
included for income, employment, education of adults in the household,
household composition, race/ethnicity of household reference person, and
geographic region.
Three logistic regression models of SNAP exit were then estimated to
assess the relative prominence of reasons for leaving SNAP in spite of hav-
ing very low food security. One model used the pooled single-year data

G
The Thrifty Food Plan—developed by the USDA—serves as a national standard for a nutritious,
low-cost diet. It represents a set of “market baskets” of food that people in specific age and gender
categories could consume at home to maintain a healthful diet that meets current dietary standards,
taking into account the food consumption patterns of US households. The cost of the Thrifty Food Plan
is based on national average prices and is revised each month to account for inflation in food prices.
442 M. Nord and A. Coleman-Jensen

and 2 used the pooled 2-year panel data. The estimation samples all com-
prised households that received SNAP at some time during the year (during
Year 1 in the 2-year panel samples) and had very low food security in the
30-day period prior to the food security survey (in Year 1 of the 2-year panel
samples). The dependent variable was whether the household exited SNAP
prior to 30 days before the food security survey (in Year 1 of the 2-year
panel samples). One of the 2-year panel data models contrasted households
that left SNAP in Year 1 and remained off in Year 2 with those that remained
on SNAP through Year 1 and into Year 2. The second contrasted households
that left SNAP in Year 1 but rejoined in Year 2 with those that remained on
the program through Year 1 and into Year 2.
All calculations used household supplement weights. Analysis of the
2-year panels used weights for Year 1.
Downloaded by [71.192.165.4] at 15:14 22 April 2012

FINDINGS
Very Low Food Security by SNAP Status
The pattern described over the past several years by Nord et al3–6 is repli-
cated in the pooled single-year data (Table 1, rows 1, 5, and 8). Very low
food security in the 30-day period prior to the food security survey was more
prevalent among recent SNAP leavers (14.7%) than among those still on
the program (13.2%) or among low-income households that did not receive
SNAP at any time during the year (7.0%). The differences were all statistically
significant with 95% confidence (single-tailed tests).
A similar pattern was seen in the 2-year panel sample, but the statistics
for SNAP leavers in Year 1 mask considerable differences between those that
rejoined SNAP in Year 2 and those that remained off SNAP. The prevalence
of very low food security was considerably higher in both years among
households that rejoined SNAP in Year 2 (20.2% in Year 1 and 19.5% in
Year 2). Households that remained off SNAP in Year 2 were also more
likely than those remaining on SNAP to experience very low food secu-
rity in the final 30 days of Year 1 (13.9% versus 11.8%). However, by the
end of Year 2, the prevalence of very low food security declined to 10.0%
for those that remained off the program but was essentially unchanged for
those continuing on SNAP (11.7%).
The higher prevalence of very low food security among recent SNAP
leavers cannot be accounted for by differences between leavers and contin-
uing SNAP recipients in economic, demographic, and other characteristics
measured in the survey. Rather, the association of very low food security
with SNAP exit was even stronger in multivariate analysis with controls
for those characteristics. The odds ratio corresponding to the difference in
prevalence rates of very low food security between recent leavers and those
still on the program as reported in Table 1 (ie, with no controls) was 1.133
Food Insecurity After Leaving SNAP 443

TABLE 1 Prevalence of Very Low Food Security in the 30 Days Prior to the Food Security
Survey, by SNAP Participation Statusa

Prevalence of very low


food security (30-day)

