Food Insecurity During the COVID-19 Pandemic: Evidence from a Survey of Low-Income Americans — Food Security
Summary
A paper in Food Security (2022) 14:165-183 by Di Fang, Michael R. Thomsen, Rodolfo M. Nayga Jr. and Wei Yang, published online 7 July 2021, titled Food insecurity during the COVID-19 pandemic: evidence from a survey of low-income Americans. The abstract states that the pandemic significantly increased food insecurity despite emergency legislation that put more resources into food assistance programs, increased unemployment benefits and provided stimulus payments. It states that the authors conducted a survey in the US on food insecurity among low-income Americans during the early months of the pandemic. It states that they cannot estimate causal effects but can show important associations between food insecurity and nutritional and economic assistance.
Summary drafted by a model from the document's text below and checked by script against that text before publication. It is a navigation aid, not a reading of what the document proves. Where AI is used
- Issuer
- Source documents
- Document type
- Food Insecurity During The Covid 19 Pandemic Evidence From A Survey Of Low Income American
- Case
- Food Insecurity During The Covid 19 Pandemic Evidence From A Survey Of Low Income Americans
Cite this document
“Food Insecurity During the COVID-19 Pandemic: Evidence from a Survey of Low-Income Americans — Food Security.”
Original at doi.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/food-insecurity-during-the-covid-19-pandemic-evidence-from-a-survey-of-low-income-americans/
, SHA-256 6d6b75b55154a5826def901c46db9a6905744d24b96bd50399c1c8b90e60f9d2.Full text
Food Security (2022) 14:165–183
https://doi.org/10.1007/s12571-021-01189-1
ORIGINAL PAPER
Food insecurity during the COVID-19 pandemic: evidence
from a survey of low-income Americans
2 3 1
Di Fang 1 & Michael R. Thomsen & Rodolfo M. Nayga Jr. & Wei Yang
Received: 17 December 2020 / Accepted: 15 June 2021 / Published online: 7 July 2021
# International Society for Plant Pathology and Springer Nature B.V. 2021
Abstract
The COVID-19 pandemic significantly increased food insecurity despite emergency legislation that put more resources into food
assistance programs, increased unemployment benefits, and provided stimulus payments. We conducted a survey in the US on
food insecurity among low-income Americans during the early months of the pandemic. While we cannot estimate causal effects,
we are able to show important associations between food insecurity and nutritional and economic assistance that highlight the
need to ensure that those newly at risk for food insecurity are able to connect to resources. For example, our results indicate that
those who lost jobs due to the pandemic reported the highest level of food insecurity and also the lowest engagement with food
assistance programs. The SNAP expansion appears to be important only among groups with higher levels of income stability
including non-minority households and those not experiencing a job loss. Thus, the SNAP expansion may not have had a
meaningful impact on those most at risk for food insecurity. Finally, our data highlight the importance of school meal programs
during normal times. Those who took advantage of school meals before the outbreak are more likely to have experienced food
insecurity during the pandemic-related school closures.
Keywords Food insecurity . COVID-19 pandemic . Low-income . Racial disparity
JEL classification C31 . I14 . Q18
1 Introduction (SNAP), being the largest nutritional assistance program in
the US, served approximately 20.9 million household with
The COVID-19 pandemic has created unprecedented chal- an average benefit of $254 per household per month in 2019
lenges. With the pandemic-induced economic downturn, lim- (USDA-FNS, 2020a). In response to the food security crisis,
ited public transportation, and disruptions to the food system, the Families First Coronavirus Act (FFCA) increased SNAP
low-income families now face adverse situations that can have benefit allotments up to the maximum allowable amount as an
significant consequences for food security. Federal nutrition emergency measure in many states (US Department of Labor,
assistance programs play an important role in sustaining food 2020). FFCA also enabled states to issue pandemic electronic
security in times of economic downturn (Bitler & Hoynes, meal-replacement benefits, or pandemic Electronic Benefits
2016). The Supplemental Nutrition Assistance Program Transfer (EBT), for households with children eligible to re-
ceive free or reduced-price school meals, equivalent to rough-
ly $114 per child a month (USDA-FNS, 2020b).1
* Di Fang More than 7 million (roughly 40% of all SNAP) house-
difang@uark.edu
holds already received the maximum benefits (Rosenbaum
et al., 2020), therefore under the emergency allotments of
1
Department of Agricultural Economics and Agribusiness, University FFCA, they received no extra aid from SNAP. On the other
of Arkansas, Fayetteville, AR, USA
2
Fay W. Boozman College of Public Health, University of Arkansas 1
for Medical Sciences, Little Rock, AR, USA Provisions also included streamlining the application process and reducing
administrative hurdles by permitting telephonic signatures and waiving the
3
Department of Agricultural Economics, Texas A&M University, application interview. These options allowed states to better manage increased
College Station, TX, USA caseloads to meet the increase in food assistance needs (Rosenbaum et al.,
2020).
166 Fang D. et al.
hand, increasing benefits alone may not be enough to address 2 Background
the barriers low-income families are facing in acquiring food
during the pandemic. Even with the larger benefit per the More than 37 million people were food insecure before the
FFCA, SNAP dollars are unlikely to last throughout the entire pandemic Coleman-Jensen et al. (2019). These numbers re-
monthly benefit period (Wilde & Ranney, 2000; Cotti et al., flect a historically low point since the Great Recession.
2020; Gregory & Todd, 2021). Additionally, stockpiling be- However, with the pandemic-induced economic downturn,
havior could worsen the situation for low-income families limited public transportation, and disruptions to the food sys-
who cannot afford to buy in bulk. For families who rely on tem, low-income families now face adverse situations that can
The Special Supplemental Nutrition Program for Women, have significant consequences for food security. At the begin-
Infants, and Children (WIC), the food security status is even ning of the pandemic, many households stockpiled shelf-
worse because WIC benefits are restricted to certain food stable foods (Information Resources, Inc., 2020), causing
items and can only be redeemed at stores that accept the ben- low-income households to visit multiple food stores to search
efits (USDA-FNS, 2020c). for the most affordable products. Poor families often travel
The objective of this paper is to understand how federal long distances to acquire food and rely heavily on public tran-
food assistance programs (SNAP, WIC, and school food pro- sit (Cannuscio et al., 2013). Moreover, with schools being
grams) have helped alleviate food insecurity among low- closed and families out of work, children in lower-income
income individuals during the pandemic. We also investigate households are at higher risk of being food insecure (Dunn
the differential associations of these programs with food inse- et al., 2020).
curity by race or ethnicity, families with children, and families On March 27 of 2020, the Coronavirus Aid, Relief, and
that experienced job loss due to COVID-19. To meet this Economic Security (CARES) Act was passed as a response to
objective, we conducted a nationwide survey in June and the economic damage caused by the virus (US Department of
July of 2020 targeting adults with incomes less than 200% the Treasury, 2020). The over $2 trillion economic relief pack-
of the Federal Poverty Line (FPL). The survey addressed age included one-time stimulus payments generally consist of
household food security status, participation in federal and $1200 per adult and an additional $500 for each child under
local assistance programs, financial hardships related to 17 years old, depending on household income. Besides the
COVID-19, food access during the pandemic, food shopping stimulus payments, 30 million unemployed workers received
behaviors, and demographic information. $600 in addition to an unemployed worker’s regular weekly
Our results indicate that losing a job due to the pandemic benefit (Stone, 2020). Raifman et al. (2021) found that the
was associated with the highest level of food insecurity but receiving unemployment insurance to be associated with large
was weakly associated with food assistance programs. reductions in food insecurity among people who lost employ-
Pandemic related job loss was highly associated with the re- ment during the pandemic.
ceipt of unemployment benefits, something most newly un- The CARES Act included a $15.8 billion appropriation for
employed would know are available to them. Hence, food SNAP and $8.8 billion for child nutrition programs. SNAP is
assistance programs may not be reaching those who have an entitlement program that places few limits on allowable
great need but little prior experience with food benefits. Our purchases. By design, SNAP responds quickly to increased
results also show a negative association between food insecu- need due to job loss or other economic shocks (Bitler et al.,
rity and SNAP benefit expansion among groups with higher 2020). As previously mentioned, SNAP families were able to
levels of income stability including non-minority households receive additional benefits (up to the maximum SNAP benefit)
and those not experiencing a job loss. Because the benefit during the pandemic. Because SNAP benefits normally de-
expansion could only help those who were not already receiv- cline as income rises, SNAP recipients with higher incomes
ing the maximum SNAP benefit, it may not have had a mean- were more likely to realize an increase in benefits under the
ingful impact on those at greatest risk for food insecurity. Our FCCA provisions. The CARES Act also provided $500 mil-
results also indicate that those who took advantage of school lion to WIC to be allocated to all WIC state agencies. To
meals before the outbreak are more likely to have experienced adhere to social distancing guidelines, the Act waived the
food insecurity after the pandemic-related school closures. requirement for individuals seeking WIC certification to visit
Short term solutions such as the Pandemic EBT program WIC clinics in person (USDA-FNS, 2020c). An earlier report
could significantly decrease the likelihood of food insecurity indicates that the availability of WIC foods in WIC approved
among the lowest-income respondents with children. retail stores was deemed to be a barrier to effectively using
The paper proceeds as follows. The next section discusses WIC during the pandemic (Grose, 2020).
the background of this study, followed by a discussion of the Before the pandemic, nearly 35 million children benefited
survey and a section of descriptive statistics. We then outline daily from the federal school meal programs (USDA-FNS,
the econometric analysis and present the results. A final sec- 2020b). With school closures, there was no federal mandate
tion concludes and discusses the findings. that schools should offer meal service. However, under the
Food insecurity during the COVID-19 pandemic: evidence from a survey of low-income Americans 167
FFCA, state agencies were allowed to disburse EBT benefits have gross income below 130% FPL. However, the “broad-
for children who qualify for free or reduced-price meals via based categorical eligibility” lets states import up to 200%
the Pandemic EBT program (Dunn et al., 2020). Some dis- FPL given TANF-funded program (USDA FNS, 2020a).
tricts continued meal delivery or pickup during pandemic- Respondents from households who reported an income higher
related school closures. Schools could temporarily waive meal than 200% of the FPL were excluded from this study.2
time requirements to make it easier to pick up meals for mul- We modified the 10-item US Adult Food Security Survey
tiple-days. The Act also allowed states, on an individual state- Module (USDA-FNS, 2020e) to elicit information about re-
by-state basis, to serve free meals to children in all schools spondents’ food insecurity status during the past 90 days.