Ref. SNAP status Year 1 (%) Year 2 (%) Number of cases

Households that left SNAP


1 SNAP leavers prior to 30 days before 14.7 NA 2230
survey, full sampleb
2 SNAP leavers prior to 30 days before 15.9 13.1 609
survey in Year 1, 2-year panel
samplec
3 SNAP leavers prior to 30 days before 13.9 10.0 414
survey in Year 1, remained off
SNAP through Year 2, 2-year panel
samplec
4 SNAP leavers prior to 30 days before 20.2 19.5 195
survey in Year 1, reentered SNAP
Downloaded by [71.192.165.4] at 15:14 22 April 2012

in Year 2, 2-year panel samplec


Households that continued to receive SNAP benefits
5 SNAP recipient in 30 days prior to 13.2 NA 11 651
survey, full sampleb
6 SNAP recipient in 30 days prior to 11.8 11.7 3209
survey in Year 1, 2-year panel
samplec
7 SNAP recipient in 30 days prior to 12.9 13.3 1995
survey in both years, 2-year panel
samplec
Households that did not receive SNAP benefits
8 No SNAP in year of survey, income 7.0 NA 26,415
<130% of poverty, full samplec
9 No SNAP in respective year, income 6.1 6.2 7,918 (y1);
<130% of poverty in respective
year, 2-year panel samplec 7,485 (y2)
10 No SNAP in either year, income 6.7 6.6 5,760 (y1);
<130% of poverty in respective
year, 2-year panel samplec 5,684 (y2)
a
NA indicates not applicable.
b
The full sample is a pooled sample of all households interviewed in 2002–2006 with valid food security
and SNAP participation data, irrespective of whether they can be matched across years.
c
The 2-year panel sample comprises households interviewed in 2 successive years, 2002–2003 to 2006–
2007, with valid food security and SNAP participation data in both years. In most analyses, very low
food security is less prevalent in the 2-year panel sample than in the full sample because better-off
households are less likely to move house and more likely to be available for interview in both years than
more economically vulnerable households.22
Source: Current Population Survey Food Security Supplements December 2002 to December 2007.24

(14.7% versus 13.2%). With controls for household characteristics, the odds
ratio was 1.325 (Table 2). That is, the odds of very low food security were
32.5% greater for households that recently exited SNAP than for otherwise
similar households that remained on the program.
444 M. Nord and A. Coleman-Jensen

TABLE 2 Logistic Regression of Very Low Food Security (During Last 30 days) on Recent
SNAP Exit With Controls for Household Characteristics, Among Households That Received
SNAP Benefits at Some Time During the Year

Characteristic Log-odds Odds ratio P

Intercept −2.529 <.001


Exited SNAP recently (at least 30 days prior 0.281 1.325 <.001
to survey, not more than 11 months)
Survey year (2002 reference)
2003 −.215 .806 .014
2004 −.018 .983 .831
2005 −.032 .968 .691
2006 .092 1.096 .258
Income relative to the poverty line (higher
than 200% reference)
Income not reported .428 1.534 .003
Downloaded by [71.192.165.4] at 15:14 22 April 2012

0% to 50% .224 1.251 .095


50% to 75% .267 1.306 .049
75% to 100% .119 1.126 .394
100% to 125% .053 1.054 .721
125% to 150% −.268 .765 .112
150% to 200% .167 1.182 .293
Employment/labor force status of primary
earner (employed full-time reference)
Not in labor force—retired −.435 .647 <.001
Employed part time—noneconomic .033 1.034 .750
reasons
Employed part time—economic reasons .309 1.362 .023
(ie, prefer to work more hours)
Unemployed (looking for work) .313 1.368 <.001
Not in labor force due to disability .589 1.802 <.001
Not in labor force—reasons other than .263 1.301 .009
retirement or disability
Education of most highly educated adult
(high school or GED reference)
Less than high school −.116 .890 .083
Some college, no 4-year degree .161 1.175 .010
Bachelor degree or higher .121 1.128 .263
Household composition (married couple
with children reference)a
Single male with child/children .003 1.003 .987
Single female with child/children .186 1.204 .024
Two or more adults, no children .591 1.805 <.001
Sole male .557 1.746 <.001
Sole female .686 1.986 <.001
Race and Hispanic ethnicity (non-Black,
non-Hispanic reference)b
Black, non-Hispanic −.250 .779 <.001
Hispanic −.247 .782 .002
(Continued)
Food Insecurity After Leaving SNAP 445

TABLE 2 (Continued)

Characteristic Log-odds Odds ratio P

Census region (Northeast reference)


Midwest .177 1.194 .034
South .017 1.017 .823
West .339 1.404 <.001
Received SNAP benefits in Januaryc −.094 .910 .115
Sample size 13 881
Somers’ D .270
a
Reference group included a small number of households with children in complex living arrangements
(such as children of a boarder or of an unmarried partner).
b
Reference group is primarily White, non-Hispanic, but includes non-Hispanic Asian/Pacific Islanders,
American Indian/Alaska Natives, and those indicating more than one race.
c
Receipt of SNAP benefits in January was included in the model as a proxy for longer-term receipt of
benefits. It is likely that a large majority of households that received SNAP benefits in January had been
receiving benefits for several months previously.
Downloaded by [71.192.165.4] at 15:14 22 April 2012

Source: Current Population Survey Food Security Supplement, 2002–2006. The sample comprised
households that received SNAP benefits at some time during the survey year.24

Why Did Households Leave SNAP in Spite of Having Very Low Food
Security?
Higher income and more favorable employment status were positively and
strongly associated with SNAP exit in the pooled single-year sample of
households with very low food security (Table 3). Households with incomes
below 125% of the poverty line were much less likely to exit than those
with incomes higher than twice the poverty line (the reference category for
the income variables), and the association was almost perfectly monotonic
across the low-income range. Households with an adult employed full-time
(the reference category for the employment/labor force status variables)
were much more likely to exit SNAP than any of the other groups.H
Able-bodied adults without dependents (ABAWDs) are eligible for only
3 months of SNAP benefits every 3 years unless they maintain employment,
but there is no evidence that ABAWD restrictions were a prominent reason
for SNAP exit. In addition to the negative association between unemployment
and SNAP exit, coefficients for the 3 groups without children—most likely to
be affected by ABAWD restrictions—differed little from that for single females
with children, who are not affected by the ABAWD restrictions.
Households that had received relatively large SNAP benefits were more
likely to exit the program than those receiving medium or small benefits.

H
Education of the most highly educated adult in the household was included in early exploratory
models as a set of dummy variables, but it added little to the models and the effect on other coefficients
was negligible. Apparently any effects of education on SNAP exit were mediated to a great extent by
income and employment.
446 M. Nord and A. Coleman-Jensen

TABLE 3 Logistic Regression of Recent SNAP Exit (at Least 30 Days, Not More Than
11 Months) on Household Characteristics, Among Households With Very Low Food Security
in the 30 Days Prior to the Survey

Characteristic Log-odds Odds ratio P

Intercept −.110 .78


Survey year (2002 reference)
2003 −.310 .733 .22
2004 .248 1.282 .24
2005 −.086 .918 .69
2006 −.072 .930 .74
Income relative to the poverty line
(higher than 200% reference)
Income not reported −1.292 .275 <.01
0% to 50% −.950 .387 <.01
50% to 75% −1.128 .324 <.01
Downloaded by [71.192.165.4] at 15:14 22 April 2012

75% to 100% −.651 .522 .03


100% to 125% −.588 .555 .07
125% to 150% −.080 .923 .82
150% to 200% −.216 .806 .51
Employment/labor force status of
primary earner (employed full
time reference)
Not in labor force—retired −1.952 .142 <.01
Employed part −1.022 .360 <.01
time—noneconomic reasons
Employed part time—economic −.504 .604 .11
reasons (ie, prefer to work more
hours)
Unemployed (looking for work) −1.006 .366 <.01
Not in labor force due to disability −1.704 .182 <.01
Not in labor force—reasons other −1.256 .285 <.01
than retirement or disability
Household composition (married
couple with children reference)a
Single male with child/children −.423 .655 .29
Single female with child/children .096 1.101 .62
Two or more adults, no children −.056 .946 .81
Sole male .018 1.018 .95
Sole female .069 1.072 .80
Race and Hispanic ethnicity
(non-Black, non-Hispanic
reference)b
Black, non-Hispanic .040 1.041 .82
Hispanic .210 1.234 .29
Census region (Northeast reference)
Midwest .347 1.415 .17
South .203 1.225 .39
West .356 1.427 .17
(Continued)
Food Insecurity After Leaving SNAP 447