(US Department of Labor, 2020). While the module has been validated for one month and one
The outbreak of the coronavirus at food processing plants year, it has not specifically been validated for three months.
affected the food supply (Gallagher & Kirkland, 2020). As a Given the timing of the survey, we assessed three months to
result, consumers saw an increase in retail food prices in April, best capture the respondents experience since the beginning of
led by meats and eggs (Johansson, 2020). In the meantime, the the pandemic. This module is comprised of indicator ques-
demand for food remained high after peaking in March tions that capture and distinguish the severity of food insecu-
(Information Resources, Inc., 2020). Although the food sup- rity. Questions are ordered by severity and include three levels
ply seemed to remain robust (Johansson, 2020), consumers of screening for adults. The USDA divides households into
saw empty shelves because of the elevated demand for staple the following categories: high food secure (all household
goods, supply chain challenges, and stockpiling behaviors members had access at all times to enough food for an active,
(Information Resources, Inc., 2020). Families living in neigh- healthy life), marginal food secure (some members reported
borhoods with limited access to grocery stores, e.g., food de- anxiety about food sufficiency or shortage of food in the
serts, may have experienced additional difficulties due to busi- house), low food secure (at least some household members
ness closures and modifications to transit services (Siddiqi reported reduced quality, variety), and very low food secure
et al., 2020). Empty shelves may be especially harmful to (one or more household members reported multiple indica-
families who rely on WIC, which limits benefits to specific tions of disrupted eating patterns and reduced food intake)
foods, as was already documented in (Grose, 2020). (Gundersen & Ziliak, 2015). Affirmative responses to ques-
Another way that households were able to maintain ade- tions were summed to one of the following four categories:
quate access to food was through online shopping (Chang & high food secure (no affirmative responses), marginal food
Meyerhoefer, 2020; Reardon et al., 2021). Redman (2020) secure (1–2 affirmative responses), low food secure (3–5 af-
reported that Walmart’s online grocery sales increased 74% firmative responses), and very low food secure (6 or more
in April. However, online shopping may not be feasible for affirmative responses). For purposes of this study, respon-
some SNAP recipients. At the beginning of the pandemic, dents in the low and very low food secure categories were
low-income SNAP recipients were prohibited from using their considered to be food insecure (USDA-FNS, 2020e).
benefits to purchase foods online. Therefore, many SNAP In this survey, we inquired about SNAP participation.
recipients could not order groceries to be delivered and were Specifically, we first asked whether the household ever re-
thus required to shop in-person, thereby putting themselves ceived SNAP and then followed up by asking when SNAP
and their families at greater risk and furthering accentuating was last received. The monthly SNAP benefit amount re-
income disparities in the effects of COVID-19 (Rosenbaum ceived before the pandemic, whether additional SNAP bene-
et al., 2020). This is now changing through the SNAP Online fits were received and the amount of additional SNAP bene-
Purchasing Pilot program (USDA-FNS, 2020d). However, fits. We also asked questions on how easy the respondent
the pilots are limited to selected retailers in participating states. could access SNAP and whether they had trouble receiving
Therefore, many recipients may not benefit from online pur- SNAP remotely during the pandemic. WIC participation, re-
chases due to availability as well as differences in timing of mote benefit collection, and whether the respondent was able
program roll-out (Rosenbaum et al., 2020). to find WIC foods in stores were also included in the survey.
In addition, we inquired about school meal participation in
2.1 The survey 2019, alternative meals provided by schools during the pan-
demic, and how frequently alternative school meals were
We designed a web-based survey targeting low-income adults utilized.
in the US. The survey was distributed to respondents using The survey also investigated whether someone in the
Dynata, an online platform. Dynata also allows researchers to household lost his/her job due to the pandemic. If the answer
screen respondents based on an income criterion.. In our was affirmative, we then asked whether the respondent had
study, we limited the sample to low-income adults in house-
holds at or below the 200% the FPL. Regular federal SNAP 2
For example, $52,400 would place a four-person household at 200% of the
rules require families with no elderly or disabled members to FPL in 2020.
168 Fang D. et al.
received unemployment benefits. Food availability was in- ethnicity. Food desert status, residential population, urbanity,
quired on whether or not the respondent switched stores due and marital status, however, did not differ significantly by
to unavailability, whether price increases were observed, and food security status.
whether online grocery purchases were an option. If the re- Based on the findings in Table 1, we created eight sub-
spondent indicated switching stores due to unavailability of samples: 1) a SNAP eligible sample based on income; 2)
foods, additional questions were asked about food availability, households with children; 3) households without children; 4)
whether price increases were observed, and whether online households experiencing a pandemic-related job loss; 5)
grocery purchases were an option. The food desert status of households who did not experience a pandemic-related job
the respondent was determined by linking the geographic co- loss; 6) African American households; 7) Hispanic house-
ordinates based on the respondent’s IP address to the census holds; and 8) Caucasian households.
tract, which is the geographic unit available in the USDA’s Approximately 45% of the full sample (n = 1214) reported
Food Environment Atlas data (Food Environment Atlas, being on SNAP, with an average monthly benefit of $241.6.
2020). Finally, a set of socio-demographic questions conclud- The average SNAP benefit is for all those received SNAP.
ed the survey. However, the expansion (or additional SNAP) is only for
those who were not already at their maximum. Among the
2.2 Descriptive statistics SNAP recipients, only 10.2% reported receiving additional
SNAP benefits with an average additional benefit of $101.8.
The survey was conducted from June 29, 2020 to July 21, Because SNAP is an entitlement program for those who are
2020. The number of income-eligible individuals completing income eligible, we constructed the SNAP eligible sample
the survey was 2772. Some participants were excluded due to based on income (i.e., under 130% of the FPL). Since income
inconsistent reporting of ZIP Code. To ensure data integrity, was reported as a category in our survey, we calculated the
we also restricted our sample based on the following criterion: upper and lower bounds of SNAP eligible income based on
1) age under 100 years old; 2) a maximum household size of household size. We then assigned respondents to be ineligible,
10; 3) maximum SNAP benefit given household size; 4) max- definitely eligible, and possibly eligible depending on where
imum stimulus payment given marital status and number of they fell within the category. The SNAP-eligible sample con-
children; and 5) driving time to secure groceries of under an sists of only the 1177 respondents who were definitely
hour. This sample drew respondents from all 50 states plus the eligible.
District of Columbia. Hence, our analysis sample consists of
2714 observations.
As explained above, we measured the four levels of food 2.3 Statistical analysis
insecurity based on the USDA’s guideline from the 10-item
food security component of the survey (USDAFNS, 2020e). To explore the associations between food insecurity and food
Table 1 characterizes our sample in terms of these four levels. assistance programs during the pandemic, we estimated struc-
Overall, 48.4% of this sample was food secure and 51.6% tural equation models (SEM).3 The dependent variable is the
were food insecure. As noted above, respondents were classi- binary food insecurity variable which is estimated simulta-
fied as “food insecure” if they fell into the low (17.39%) or neously with program participation in SNAP, WIC, school
very low (34.23%) food security categories. Food insecurity meals, receipt of alternative school meals, and the receipt of
in 2019 among households below the 185 FPL was 28% community foods. Specifically, we additionally estimate
(USDA-ERS, 2020) so incidence in our sample is much larger equations for SNAP, WIC, School Meals, and Community
than this national figure. Some of this difference is likely due Foods. Each of these equations includes food insecurity as a
to economic hardship brought on by the pandemic, but it is regressor along with all other covariates. The motivation for
also important to keep in mind that our survey asked about estimating the equations in a system is to correct for correlated
food insecurity in the last three months as opposed to the last and unobserved effects in the errors (Greene, 2008) that could
month. jointly determine food insecurity and program participation.
Table 1 also presents p-values from tests to examine het- We do not claim causal inference from the SEM estimates.
erogeneity in selected variables with respect to food security However, the estimation should provide more accurate asso-
status. Results show there were significant differences be- ciations than single-equation linear probability or logit
tween food insecure and food secure respondents in terms of 3
It is well documented that the selection issue makes it difficult to identify a
federal and local nutrition programs, stimulus payments, un- causal relationship between SNAP participation and food insecurity (Ratcliffe
employment benefits, pandemic related job loss, income sta- & McKernan, 2010). Past literature has employed instrumental variables (IV)
bility, number of children, the ability to purchase groceries to attempt to tease out the causal effects. Common IVs for SNAP participation
include state variation in eligibility (Bartfeld & Dunifon, 2006; Ratcliffe &
online, observed food price increases, and driving time to McKernan, 2010). However, this is not applicable because COVID-related
the store. There were also significant differences by race and SNAP expansion has completed for almost all States by the time of survey.
Food insecurity during the COVID-19 pandemic: evidence from a survey of low-income Americans 169
Table 1 Food Security Status
All High Security Marginal Security Low Security Very low Security
N (%) N (%) N (%) N(%) N (%) p value
Total 2714(100) 866(31.91) 447(16.47) 472(17.39) 929(34.23)
SNAP
Yes 1214(45) 280(23.06) 188(15.49) 235(19.36) 511(42.09) <0.001
No 1500(55) 586(39.07) 259(17.27) 237(15.8) 418(27.87)
Additional SNAP
Yes 662(24) 157(23.72) 104(15.71) 138(20.85) 263(39.73) <0.001
No 2052(76) 709(34.55) 343(16.72) 334(16.28) 666(32.46)
WIC
Yes 249(9) 33(13.25) 24(9.64) 48(19.28) 144(57.83) <0.001
No 2465(91) 833(33.79) 423(17.16) 424(17.2) 785(31.85)
School Meals
Yes 533(20) 60(11.26) 71(13.32) 102(19.14) 300(56.29) <0.001
No 2181(80) 806(36.96) 376(17.24) 370(16.96) 629(28.84)
Alternative School Meals
Yes 453(17) 52(11.48) 57(12.58) 82(18.1) 262(57.84) <0.001
No 2261(83) 814(36) 390(17.25) 390(17.25) 667(29.5)
Charitable Foods
Yes 839(31) 132(15.73) 109(12.99) 144(17.16) 454(54.11) <0.001
No 1875(69) 734(39.15) 338(18.03) 328(17.49) 475(25.33)
Lost Job
Yes 534(20) 77(14.42) 46(8.61) 96(17.98) 315(58.99) <0.001
No 2180(80) 789(36.19) 401(18.39) 376(17.25) 614(28.17)
Received UI
Yes 276(10) 68(24.64) 42(15.22) 53(19.2) 113(40.94) 0.019
No 2438(90) 798(32.73) 405(16.61) 419(17.19) 816(33.47)
Received Stimulus
Yes 1997(74) 719(36) 328(16.42) 314(15.72) 636(31.85) <0.001
No 717(26) 147(20.5) 119(16.6) 158(22.04) 293(40.86)
Online Shopping
Yes 940(35) 258(27.45) 143(15.21) 154(16.38) 385(40.96) <0.001
No 1774(65) 608(34.27) 304(17.14) 318(17.93) 544(30.67)
Food Dessert
Yes 374(14) 118(31.55) 62(16.58) 61(16.31) 133(35.56) 0.904
No 2332(86) 747(32.03) 383(16.42) 410(17.58) 792(33.96)
Note: p values indicate the x2 test between subsample who answered “Yes” and those who answered “No”
models. We did not include state fixed effects in these models charitable food programs. About 22.3% of the respondents
because we only have a few observations in certain states, in the SNAP-eligible sample reported pandemic related job
hence including state fixed effects will introduce singularity loss, compared to 19.7% in the overall sample. However, the
in the subsamples. Also, the lockdown policies were generally percentage receiving unemployment benefits is comparable.
mild during the sampling period, so that the different states are Food access variables are comparable between the SNAP
more or less comparable. The system is estimated using a sample and the overall sample with similar statistics on dis-
generalized least square estimator (GLS) and estimates are tance to store, the ability to shop online, and food desert status.
produced by the systemfit package in R (Henningsen et al., The SNAP sample has a higher proportion of Black and
2007). Hispanic respondents compared to the overall sample, consis-
tent with national statistics (USDA-ERS, 2020). Overall, our
data show that the SNAP sample is of lower-income and less
3 Results financially secure.
We also compare households who lost jobs due to the pan-
As shown in Table 2, the average food insecurity score is demic to those who did not. About 19.7% (n = 534) reported
higher (meaning more food insecure) in the SNAP eligible unemployment due to the pandemic. Food insecurity is much
sample, with 61.4% of respondents being food insecure, com- higher for the unemployed sample, with 77% being food in-
pared to 51.6% in the overall sample. Among SNAP eligible secure compared to 45.4% in the employed sample. Among
respondents, we also observe a higher percentage of partici- those who lost jobs, 31.3% received unemployment benefits.
pation in other food assistance programs, namely WIC, school The unemployed sample reported a lower level of income
meals, alternative school meals during the pandemic, and stability (40.4%) compared to the employed sample (69%).
170 Fang D. et al.
They also reported a higher level of SNAP participation than Results for the SEM are presented in Table 3. Appendix
those who were employed, consistent with the finding of Tables 4 and 5 present estimates from the single equation
(Bitler et al., 2020). Those who lost jobs were younger in models: linear probability models and logit models, respec-
age (36.4 years old compared to 49.3 years) and were less tively for interested readers. Estimates from the single equa-
educated. This is consistent with the notion that most layoffs tion models presented in the appendix were similar results in
are temporary jobs assumed by lower-skilled or less experi- terms of sign and significance. To facilitate comparison across
enced workers (Ganong et al., 2020). There was also a higher the SEM and single-equation model specifications, key asso-
proportion of respondents reporting one or more children ciations are reported as marginal effects in Appendix Figs. 1 to
among the unemployed sample. This could be due to the fact 9.
that people had to leave jobs to take care of children when
schools closed due to the pandemic. 3.1 SNAP and WIC
Table 2 also compares respondents with children to those
without. Approximately 32.7% of our sample (n = 887) had Given the recent onset of the pandemic, the loss of a job may
children, with an average of 1.8 children per household. Food have been a large enough disruption to induce food insecurity
insecurity was higher among respondents with children com- regardless of the respondents’ prior SNAP status. It is also
pared to those without (67% compared to 44.2%). possible that recently unemployed who were newly eligible
Respondents with children also reported a much higher par- for SNAP had not yet received benefits. The association be-
ticipation in SNAP, WIC, and school meals, consistent with tween receipt of additional SNAP benefits and food insecurity
the national averages (USDA-FNS, 2020a; USDA-FNS, is also negative but is less consistent across the different
2020c). Respondents with children also reported a higher uti- subsamples.
lization of charitable foods compared to those without chil- The association between receipt of additional SNAP bene-
dren. In our sample, households with children are less income fits and food insecurity is statistically different from zero in the
stable (48.4%, compared to 70.7%) and as noted above were full sample and is particularly strong among households with
more likely to be unemployed (29.7% compared to 14.8%). children and may reflect additional benefits via the Pandemic
The proportion of Hispanics is higher among the subsample EBT. Interestingly, there is no significant association between
with children. additional SNAP benefits and food insecurity among the
Finally, we also examined whether there is racial dispar- SNAP-eligible sample. Because membership in this sample
ity in our sample in terms of food insecurity. Table 2 shows is contingent on income, it is possible that many SNAP recip-
the subsamples of Blacks, Hispanics, and Whites. Food ients in this subsample were already near the maximum SNAP
insecurity is highest among the Hispanic sample (57.4%) benefit allotment and hence were not meaningfully impacted
and lowest in the White sample (46.2%). The Black sample by the additional SNAP benefits provided by the emergency
reported the highest SNAP participation at 50.6%, follow- COVID-19 legislation.
ed by Whites (42.8%) and then Hispanics (41.5%). WIC Finally, WIC participation is not strongly associated with
participation and pre-pandemic participation in school food insecurity in our data. As in the case of participation in
meals, post-pandemic participation in alternative school SNAP or community food programs, there could be
meal delivery programs, and use of charitable food assis- bidirectionality in this association wherein eligible food-
tance was also higher among respondents in the Black insecure households are more likely to enroll in WIC.
sample. Our data show that Blacks and Hispanics were less Evidence of this is provided by the strong positive association
income stable, 58.6% and 52.8% respectively, compared to between WIC participation and food insecurity within the
Whites at 73.9%. However, 81.2% of Whites received the Hispanic subsample. Historically, Hispanic participation and
stimulus payment, whereas only 69.5% of Blacks and coverage rates have exceeded other ethnic groups (USDA-
65.8% of Hispanics did. Nichols and Simms (2012) argues ERS, 2020).
that after adjusting for education, past employment, and
reasons for unemployment, Black and Hispanic workers 3.2 School meals
are significantly less likely to receive unemployment ben-
efits than their White counterparts. This is not the case in School meals are positively associated with food insecurity in
our sample. Our sample reflected higher utilization of un- the full sample and across most subsamples. Recall that school
employment benefits among Blacks (11.9%) and Hispanics meals is measured as an indicator of whether anyone in the
(14.4%) than Whites (6.5%), which suggests COVID- household received school meals during 2019, a period unaf-
related job losses disproportionately impacted minorities fected by the pandemic. Hence, the positive association is
in our sample. We also note that 17.8% of Blacks reside consistent with the loss of school meals during the pandemic
in a food desert, compared to 13.9% for White and 9% for contributing to higher food insecurity. This association is par-
Hispanics. ticularly strong and significant among the SNAP-eligible
Food insecurity during the COVID-19 pandemic: evidence from a survey of low-income Americans 171
Table 2 Descriptive Statistics
Full Sample (N =2714)
Variables mean s.d.