TABLE 3 (Continued)

Characteristic Log-odds Odds ratio P

Size of SNAP benefit (less than one


third maximum benefit reference)
Medium (one third to two thirds −.002 .998 .99
maximum benefit)
Large (two thirds maximum .382 1.466 .04
benefit or more)
Received SNAP benefits in Januaryc −.914 .401 <.01
Sample size 1882
Somers’ D .555
a
Reference group included a small number of households with children in complex living arrangements
(such as children of a boarder or of an unmarried partner).
b
Reference group is primarily White, non-Hispanic, but includes non-Hispanic Asian/Pacific Islanders,
American Indian/Alaska Natives, and those indicating more than one race.
c
Receipt of SNAP benefits in January was included in the model as a proxy for longer-term receipt of
benefits. It is likely that a large majority of households that received SNAP benefits in January had been
Downloaded by [71.192.165.4] at 15:14 22 April 2012

receiving benefits for several months previously.


Source: Current Population Survey Food Security Supplement, 2002–2006. The sample comprised house-
holds that received SNAP benefits at some time during the survey year and had very low food security
in the 30-day period prior to the food security survey in mid-December of the survey year.24

In combination, the positive associations of leaving SNAP with both larger


SNAP benefits and higher income may reflect substantial recent increases in
income that either made the household ineligible or would have reduced
benefits to such a small amount that there was little motivation to recertify.
The positive associations of leaving SNAP with higher income and full-
time employment were also observed in the 2-year panel sample and were
stronger and more consistent for households that remained off SNAP in
Year 2 than for households that reentered SNAP (Table 4). Compared with
households that remained on SNAP, those that exited SNAP and remained
off the program in Year 2 were much more likely to have had incomes
higher than twice the poverty line (Table 4, model 1), and households in
which income increased by at least 10% of the poverty line from Year 1 to
Year 2 were much more likely to exit SNAP in Year 1 than those with smaller
increases or declines in income. The association of SNAP exit with full-time
employment was also very strong.
These associations reflect the experience of substantial proportions of
households. About 25% of households that exited and remained off SNAP
had incomes higher than twice the poverty line, 52% had incomes higher
than 125% of the poverty line, and 82% had a full-time worker (analysis
not shown). The corresponding percentages for continuing SNAP recipients
were 1.6% with incomes higher than 200% of the poverty line, 6.5% with
incomes higher than 125% of the poverty line, and 14% with a full-time
worker.
448 M. Nord and A. Coleman-Jensen

TABLE 4 Characteristics of Households With Very Low Food Security That Left SNAP 30 Days
or More Prior to the Food Security Survey in Year 1, by SNAP Receipt in Year 2: Logistic
Regression Resultsa

Model 1 Model 2

Left SNAP Year 1 and Left SNAP Year 1 and


remained off in Year 2 reentered in Year 2 versus
versus remained on SNAP remained on SNAP

Characteristic Coeff. P Coeff. P

Intercept 4.304 2.548


Survey years (2002–2003 reference)
2003–2004 −1.630 .16 −.907 .27
2004–2005 .128 .89 −.825 .30
2005–2006 −.733 .44 −1.202 .14
−1.289 −1.030
Downloaded by [71.192.165.4] at 15:14 22 April 2012