Food insecurity status 0.516 0.500
SNAP 0.447 0.497
Additional SNAP 0.244 0.430
WIC 0.092 0.289
School Meals 0.196 0.397
Alternative School Meals 0.167 0.373
Charitable Foods 0.309 0.462
Lost Job 0.197 0.398
Income Stable 0.634 0.482
Unemployment Benefit 0.102 0.302
Receive Stimulus 0.736 0.441
Online Shopping 0.346 0.476
Food Dessert 0.138 0.345
Female 0.727 0.446
Age 46.784 18.626
Married 0.238 0.426
Black 0.316 0.465
Hispanic 0.241 0.428
White 0.416 0.493
College and Higher 0.188 0.391
High School 0.347 0.476
Less High School 0.050 0.219
Some College 0.415 0.493
Less $15,000 0.343 0.475
$15,000–$29,999 0.445 0.497
$30,000–$44,999 0.174 0.379
$45,000 and Above 0.038 0.191
SNAP Eligiblea Lost Job Has Job With Children Without Children
Sample Size 1177 534 2180 887 1872
mean s.d. mean s.d. mean s.d. mean s.d. mean s.d.
Food Insecurity (binary) 0.614 0.487 0.770 0.421 0.454 0.498 0.670 0.471 0.442 0.497
SNAP 0.569 0.495 0.536 0.499 0.426 0.495 0.563 0.496 0.391 0.488
Additional SNAP 0.304 0.460 0.281 0.450 0.235 0.424 0.337 0.473 0.199 0.399
WIC 0.118 0.323 0.187 0.390 0.068 0.252 0.218 0.413 0.031 0.172
School Meals 0.223 0.416 0.350 0.477 0.159 0.365 0.479 0.500 0.059 0.236
Alternative School Meals 0.198 0.399 0.309 0.463 0.132 0.339 0.409 0.492 0.049 0.216
Charitable Foods 0.359 0.480 0.457 0.499 0.273 0.446 0.384 0.487 0.273 0.445
Lost Job 0.223 0.417 0.297 0.457 0.148 0.356
Income Stable 0.572 0.495 0.404 0.491 0.690 0.463 0.484 0.500 0.707 0.455
Unemployment Benefit 0.103 0.304 0.313 0.464 0.050 0.218 0.135 0.342 0.085 0.280
Receive Stimulus 0.667 0.472 0.693 0.462 0.746 0.435 0.708 0.455 0.749 0.434
Online Shopping 0.336 0.473 0.429 0.495 0.326 0.469 0.408 0.492 0.316 0.465
Food Dessert 0.144 0.352 0.139 0.346 0.138 0.345 0.152 0.359 0.132 0.338
Female 0.744 0.436 0.727 0.446 0.727 0.446 0.806 0.396 0.689 0.463
Age 42.661 17.674 36.423 15.224 49.322 18.507 36.230 13.430 51.908 18.643
Married 0.163 0.370 0.215 0.411 0.243 0.429 0.327 0.469 0.194 0.396
Black 0.351 0.477 0.335 0.473 0.311 0.463 0.307 0.461 0.320 0.467
172 Fang D. et al.
Table 2 (continued)
Full Sample (N = 2714)
Variables mean s.d.
Hispanic 0.260 0.439 0.337 0.473 0.217 0.412 0.340 0.474 0.192 0.394
White 0.360 0.480 0.290 0.454 0.447 0.497 0.317 0.465 0.465 0.499
College and Higher 0.149 0.356 0.165 0.371 0.194 0.396 0.166 0.372 0.199 0.400
High School 0.387 0.487 0.356 0.479 0.344 0.475 0.359 0.480 0.341 0.474
Less High School 0.088 0.284 0.047 0.211 0.051 0.221 0.061 0.239 0.045 0.208
Some College 0.376 0.485 0.433 0.496 0.410 0.492 0.415 0.493 0.414 0.493
Less $15,000 0.791 0.407 0.375 0.484 0.335 0.472 0.275 0.447 0.376 0.485
$15,000–$29,999 0.172 0.378 0.397 0.490 0.456 0.498 0.374 0.484 0.479 0.500
$30,000–$44,999 0.036 0.186 0.189 0.392 0.171 0.376 0.256 0.437 0.135 0.341
$45,000 and Above 0.001 0.029 0.039 0.195 0.038 0.190 0.095 0.293 0.010 0.101
Black Hispanic White
Sample Size 857 653 1130
mean s.d. mean s.d. mean s.d.
Food Insecurity (binary) 0.538 0.499 0.574 0.495 0.462 0.499
SNAP 0.506 0.500 0.415 0.493 0.428 0.495
Additional SNAP 0.289 0.454 0.211 0.409 0.238 0.426
WIC 0.123 0.328 0.109 0.312 0.055 0.228
School Meals 0.235 0.424 0.250 0.433 0.135 0.341
Alternative School Meals 0.197 0.398 0.228 0.420 0.102 0.302
Charitable Foods 0.372 0.484 0.323 0.468 0.245 0.430
Lost Job 0.209 0.407 0.276 0.447 0.137 0.344
Income Stable 0.586 0.493 0.528 0.500 0.739 0.439
Unemployment Benefit 0.119 0.324 0.144 0.351 0.065 0.246
Receive Stimulus 0.695 0.460 0.658 0.475 0.812 0.391
Online Shopping 0.342 0.475 0.377 0.485 0.329 0.470
Food Dessert 0.178 0.382 0.090 0.286 0.139 0.346
Female 0.750 0.433 0.727 0.446 0.715 0.452
Age 44.076 18.492 37.894 15.901 54.367 17.292
Married 0.168 0.374 0.257 0.437 0.273 0.446
College and Higher 0.181 0.385 0.181 0.385 0.190 0.393
High School 0.345 0.476 0.308 0.462 0.378 0.485
Less High School 0.054 0.226 0.060 0.237 0.041 0.198
Some College 0.420 0.494 0.452 0.498 0.391 0.488
Less $15,000 0.400 0.490 0.326 0.469 0.312 0.463
$15,000–$29,999 0.433 0.496 0.383 0.486 0.491 0.500
$30,000–$44,999 0.128 0.335 0.245 0.430 0.165 0.371
$45,000 and Above 0.039 0.193 0.046 0.210 0.033 0.178
a.
Since income was reported as a category in our survey, we calculated the upper and lower bounds of SNAP eligible income based on household size.