2006–2007 .19 .20


Income relative to the poverty
line in Year 1 (200% or
higher reference)
Income not reported −5.690 <.01 −1.691 .25
0% to 50% −5.208 <.01 −2.162 .07
50% to 75% −5.963 <.01 −2.776 .01
75% to 100% −5.030 <.01 −1.748 .11
100% to 125% −4.797 <.01 −2.125 .06
125% to 150% −2.002 .18 −1.346 .26
150% to 200% −2.862 .05 −4.478 .13
Change in income from Year 1
to Year 2 (changed <10%
reference)
Increased more than 10% of 1.638 .04 .979 .12
poverty line
Declined more than 10% of −1.681 .11 .217 .73
poverty line
Income not reported in one 2.599 .02 NI NI
or both years
Labor force status of primary
earner in Year 1 (employed
full time reference)
Not in labor force—retired −3.445 <.01 −2.475 .03
Employed part −3.825 .03 −.841 .29
time—noneconomic
reasons
Employed part −.811 .61 −2.624 .40
time—economic reasons
(ie, prefer to work more
hours)
(Continued)
Food Insecurity After Leaving SNAP 449

TABLE 4 (Continued)

Model 1 Model 2

Left SNAP Year 1 and Left SNAP Year 1 and


remained off in Year 2 reentered in Year 2 versus
versus remained on SNAP remained on SNAP

Characteristic Coeff. P Coeff. P

Unemployed (looking for −4.025 .01 −2.051 .03


work)
Not in labor force due to −3.730 <.01 −2.777 <.01
disability
Not in labor force—reasons −3.785 .01 −3.679 .02
other than retirement or
disability
Change in full-time
Downloaded by [71.192.165.4] at 15:14 22 April 2012

employment from Year 1 to


Year 2 (no change
reference)
Primary earner full-time in .812 .44 .183 .86
Year 2, not in Year 1
Primary earner full-time in −.579 .47 .009 .99
Year 1, not in Year 2
Household composition
(married couple with
children reference)b
Single female with .012 .99 −.259 .67
child/children
Two or more adults, no .536 .59 .786 .28
children
Sole male .268 .85 .284 .79
Sole female .118 .92 −.988 .32
Race and Hispanic ethnicity
(non-Black, non-Hispanic
reference)c
Black, non-Hispanic −1.042 .16 −.016 .98
Hispanic .622 .46 .789 .34
Census region (Northeast
reference)
Midwest .559 .61 −.678 .42
South 2.048 .08 −.004 >.99
West .483 .69 −.234 .78
Size of SNAP benefit in Year 1
(less than one third
maximum benefit reference)
Medium (one third to two −.476 .45 −.222 .69
thirds maximum benefit)
Large (two thirds maximum .053 .94 .726 .24
benefit or more)
(Continued)
450 M. Nord and A. Coleman-Jensen

TABLE 4 (Continued)

Model 1 Model 2

Left SNAP Year 1 and Left SNAP Year 1 and


remained off in Year 2 reentered in Year 2 versus
versus remained on SNAP remained on SNAP

Characteristic Coeff. P Coeff. P

Received SNAP benefits in −.801 .21 −.637 .20


January of Year 1d
Sample size 344 333
Somers’ D .862 .721
a
NI indicates variable not included in model.
b
Reference group included a small number of single males with child/children and other households
with children in complex living arrangements (such as children of a boarder or of an unmarried partner).
Downloaded by [71.192.165.4] at 15:14 22 April 2012

c
Reference group was primarily White, non-Hispanic, but included non-Hispanic Asian/Pacific Islanders,
American Indian/Alaska Natives, and those indicating more than one race.
d
Receipt of SNAP benefits in January of Year 1 was included in the model as a proxy for longer-term
receipt of benefits. It is likely that a large majority of households that received SNAP benefits in January
had been receiving benefits for several months previously.
Source: Panels of matched households in Current Population Survey Food Security Supplements, from
2002–2003 to 2006–2007. The samples were limited to households that received SNAP benefits at some
time during Year 1 and had very low food security in the 30-day period prior to the food security survey
in mid-December of Year 1.24

The analysis comparing households that exited in Year 1 and returned


to SNAP in Year 2 with those that remained on SNAP throughout Year 1 and
into Year 2 revealed associations similar to those described above, although
they were generally weaker and less consistent (Table 4, model 2). The
weaker associations may reflect a mixture of reasons for leaving SNAP.

CONCLUSIONS

The higher prevalence of food insecurity of recent SNAP leavers compared


with current SNAP recipients reflects 2 processes. In part, it reflects a con-
siderably higher prevalence of food insecurity among households that exit
in spite of continuing unmet food needs and will reenter the program sev-
eral months later—about one third of households that exit in a given year.
Second, it reflects continuing or short-term worsening food insecurity of
households that have left the program and will remain off the program
the following year. On average, the food security of those households will
improve in the following year.
The poorer food security among recent SNAP leavers does not appear
to result from their having less favorable income, employment, or other
Food Insecurity After Leaving SNAP 451

household characteristics. Their circumstances should, on average, result in


their being more food secure than households that remain on the program.
The findings strongly suggest that most households that leave SNAP in
spite of having very low food security do so because their income increases
and employment improves. Many probably become ineligible for SNAP as
a result, and for others, benefits would decline to a level insufficient to
motivate recertification. The role of improved income and employment in
SNAP exit of households with very low food security is especially prominent
for those that remain off the program in the following year.
The findings suggest that continuing SNAP benefits during a transitional
period after income rises beyond the eligibility limit might help households
avoid very low food security during the initial few months after leaving
SNAP. The relatively poor food security of households soon after leaving
SNAP followed by an improvement in the second year could simply reflect
further increases in income in the second year. Alternatively, it may suggest
Downloaded by [71.192.165.4] at 15:14 22 April 2012

that during the time households were receiving SNAP benefits, some of
them put off paying utility bills or rent and accumulated credit card debt
and other consumer debt. Even after their income increased beyond the
SNAP eligibility limit, such households may have had insufficient resources
for food for several months because of the need to pay overdue bills and
pay down loans. In either case, transitional eligibility for SNAP benefits—
perhaps at a somewhat reduced level—might head off food insecurity for
many of these households during the period of transition after leaving SNAP.
Transitional SNAP benefits might also reduce very low food security among
households with a temporary spike in income that makes them ineligible for
SNAP and smooth their transition back onto the regular program if income
again falls below the eligibility level.

REFERENCES

1. Nord M. Characteristics of Low-Income Households With Very Low Food


Security: An Analysis of the USDA GPRA Food Security Indicator. Available at:
http://www.ers.usda.gov/publications/EIB25/ Accessed November 2, 2010.
2. Nord M. Food Insecurity in Households With Children: Prevalence, Seve-
rity, and Household Characteristics. Available at: http://www.ers.usda.gov/
publications/eib56/. Accessed November 2, 2010.
3. Nord M, Andrews M, Carlson S. Household Food Security in the United States,
2005. Available at: http://www.ers.usda.gov/publications/err29/. Accessed
November 2, 2010.
4. Nord M, Andrews M, Carlson S. Household Food Security in the United States,
2006. Available at: http://www.ers.usda.gov/publications/err66/. Accessed
November 2, 2010.
5. Nord M, Andrews M, Carlson S. Household Food Security in the United States,
2007. Available at: http://www.ers.usda.gov/publications/err66/. Accessed
November 2, 2010.
452 M. Nord and A. Coleman-Jensen