We then assigned respondents to be ineligible, definitely eligible, and possibly eligible depending on where they fell within the category. The SNAP-
eligible sample consists of only the respondents who were definitely eligible
sample. Based on income, children in these households would related school closures. In our sample, taking advantage of
be eligible for free school meals which may explain the strong these alternative school meals is negatively associated with
association. food insecurity across most subsamples but the magnitudes
In addition to Pandemic EBT, many school districts pro- of the effect are small and insignificant. The one exception
vided alternative meals for pickup or delivery during COVID- is the SNAP-eligible sample. Because this sample is
Food insecurity during the COVID-19 pandemic: evidence from a survey of low-income Americans 173
Table 3 SEM Results
Full SNAP Eligiblea Lost Job Job Children No Children Black Hispanic White
Intercept 0.657*** 0.682*** 0.895*** 0.616*** 0.464*** 0.779*** 0.759*** 0.502*** 0.849***
(0.077) (0.112) (0.150) (0.090) (0.127) (0.098) (0.104) (0.134) (0.094)
SNAP 0.241*** 0.237*** 0.021 0.302*** 0.331*** 0.217*** 0.167*** 0.360*** 0.216***
(0.028) (0.040) (0.051) (0.033) (0.046) (0.039) (0.046) (0.063) (0.045)
Additional SNAP −0.076** −0.036 −0.051 −0.089** −0.125** −0.059 −0.033 −0.104 −0.084*
(0.026) (0.036) (0.048) (0.030) (0.041) (0.034) (0.044) (0.057) (0.040)
WIC −0.014 −0.032 −0.001 0.026 0.000 0.113 −0.024 0.142* −0.055
(0.032) (0.046) (0.050) (0.041) (0.037) (0.069) (0.054) (0.062) (0.060)
School Meals 0.055 0.127** 0.053 0.078* 0.062 −0.069 0.078 −0.065 0.149**
(0.029) (0.043) (0.050) (0.036) (0.034) (0.053) (0.050) (0.053) (0.056)
Alternative Meals −0.040 −0.117** −0.007 −0.062 0.009 −0.045 0.035 −0.082
(0.030) (0.044) (0.050) (0.037) (0.035) (0.052) (0.055) (0.058)
Charitable Food 0.290*** 0.233*** 0.248*** 0.297*** 0.191*** 0.332*** 0.251*** 0.232*** 0.386***
(0.020) (0.029) (0.037) (0.023) (0.033) (0.025) (0.034) (0.042) (0.031)
Lost Job 0.143*** 0.148*** 0.174*** 0.119*** 0.183*** 0.168*** 0.043
(0.024) (0.035) (0.034) (0.033) (0.041) (0.045) (0.042)
Income Stable −0.129*** −0.095** −0.089* −0.133*** −0.151*** −0.122*** −0.123*** −0.166*** −0.097**
(0.019) (0.030) (0.035) (0.023) (0.031) (0.025) (0.034) (0.038) (0.033)
Rec. UI −0.114*** −0.086 −0.155*** −0.088 −0.165*** −0.076 −0.168** −0.055 −0.080
(0.030) (0.046) (0.038) (0.046) (0.045) (0.041) (0.051) (0.056) (0.055)
Rec. Stimulus −0.037 −0.107*** −0.034 −0.044 −0.029 −0.031 0.004 0.003 −0.122***
(0.021) (0.030) (0.039) (0.024) (0.034) (0.027) (0.035) (0.041) (0.035)
Online Shopping 0.006 −0.000 0.029 0.004 0.002 0.001 −0.018 0.007 0.028
(0.018) (0.029) (0.036) (0.021) (0.031) (0.023) (0.034) (0.038) (0.028)
Price Increase 0.137*** 0.138*** 0.122** 0.140*** 0.139*** 0.138*** 0.090* 0.131** 0.184***
(0.021) (0.031) (0.043) (0.024) (0.035) (0.026) (0.038) (0.044) (0.032)
Food Dessert −0.015 −0.037 0.055 −0.031 0.005 −0.033 −0.054 0.003 0.033
(0.025) (0.038) (0.049) (0.028) (0.041) (0.031) (0.040) (0.064) (0.037)
Female 0.001 0.015 −0.007 0.005 −0.023 0.004 −0.041 0.059 −0.002
(0.019) (0.031) (0.039) (0.022) (0.038) (0.023) (0.036) (0.041) (0.029)
Age −0.007*** −0.006*** −0.005*** −0.007*** −0.005*** −0.007*** −0.007*** −0.005*** −0.007***
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Married −0.004 0.016 −0.098* 0.023 −0.026 0.016 −0.011 −0.036 0.026
(0.021) (0.038) (0.043) (0.024) (0.033) (0.028) (0.042) (0.044) (0.031)
Black 0.016 0.003 0.004 0.028 0.038 −0.009
(0.053) (0.081) (0.095) (0.064) (0.082) (0.071)
Hispanic 0.024 −0.006 0.025 0.029 0.046 −0.004
(0.054) (0.082) (0.093) (0.065) (0.081) (0.072)
White 0.095 0.054 0.014 0.121 0.139 0.055
(0.053) (0.082) (0.096) (0.064) (0.082) (0.071)
Child 3 More 0.178** 0.116 −0.022 0.273*** 0.123 0.200 0.168 0.112
(0.063) (0.085) (0.098) (0.079) (0.063) (0.114) (0.124) (0.113)
High School −0.055 −0.021 −0.022 −0.056 0.057 −0.113* −0.041 0.054 −0.169*
(0.041) (0.049) (0.084) (0.046) (0.066) (0.052) (0.070) (0.082) (0.067)
Some College −0.006 0.002 −0.062 0.018 0.109 −0.066 0.012 0.035 −0.098
(0.040) (0.050) (0.084) (0.046) (0.066) (0.052) (0.070) (0.080) (0.068)
College Higher −0.020 −0.028 −0.039 −0.003 0.075 −0.068 0.018 0.022 −0.120
(0.043) (0.057) (0.090) (0.049) (0.072) (0.055) (0.076) (0.088) (0.072)
Inc: $15,000–$29,999 −0.046* −0.008 −0.047 −0.046* −0.000 −0.062** −0.092** 0.005 −0.012
174 Fang D. et al.
Table 3 (continued)
Full SNAP Eligiblea Lost Job Job Children No Children Black Hispanic White
(0.020) (0.040) (0.040) (0.023) (0.038) (0.024) (0.034) (0.044) (0.031)
Inc: $30,000–$44,999 −0.099*** 0.001 −0.018 −0.116*** −0.056 −0.129*** −0.156** −0.058 −0.087*
(0.027) (0.081) (0.050) (0.031) (0.043) (0.035) (0.051) (0.050) (0.044)
Inc: >$45,000 −0.095* −0.363 −0.048 −0.112* −0.052 −0.158 −0.057 −0.036 −0.197*
(0.048) (0.457) (0.092) (0.056) (0.058) (0.104) (0.083) (0.093) (0.082)
N 2706 1170 534 2172 883 1823 856 648 1128
Notes: coefficients are marginal effects. Standard errors are in the parentheses. *,**, and *** indicate significance levels at 0.5, 0.1 and 0.01, respectively
a.
Since income was reported as a category in our survey, we calculated the upper and lower bounds of SNAP eligible income based on household size.
We then assigned respondents to be ineligible, definitely eligible, and possibly eligible depending on where they fell within the category. The SNAP-
eligible sample consists of only the respondents who were definitely eligible
comprised of low-income households, this provides some ev- unemployment benefits had a much stronger negative associ-
idence that alternative meal delivery programs were helping ation with food insecurity than did stimulus payments.
lower-income families during the pandemic. The fact that the
association is significant only among the SNAP-eligible sam- 3.4 Other covariates
ple provides some evidence that alternative meal delivery pro-
grams were not sufficient to make up for the benefit of having Other controls in the model can shed light on how food access
meals in school among the lowest income households. and food procurement may relate to food insecurity. Residing
The associations between school meals are not significant in a food desert was not significantly associated with food
among the subsample of households with children, but these insecurity in our data. 34.6% of our respondents were able
households were more likely to receive Pandemic EBT sup- to get groceries online. However, the estimates in Table 3
plements. This may have limited the impact of losing access to show that online shopping is not meaningfully associated with
meals in the school environment and lessened their reliance on food insecurity. On the other hand, there is a small but positive
alternative school meal delivery and pickup programs. association between driving distance to the store and food
Moreover, data presented earlier in Table 2 indicate that nearly insecurity. This association is significant for the full sample
as many households with children that reported utilizing and some of the subsamples and could indicate that access to
school meals before the pandemic took advantage of alterna- stores during the pandemic impacted the ability of some re-
tive meal delivery and pickup programs. spondents to obtain food. Notably, 77.8% of our respondents
saw food price increases in their primary store, and this is
3.3 Direct cash assistance strongly and positively associated with food insecurity.
These controls are also informative on the role of house-
The primary cash assistance programs in our data are the un- hold structure. Among the sample who did not experience
employment benefits and stimulus payments. Across the full COVID-19 related job loss, having more than three children
sample and all subsamples, receipt of unemployment benefits is significantly associated with food insecurity. It is possible
is negatively associated with food insecurity with the strongest that some of these households may have had to cut back work
associations and significant associations are among those who hours to take care of children who would otherwise have been
lost a job due to the pandemic, households with children, and in school. For those who did lose jobs, being married is neg-
Black households. These strong associations suggest that un- atively associated with food insecurity. Finally, age is nega-
employment benefits are helping to alleviate food insecurity tively associated with food insecurity, likely due to the accu-
and are consistent with recent findings by Bitler et al. (2020). mulation of savings. Support for this conjecture is provided by
The association in the Black sample is also consistent with the strong negative association between the income stability
Stone (2020) who finds the unemployment benefit was able indicator and food insecurity across all subsamples.
to reduce the disparities in many southern states, with large
Black populations, that have weak unemployment insurance
(UI) systems. Receipt of economic stimulus payments is also 4 Discussion
negatively associated with food insecurity but this association
is significant only among the SNAP-eligible sample and sam- Economic hardship brought on by the COVID-19 pandemic
ple of White households. Economic assistance in the form of has increased food insecurity. This study explored how food
Food insecurity during the COVID-19 pandemic: evidence from a survey of low-income Americans 175
and economic assistance programs are associated with food Our data highlight the importance of school meal pro-
insecurity during very difficult times. Overall, our data show grams during normal times. Those in our survey that took
strong associations between assistance programs and food in- advantage of school meals before the outbreak were more
security and point to several insights for food security and likely to have experienced food insecurity after the
policy. pandemic-related school closures. Short-term solutions
One implication is that there is a need to ensure that those such as alternative meal delivery or pickup programs,
newly at risk for food insecurity are able to connect to re- however, did not significantly decrease the likelihood of
sources. Across our subsamples, those who lost jobs due to food insecurity except among the lowest-income respon-
the pandemic had the highest level of food insecurity but the dents. Our results provide evidence that Pandemic EBT,
weakest associations with food assistance programs. Among designed to help offset the cost of meals that would oth-
this group, the only significant association was with the re- erwise have been provided in the schools, could be more
ceipt of unemployment benefits, something most newly un- effective at reducing food insecurity among households
employed would know are available to them. The point is that with children. The impact of emergency policy responses
food assistance programs may not be reaching those who have on health outcomes should also be investigated in future
great need but little prior experience with food benefits. research. For example, shifting the responsibility of feed-
Hence, outreach to help these individuals better access food ing children from school to parents may lead to negative
assistance is needed. health outcomes because school meals have been shown
Another implication is that targeted assistance may be more to improve the diets of children from homes with poor
effective than general assistance when it comes to food inse- food environments (Smith, 2017).