6. Nord M, Andrews M, Carlson S. Household Food Security in the United


States, 2008. Available at: http://ers.usda.gov/Publications/Err83/. Accessed
November 2, 2010.
7. Jensen HH, Garasky S, Wessman C, Nusser SM. Iowa Food Stamp Leavers Survey:
Final Report. Available at: http://www.ers.usda.gov/publications/efan02014-1/.
Accessed November 2, 2010.
8. Rangarajan A, Gleason PM. Food Stamp Leavers in Illinois—How Are They
Doing Two Years Later? Final Report. Available at: http://www.ers.usda.gov/
publications/efan01002/. Accessed November 2, 2010.
9. Mills G, Kornfeld R. Study of Arizona Adults Leaving the Food Stamp Program
Final Report. Available at: http://www.ers.usda.gov/publications/efan01001/.
Accessed November 2, 2010.
10. Richardson P, Schoenfeld G, LaFever S, Jackson F, Tecco M. Food Stamp Leavers
Research Study—Study of Nonwelfare Families Leaving the Food Stamp Program
in South Carolina: Final Report. Available at: http://www.ers.usda.gov/
publications/efan03003/. Accessed November 2, 2010.
Downloaded by [71.192.165.4] at 15:14 22 April 2012

11. Ribar DC, Edelhoch M. Earnings volatility and the reasons for leaving the
Food Stamp Program. In: Jolliffe D, Ziliak JP, eds. Income Volatility and Food
Assistance in the United States. Kalamazoo, Mich: W.E. Upjohn Institute for
Employment Research; 2008.
12. Zedlewski S, Gruber A. Former Welfare Families Continue to Leave the Food
Stamp Program. Washington, DC: The Urban Institute; 2001.
13. Mills B, Dorai-Raj S, Peterson E, Alwang J. Determinants of Food Stamp Program
exits. Soc Serv Rev. 2001;75:539–558.
14. Cody S, Castner L, Mabli J, Sykes J. Dynamics of Food Stamp Program Particip-
ation, 2001–2003. Available at: http://www.fns.usda.gov/ora/menu/Published/
SNAP/FILES/Participation/Dynamics2001–2003.pdf. Accessed November 2,
2010.
15. Burstein N. Dynamics of the Food Stamp Program as Reported in the Survey of
Income and Program Participation. Cambridge, Mass: Abt Associates;1993.
16. DePolt RA, Moffitt RA, Ribar DC. Food Stamps, Temporary Assistance for
Needy Families and food hardships in three American cities. Pac Econ Rev.
2009;14:445–473.
17. Stegman MA, Faris R. Welfare, work and banking: the use of consumer credit
by current and former TANF recipients in Charlotte, North Carolina. J Urban
Aff . 2005;27:379–402.
18. Hamilton WL, Cook JT, Thompson WW, et al. Household Food Security in
the United States in 1995: Summary Report of the Food Security Measure-
ment Project. Available at: http://www.fns.usda.gov/oane/menu/published/
foodsecurity/sumrpt.pdf. Accessed November 2, 2010.
19. Hamilton WL, Cook JT, Thompson WW, et al. Household Food Security in the
United States in 1995: Technical Report. Available at: http://www.fns.usda.gov/
oane/menu/published/foodsecurity/tech_rpt.pdf. Accessed November 2, 2010.
20. Bickel G, Nord M, Price C, Hamilton WL, Cook JT. Guide to Measuring
Household Food Security, Revised 2000. Available at: www.fns.usda.gov/fsec/
files/fsguide.pdf. Accessed November 2, 2010.
21. Nord M. A 30-Day Food Security Scale for Current Population Survey Food
Security Supplement Data. Available at: http://www.ers.usda.gov/publications/
efan02015. Accessed November 2, 2010.
Food Insecurity After Leaving SNAP 453

22. Wilde P, Nord M. The effect of Food Stamps on food security: a panel data
approach. Rev Agric Econ. 2005;27:425–432.
23. Center for Nutrition Policy and Promotion. Cost of Food at Home: US
Average at Four Cost Levels. US Department of Agriculture, Center for
Nutrition Policy and Promotion. Available at: http://www.cnpp.usda.gov/
USDAFoodCost-Home.htm. Accessed November 2, 2010.
24. United States Census Bureau. Current Population Survey Food Security Supple-
ment Microdata Files. Technical Documentation available at http://www.
census.gov/apsd/techdoc/cps/cps-main.html. Accessed November 2, 2010
Downloaded by [71.192.165.4] at 15:14 22 April 2012

Вам также может понравиться