curity. For example, extending unemployment benefits may Although recent studies have reported stockpiling behav-
be a more effective strategy than additional broad-based cash iors among consumers during the early phases of the pandem-
transfers. Extending unemployment benefits would also be ic (Grose, 2020; Information Resources, Inc., 2020), the ma-
more cost effective by concentrating assistance to those in jority of our respondents had reported no trouble finding
greatest need. Another example of effective targeting is with foods, even WIC food items. 77.8% of our respondents, how-
the Pandemic EBT Program to replace school meals. Behind ever, have reported price increase in their primary stores, and
the newly unemployed, households with children experienced this has contributed to higher food insecurity. We did not find
the next highest level of food insecurity in our data and results food access, i.e. food desert status or distance to the store, to
presented above suggest that additional EBT transfers to these be associated with food insecurity. Online food shopping was
families helps reduce food insecurity. not associated with food insecurity. Online grocery industry
The SNAP expansion is an example of a policy that seems was accelerated by the pandemic (Reardon et al., 2021).
to be not meeting the neediest Americans. This policy expand- However, the delayed rollout of SNAP online pilots by the
ed benefits for those who were not already receiving the max- time of the survey and food access issue for SNAP partici-
imum benefit. In this respect, it did not impact those SNAP pants (Fitzpatrick et al., 2016) may have contributed to this
recipients who were in the worst economic condition prior to small percentage.
the policy change. We find little evidence that the SNAP ex- This study examined how the food assistance and eco-
pansion reduced food insecurity. The simplest explanation is nomic assistance programs are associated with food in-
that it had a limited impact on those at greatest economic security in the US. While we cannot estimate the causal
disadvantage. In contrast, evidence of a negative association effects of these programs on food insecurity, we are able
between food insecurity and SNAP expansion was observed to show important associations that provide useful in-
among groups with higher levels of income stability including sights into how these programs could alleviate food in-
non-minority households and those not experiencing a job security during catastrophic periods such as during pan-
loss. Again, this is consistent with the SNAP expansion not demics. In the absence of a large-scale nationwide data-
impacting Americans most at risk for food insecurity. In fact, base on food insecurity, our study provides timely and
this conclusion is consistent with the Biden Administration’s important insights into this issue during the pandemic.
decision in December 2020 to increase SNAP benefits by 15% This is the first step in understanding how the food and
from January through June 2021 (Rosenbaum, 2021). On the economic assistance initiatives helped alleviate food in-
other hand, food insecurity is still an issue among low-income security during the pandemic. Future studies should ex-
but more financially stable households. For example, 45% of amine whether any of these initiatives had a causal effect
those not experiencing a job loss reported food insecurity on the incidence of food insecurity in the US. A timely
during the first months of the pandemic. Workers in these data source to study a similar topic during the pandemic
households may have lost hours or income even if they con- is the Household Pulse Survey by the Census, which
tinued to be employed or may struggle to purchase foods as incorporated questions on food insufficiency and food
school meals become less available. hardship.
176 Fang D. et al.
Appendix
Table 4 Liner Probability Model Estimates
Full SNAP Lost Job Employed Children No Children Black Hispanic White
SNAP 0.047** 0.035 −0.047 0.080*** 0.104** 0.044 0.025 0.106** 0.044
(0.024) (0.031) (0.042) (0.028) (0.041) (0.034) (0.040) (0.053) (0.038)
Add. SNAP 0.003 0.007 −0.022 0.009 −0.021 −0.004 −0.010 0.040 −0.022
(0.027) (0.036) (0.050) (0.032) (0.042) (0.037) (0.045) (0.063) (0.039)
WIC −0.020 −0.005 −0.034 0.015 −0.012 −0.014 −0.015 0.022 −0.009
(0.027) (0.038) (0.038) (0.037) (0.031) (0.061) (0.048) (0.049) (0.052)
School Meal 0.032 0.045 −0.011 0.069** 0.041 0.005 0.019 −0.063 0.150***
(0.026) (0.037) (0.044) (0.032) (0.031) (0.046) (0.045) (0.049) (0.045)
Alt.School Meal −0.020 −0.029 0.026 −0.034 −0.022 −0.009 0.019 −0.086**
(0.027) (0.038) (0.042) (0.034) (0.031) (0.047) (0.051) (0.044)
Charitable Food 0.087*** 0.069*** 0.113*** 0.076*** 0.076*** 0.081*** 0.072** 0.109*** 0.083***
(0.017) (0.024) (0.033) (0.020) (0.029) (0.022) (0.031) (0.038) (0.026)
Lost Job 0.102*** 0.128*** 0.118*** 0.096*** 0.117*** 0.162*** 0.007
(0.020) (0.028) (0.027) (0.029) (0.035) (0.038) (0.033)
Income Stable −0.096*** −0.059** −0.069** −0.103*** −0.118*** −0.081*** −0.086*** −0.122*** −0.084***
(0.018) (0.025) (0.033) (0.021) (0.028) (0.023) (0.031) (0.034) (0.030)
Rec. UI −0.055** −0.026 −0.090** −0.049 −0.123*** −0.006 −0.124*** −0.009 −0.019
(0.028) (0.041) (0.038) (0.043) (0.040) (0.039) (0.046) (0.051) (0.050)
Rec. Stimulus −0.013 −0.050 −0.034 −0.004 0.036 −0.052 −0.009 0.020 −0.072
(0.027) (0.038) (0.048) (0.033) (0.042) (0.040) (0.050) (0.053) (0.046)
Online Shopping −0.015 −0.035 −0.013 −0.008 −0.011 −0.023 −0.052* 0.0003 0.003
(0.015) (0.024) (0.031) (0.018) (0.027) (0.019) (0.029) (0.034) (0.022)
Price Increase 0.074*** 0.053** 0.067* 0.073*** 0.076** 0.079*** 0.060* 0.083** 0.081***
(0.018) (0.027) (0.038) (0.020) (0.032) (0.022) (0.034) (0.039) (0.026)
Food Dessert −0.005 0.0001 0.090* −0.031 −0.002 −0.007 −0.049 0.027 0.027
(0.021) (0.031) (0.047) (0.023) (0.034) (0.027) (0.034) (0.056) (0.032)
Min. to Store 0.001 −0.001 0.001 0.0004 0.0004 0.001 0.002* −0.00001 −0.002
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.002) (0.001)
Female −0.032** −0.031 −0.022 −0.037** −0.054* −0.028 −0.061** −0.020 −0.006
(0.016) (0.025) (0.037) (0.018) (0.032) (0.019) (0.030) (0.037) (0.023)
Age −0.004*** −0.002** −0.005*** −0.003*** −0.003** −0.004*** −0.005*** −0.003** −0.003***
(0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001) (0.001)
Married −0.015 −0.009 −0.078* 0.005 −0.014 −0.014 −0.017 −0.023 −0.008
(0.019) (0.032) (0.042) (0.021) (0.031) (0.024) (0.039) (0.041) (0.026)
Black 0.030 0.015 0.023 0.035 0.008 0.055
(0.042) (0.061) (0.084) (0.049) (0.070) (0.054)
Hispanic 0.015 −0.021 0.047 −0.004 −0.012 0.038
(0.043) (0.062) (0.083) (0.050) (0.068) (0.056)
White 0.065 0.033 0.004 0.074 0.029 0.094*
(0.041) (0.061) (0.084) (0.048) (0.068) (0.053)
Child3 More 0.117** 0.048 0.017 0.158** 0.090* 0.123 0.115 0.035
(0.054) (0.069) (0.091) (0.070) (0.054) (0.106) (0.075) (0.117)
High School −0.001 0.019 −0.008 0.011 0.058 −0.030 0.013 0.035 −0.059
(0.034) (0.040) (0.066) (0.039) (0.054) (0.043) (0.062) (0.067) (0.051)
Some College 0.041 0.059 −0.004 0.057 0.109** 0.009 0.079 0.045 −0.023
(0.034) (0.041) (0.068) (0.039) (0.056) (0.043) (0.063) (0.066) (0.051)
Food insecurity during the COVID-19 pandemic: evidence from a survey of low-income Americans 177
Table 4 (continued)
Full SNAP Lost Job Employed Children No Children Black Hispanic White
College Higher 0.039 0.052 0.013 0.061 0.064 0.030 0.107 0.029 −0.034
(0.037) (0.047) (0.074) (0.042) (0.062) (0.046) (0.069) (0.075) (0.055)
15,000-29,999 −0.021 0.031 −0.022 −0.021 0.015 −0.033* −0.050* 0.025 −0.013
(0.017) (0.035) (0.036) (0.019) (0.032) (0.020) (0.030) (0.039) (0.025)
30,000-44,999 −0.046* 0.115* −0.014 −0.046* −0.011 −0.066** −0.100** 0.012 −0.049
(0.024) (0.066) (0.045) (0.028) (0.038) (0.030) (0.045) (0.044) (0.038)
>45,000 −0.048 −0.137 −0.031 −0.055 −0.002 −0.139 −0.029 −0.007 −0.062
(0.041) (0.083) (0.084) (0.047) (0.049) (0.097) (0.080) (0.073) (0.064)
Intercept 0.404*** 0.386*** 0.610*** 0.366*** 0.340*** 0.425*** 0.563*** 0.250** 0.506***
(0.066) (0.095) (0.137) (0.075) (0.112) (0.083) (0.098) (0.112) (0.083)
N 2706 1170 534 2172 883 1823 856 648 1128
Notes: coefficients are marginal effects. Standard errors are in the parentheses. *,**, and *** indicate significance levels at 0.5, 0.1 and 0.01, respectively
Table 5 Logit Model Results
Full SNAP Lost Job Job Children No Children Black Hispanic White
SNAP 0.419** 0.348 −0.462 0.635*** 0.881*** 0.382 0.339 0.866** 0.364
(0.182) (0.263) (0.484) (0.203) (0.329) (0.245) (0.309) (0.393) (0.338)
Add. SNAP 0.032 0.148 −0.608 0.017 −0.117 −0.029 −0.231 0.421 −0.199
(0.203) (0.285) (0.568) (0.223) (0.362) (0.261) (0.351) (0.427) (0.363)
WIC 0.043 0.213 −0.279 0.161 0.067 0.150 0.231 0.384 −0.078
(0.228) (0.340) (0.523) (0.260) (0.268) (0.526) (0.389) (0.423) (0.478)
School Meal 0.376** 0.641** −0.171 0.536** 0.453* 0.095 0.247 −0.257 1.558***
(0.190) (0.299) (0.469) (0.215) (0.231) (0.355) (0.325) (0.328) (0.419)
Alt.School Meal −0.103 −0.173 0.456 −0.213 −0.173 0.047 0.034 −0.692
(0.195) (0.289) (0.487) (0.221) (0.235) (0.332) (0.327) (0.457)
Charitable Food 0.620*** 0.553*** 1.210*** 0.528*** 0.673*** 0.577*** 0.529** 0.800*** 0.733***
(0.126) (0.189) (0.343) (0.140) (0.230) (0.155) (0.214) (0.247) (0.234)
Lost Job 0.883*** 1.228*** 1.244*** 0.686*** 1.111*** 1.176*** 0.084
(0.163) (0.257) (0.277) (0.210) (0.296) (0.285) (0.313)
Income Stable −0.698*** −0.525*** −0.722** −0.724*** −0.885*** −0.563*** −0.650*** −0.812*** −0.745***
(0.121) (0.190) (0.296) (0.138) (0.205) (0.155) (0.215) (0.221) (0.230)
Rec. UI −0.447** −0.199 −0.888*** −0.205 −1.163*** 0.009 −1.011*** −0.054 −0.016
(0.200) (0.311) (0.326) (0.285) (0.328) (0.255) (0.346) (0.343) (0.407)
Rec. Stimulus −0.025 −0.280 0.079 −0.051 0.298 −0.339 0.081 0.273 −0.687*
(0.211) (0.321) (0.539) (0.238) (0.346) (0.307) (0.408) (0.382) (0.393)
Stimulus Amount −0.088 −0.153 0.079 −0.107 −0.267* 0.175 −0.050 −0.116 −0.003
(0.122) (0.192) (0.316) (0.138) (0.162) (0.211) (0.272) (0.215) (0.216)
Online Shopping −0.082 −0.200 −0.338 −0.055 −0.050 −0.144 −0.391* 0.025 0.033
(0.117) (0.188) (0.327) (0.128) (0.208) (0.145) (0.218) (0.229) (0.199)
Price Increase 0.511*** 0.369* 0.476 0.510*** 0.502** 0.550*** 0.485** 0.461* 0.735***
(0.132) (0.196) (0.353) (0.147) (0.232) (0.165) (0.244) (0.253) (0.237)
Food Dessert −0.104 −0.122 0.838* −0.227 −0.131 −0.118 −0.452* 0.139 0.112
(0.158) (0.243) (0.442) (0.176) (0.284) (0.196) (0.253) (0.376) (0.265)
Min. to Store 0.004 −0.016 0.023 0.003 0.004 0.006 0.021** −0.0002 −0.019*
(0.006) (0.010) (0.017) (0.007) (0.012) (0.008) (0.010) (0.014) (0.011)
Female −0.246** −0.256 −0.100 −0.279** −0.442* −0.208 −0.487** −0.058 −0.081
(0.121) (0.198) (0.328) (0.134) (0.260) (0.140) (0.232) (0.239) (0.200)
Age −0.024*** −0.011* −0.039*** −0.023*** −0.019** −0.029*** −0.034*** −0.017** −0.024***
(0.004) (0.006) (0.011) (0.004) (0.008) (0.005) (0.007) (0.008) (0.007)
Married −0.109 −0.112 −0.836** 0.034 −0.135 −0.106 −0.150 −0.088 −0.047
(0.141) (0.257) (0.376) (0.156) (0.236) (0.185) (0.272) (0.277) (0.236)
Black 0.315 0.245 0.285 0.347 0.169 0.475
(0.344) (0.502) (0.879) (0.386) (0.526) (0.479)
Hispanic 0.182 −0.081 0.571 0.089 −0.014 0.337
178 Fang D. et al.
Table 5 (continued)
Full SNAP Lost Job Job Children No Children Black Hispanic White
(0.347) (0.506) (0.854) (0.391) (0.521) (0.485)
White 0.580* 0.396 0.303 0.645* 0.349 0.757
(0.345) (0.506) (0.901) (0.386) (0.536) (0.477)
Child 3 More 0.910** 0.442 −0.085 1.175** 0.712 0.638 0.844 0.372
(0.424) (0.589) (0.828) (0.500) (0.446) (0.716) (0.841) (0.848)
High School 0.115 0.291 0.361 0.147 0.592 −0.100 0.127 0.311 −0.386
(0.261) (0.321) (0.793) (0.284) (0.444) (0.334) (0.433) (0.504) (0.519)
Some College 0.410 0.595* 0.595 0.493* 1.015** 0.149 0.666 0.366 −0.151
(0.260) (0.323) (0.801) (0.282) (0.455) (0.330) (0.428) (0.493) (0.523)
College Higher 0.386 0.508 0.725 0.505* 0.669 0.302 0.882* 0.273 −0.241
(0.279) (0.371) (0.848) (0.304) (0.494) (0.351) (0.477) (0.536) (0.546)
$15,000–$29,999 −0.155 0.151 −0.253 −0.128 0.025 −0.243 −0.362* 0.207 −0.084
(0.127) (0.266) (0.359) (0.140) (0.271) (0.148) (0.215) (0.265) (0.222)
$30,000–$44,999 −0.310* 0.782 −0.248 −0.307 −0.097 −0.427* −0.823** 0.196 −0.402
(0.170) (0.550) (0.439) (0.189) (0.299) (0.220) (0.333) (0.303) (0.313)
>$45,000 −0.437 −11.724 −0.548 −0.470 −0.070 −0.943 −0.402 0.040 −0.756
(0.297) (535.412) (0.808) (0.332) (0.401) (0.588) (0.495) (0.558) (0.606)
Intercept −0.462 −0.505 −0.117 −0.574 −1.009 −0.212 0.810 −1.498* 0.875
(0.488) (0.692) (1.379) (0.538) (0.843) (0.638) (0.647) (0.822) (0.684)
N 2706 1170 534 2172 883 1823 856 648 1128
Log Likelihood −1119.032 −458.025 −161.834 −930.779 −338.872 −761.031 −353.095 −287.649 −405.450
Akaike Inf. Crit. 2316.065 994.051 399.668 1937.557 753.743 1590.062 778.190 647.297 882.901
Notes: coefficients are marginal effects. Standard errors are in the parentheses. *,**, and *** indicate significance levels at 0.5, 0.1 and 0.01, respectively
Fig. 1 Marginal Effects Across
Different Models: Full Sample
Fig. 2 Marginal Effects Across
Different Models: SNAP Eligible
Food insecurity during the COVID-19 pandemic: evidence from a survey of low-income Americans 179
Fig. 3 Marginal Effects Across
Different Models: Lost Job
Fig. 4 Marginal Effects Across
Different Models: Has Job
Fig. 5 Marginal Effects Across
Different Models: Household
with Children
180 Fang D. et al.
Fig. 6 Marginal Effects Across
Different Models: Household
without Children
Fig. 7 Marginal Effects Across
Different Models: Black Sample
Fig. 8 Marginal Effects Across
Different Models: Hispanic
Sample
Food insecurity during the COVID-19 pandemic: evidence from a survey of low-income Americans 181
Fig. 9 Marginal Effects Across
Different Models: White Sample
Author contributions Fang: Conceptualization; Formal analysis; Writing Coleman-Jensen, A., Rabbitt, M., Gregory, C., & Singh, A. (2019).
– original draft; Writing – review & editing. Household food security in the United States in 2018. USDA-ERS
Thomsen: Conceptualization; Writing – original draft; Writing – re- Economic Research Report, 270, 1–35.
view & editing. Cotti, C. D., Gordanier, J. M., & Ozturk, O. D. (2020). Hunger pains?
Nayga Jr.: Conceptualization; Writing – review & editing. SNAP timing and emergency room visits. Journal of Health
Yang: Data curation. Economics, 71, 102313.
Dunn, C. G., Kenney, E., Fleischhacker, S. E., & Bleich, S. N. (2020).
Data availability available. Feeding low-income children during the covid-19 pandemic. New
England journal of medicine, 382 (18), https://doi.org/10.1056/
NEJMp2005638, e40.
Declarations Fitzpatrick, K., Greenhalgh-Stanley, N., & Ver Ploeg, M. (2016). The
impact of food deserts on food insufficiency and SNAP participation
Conflict of interest The authors declared that they have no conflict of among the elderly. American Journal of Agricultural Economics,
interest. 98(1), 19–40.
Food Environment Atlas (2020). USDA-ERS. Retrieved Aug 20, 2020
from https://www.ers.usda.gov/data-products/food-environment-
atlas/
Gallagher, D., & Kirkland, P. (April 27, 2020). Meat processing plants
References across the US are closing due to the pandemic. Will consumers feel
the impact. CNN business. Retrieved Aug 20, 2020, from https://
Bartfeld, J., & Dunifon, R. (2006). State-level predictors of food insecu- www.cnn.com/2020/04/26/business/meat-processing-plants-
rity among households with children. Journal of Policy Analysis and coronavirus/index.html
Management, 25(4), 921–942. Ganong, P., Noel, P., & Vavra, J. (2020). US unemployment insurance
Bitler, M., & Hoynes, H. (2016). The more things change, the more they replacement rates during the pandemic. Journal of Public
stay the same? The safety net and poverty in the great recession. Economics, 191, 104273.
Journal of Labor Economics, 34(S1), S403–S444. Greene, W. H. (2008). The econometric approach to efficiency analysis.
Bitler, M., Hoynes, H. W., & Schanzenbach, D. W. (2020). The social The measurement of productive efficiency and productivity growth,
safety net in the wake of covid-19. BPEA conference drafts. 1(1), 92–250.
Brookings papers on economic activity. Retrieved Aug 20, 2020, Gregory, C. A., & Todd, J. E. (2021). SNAP timing and food insecurity.
from https://www.brookings.edu/wp-content/uploads/2020/06/ PLoS One, 16(2), e0246946.
Bitler-et- al-conference-draft.pdf. Grose, J. (2020). Families scramble to find baby formula, diapers and
Cannuscio, C. C., Tappe, K., Hillier, A., Buttenheim, A., Karpyn, A., & wipes. New York Times. Retrieved Aug 20, 2020, from https://
Glanz, K. (2013). Urban food environments and residents’ shopping www.nytimes.com/2020/03/30/parenting/coronavirus-baby-
behaviors. American Journal of Preventive Medicine, 45(5), 606– formula-shortages-wipes-diapers.html
614. Gundersen, C., & Ziliak, J. P. (2015). Food insecurity and health out-
Chang, H. H., & Meyerhoefer, C. D. (2020). COVID-19 and the demand comes. Health Affairs, 34(11), 1830–1839.
for online food shopping services: Empirical evidence from Taiwan. Henningsen, A., Hamann, J. D., et al. (2007). Systemfit: A package for
American Journal of Agricultural Economics., 103, 448–465. estimating systems of simultaneous equations in R. Journal of
https://doi.org/10.1111/ajae.12170. Statistical Software, 23(4), 1–40.
182 Fang D. et al.
Information Resources, Inc. (March 25, 2020). Then and now: Consumer USDA-FNS. (2020a). Supplemental nutrition assistance program
CPG behavior during economic downturns. Retrieved Aug 20, (SNAP). Retrieved Aug 20, 2020 from https://www.fns.usda.gov/
2020 from: https://www.iriworldwide.com/IRI/media/Library/ snap/supplemental-nutrition-assistance-program
COVID-19-Thought-Leadership-3-25-2020.pdf USDA-FNS. (2020b). Child and adult care food program. Retrieved
Johansson, R. (May 28, 2020). Another look at availability and prices of Aug 20, 2020 from https://www.fns.usda.gov/cacfp.
food amid the covid- 19 pandemic. USDA. Retrieved Aug 20, 2020 USDA-FNS. (2020c). Special supplemental nutrition program for wom-
from https://www.usda.gov/media/blog/2020/05/28/another-look- en, infants, and children (WIC). Retrieved Aug 20, 2020 from
availability-and-prices-food-amid-covid-19-pandemic https://www.fns.usda.gov/wic
Nichols, A., & Simms, M. (2012). racial and ethnic Differences in receipt USDA-FNS. (2020d). FNS launches the online purchasing pilot.
of unemployment insurance benefits during the Great recession. Retrieved Aug 20, 2020 from https://www.fns.usda.gov/snap/
Unemployment and Recovery Project Brief, 4. online-purchasing-pilot
Raifman, J., Bor, J., & Venkataramani, A. (2021). Association between USDA-FNS. (2020e). U.S. Adult Food Security Survey Module.
receipt of unemployment insurance and food insecurity among peo- Retrieved Aug 20, 2020 from https://www.ers.usda.gov/topics/
ple who lost employment during the COVID-19 pandemic in the food-nutrition-assistance/food-security-in-the-us/survey-tools/ f.
United States. JAMA Network Open, 4(1), e2035884–e2035884. Wilde, P. E., & Ranney, C. K. (2000). The monthly food stamp cycle:
Ratcliffe, C., & McKernan, S. M. (2010). How much does SNAP reduce Shopping frequency and food intake decisions in an endogenous
food insecurity? (no. 2239-2019-2874). Retreived march 17, 2021 switching regression framework. American Journal of Agricultural
from https://ageconsearch.umn.edu/record/292080/. Economics, 82(1), 200–213.
Reardon, T., Belton, B., Liverpool-Tasie, L. S. O., Lu, L., Nuthalapati, C.
S., Tasie, O., & Zilberman, D. (2021). E-commerce’s fast-tracking
diffusion and adaptation in developing countries. Applied Economic
Perspectives and Policy. https://doi.org/10.1002/aepp.13160. Di Fang is Assistant Professor in
Redman, R. (2020). Online grocery sales to grow 40% in 2020. super- the Department of Agricultural
market news. Retrieved Aug 20, 2020 from https://www. Economics and Agribusiness at
supermarketnews.com/online-retail/online-grocery-sales-grow-40- the University of Arkansas. She
2020 received a Ph.D. in Business
Rosenbaum, D. (2021). Food assistance in president Biden’s COVID Administration (Agribusiness)
relief plan would reduce hardship, Provide Economic Stimulus. from W.P. Carey School of
Center on budget and policy priorities. Retrieved March 17, 2021, Business at Arizona State
from https://www.cbpp.org/research/food-assistance/food- University. Dr. Fang’s research
assistance-in-president-bidens-covid-relief-plan-would-reduce focus on food and health econom-
Rosenbaum, D., Bolen, E., Neuberger, Z., & Dean, S. (2020). States must ics, consumer economics and
act swiftly to deliver food assistance allowed by families first act. marketing, and issues related to
Center on budget and policy priorities. Retrieved Aug 20, 2020 from childhood obesity.
https://www.cbpp.org/research/food-assistance/states-are-using-
much-needed-temporary-flexibility-in-snap-to-respond-to
Siddiqi, S. M., Cantor, J., Dubowitz, T., Richardson, A., Stapleton, P. A.,
& Katz, Y. (March 31, 2020). Food access: Challenges and solutions
brought on by covid-19. Rand Corporation. Retrieved Aug 20, 2020
from https://www.rand.org/blog/2020/03/food-access-challenges-
and-solutions-brought-on-by.html Michael Thomsen is Professor
Smith, T. A. (2017). Do school food programs improve child dietary and Governor Sidney S. McMath
quality? American Journal of Agricultural Economics, 99(2), 339– Endowed Chair in Obesity
356. Prevention. He is in the
Stone, C. (2020). Cares act measures strengthening unemployment in- Department of Health Policy and
surance should continue while need remains. CBPP. Retrieved Management, Fay W. Boozman
Aug 20, 2020 from https :// www. cbpp. org/ research / federal- College of Public Health,
budget/cares-act-measures-strengthening-unemployment- University of Arkansas for
insurance- should-continue Medical Sciences. He completed
US Department of Labor. (2020). Families first coronavirus response act. his PhD at the University of
Retrieved Aug 20, 2020 from https://www.dol.gov/agencies/whd/ Minnesota in agricultural and ap-
pandemic/ffcra-employee-paid-leave plied economics with a minor in
US Department of the Treasury. (2020). The cares act works for all business administration. Dr.
Americans. Retrieved Aug 20, 2020 from https://home.treasury. Thomsen’s research program em-
gov/policy-issues/cares. phasizes linkages between the
food marketing system and human health. Topic areas include the effect
USDA-ERS (2020) Food Security in the US: Key Statistics & Graphics.
of the commercial food environment on childhood obesity and the eco-
Retrieved Aug 20, 2020 from https://www.ers.usda.gov/topics/
nomics of nutrition programs.
food-nutrition-assistance/food-security-in-the-us/key-statistics-
graphics.aspx
Food insecurity during the COVID-19 pandemic: evidence from a survey of low-income Americans 183
Rodolfo M. Nayga , Jr. is Wei Yang is a program associate
Professor and Head, Department in the Department of Agricultural
of Agricultural Economics at Economics and Agribusiness at
Texas A&M University. He re- the University of Arkansas. He re-
ceived his Ph.D. from Texas ceived M.S. degrees in
A&M University. His research in- Agricultural Economics and
terests are focused on the eco- Statistics. His specialties are ex-
nomics of food policy and quality perimental design and data analy-
and on obtaining an understand- sis of choice experiment.
ing of how emerging consumer
issues affect food and nutrient
consumption/demand, obesity,
and public policies.
File and source
- File
- food-insecurity-during-the-covid-19-pandemic-evidence-from-a-survey-of-low-income-americans.pdf
- Size
- 1,193,827 bytes
- SHA-256
- 6d6b75b55154a5826def901c46db9a6905744d24b96bd50399c1c8b90e60f9d2
- Our copy
- food-insecurity-during-the-covid-19-pandemic-evidence-from-a-survey-of-low-income-americans.pdf
- Original
- doi.org