An, Gabriel, Tzur-Ilan, More Than Shelter: The Effects of Rental Eviction Moratoria on Household Well-Being (Philadelphia Fed WP 22-10; AEA Papers and Proceedings 112 (2022), 2022-03)
Summary
Philadelphia Fed working paper WP 22-10, More Than Shelter: The Effects of Rental Eviction Moratoria on Household Well-Being, by Xudong An, Stuart A. Gabriel and Nitzan Tzur-Ilan, dated March 2022. The abstract says new panel data on 2020 COVID-19 rental eviction moratoria show that the moratoria reduced eviction filings, redirected household resources to immediate consumption needs including food and groceries, and reduced household food insecurity and mental stress, with larger effects among African American households. The data sources include the Federal Reserve Y-14M regulatory report, Opportunity Insights spending data, the COVID-19 Household Pulse Survey and Google Trends. The study period is March-August 2020.
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
Cite this document
“An, Gabriel, Tzur-Ilan, More Than Shelter: The Effects of Rental Eviction Moratoria on Household Well-Being (Philadelphia Fed WP 22-10; AEA Papers and Proceedings 112 (2022), 2022-03).”
Original at www.philadelphiafed.org. Archived at Pandemic Darlings, https://pandemicdarlings.com/source-documents/an-gabriel-tzur-ilan-more-than-shelter-the-effects-of-rental-eviction-06565f614d/
, SHA-256 49be7d4149c48ed8c5355eedecbc6f938b1ac07d26ecbdc9949c3216b553c16c.Full text
Working Papers
RESEARCH DEPARTMENT
More Than Shelter:
The Effects of Rental Eviction
Moratoria on Household
Well-Being
WP 22-10
Xudong An
Federal Reserve Bank of Philadelphia Supervision, Regulation, and Credit
Department
PUBLISHED
Stuart A. Gabriel
UCLA Anderson School of Management
March 2022
Nitzan Tzur-Ilan
Federal Reserve Bank of Dallas
ISSN: 1962-5361
Disclaimer: This Philadelphia Fed working paper represents preliminary research that is being
circulated for discussion purposes. The views expressed in these papers are solely those of
the authors and do not necessarily reflect the views of the Federal Reserve Bank of Philadelphia
or the Federal Reserve System. Any errors or omissions are the responsibility of the authors.
Philadelphia Fed working papers are free to download at: https://philadelphiafed.org/research- DOI: https://doi.org/
and-data/publications/working-papers. 10.21799/frbp.wp.2022.10
More Than Shelter: The Effects of Rental Eviction Moratoria
on Household Well-Being
Xudong An, Stuart A. Gabriel, and Nitzan Tzur-Ilan∗
March 2022
ABSTRACT
We investigate the impact of 2020 COVID-19 rental eviction moratoria on household well-being.
Analysis of new panel data indicates that eviction moratoria reduced evictions filings and resulted
in redirection of scarce household financial resources to immediate consumption needs, notably
including food and grocery spending. We also find that eviction moratoria reduced household food
insecurity and mental stress, with larger effects evidenced among African American households.
Findings suggest broad salutary effects of eviction moratoria during a period of widespread virus
and economic distress.
JEL Classification: G28, R30, I38
Keywords: Eviction moratorium, consumption, food security, mental health, COVID-19
∗
An: Federal Reserve Bank of Philadelphia (email: xudong.an@phil.frb.org) Gabriel: UCLA Anderson
School of Management (email: sgabriel@anderson.ucla.edu). Tzur-Ilan: Federal Reserve Bank of Dallas (email:
Nitzan.TzurIlan@dal.frb.org). We thank Kartik Agarwal and Susu Zhu for excellent research assistance. The views
expressed in this paper are solely those of the authors and do not necessarily reflect the views of the Federal Reserve
Bank of Philadelphia or the Federal Reserve System.
I. Introduction
In the wake of the onset of the 2020 COVID-19 pandemic, many state and local governments in the
U.S. enacted rental eviction moratoria. Their objective was to assure shelter of households idled by
the pandemic and to damp virus spread. In this paper, we provide new panel information on the
enactment of eviction moratoria and evaluate the effects of those measures on household well-being,
including consumer spending, food insecurity, and mental health.
Upon the pandemic onset, the share of affordability-constrained renters, defined as households
payng more than one-half of their income for rent, jumped to one-half of all renter households.1
Moratoria on eviction and related deferral of rent may have provided treated households with fi-
nancial relief in the form of positive shocks to household liquidity. Renters benefiting from such
interventions could have redirected scarce resources to other immediate consumption needs such as
food purchases. Eviction moratoria similarly assured renters of continued shelter during a period of
elevated COVID-19 virus transmission, likely easing their mental stress and anxiety.
The staggered implementation of state and county rental eviction moratoria enabled construc-
tion of data panels to identify policy effects. Our study focuses on the March-August 2020 period
prior to the enactment of a nationwide federal eviction moratorium.2
We used web scraping and text parsing protocols to conduct an automated search of COVID-19
rental policy interventions at state-level governor, court, and legislation websites over the period
of analysis. Ultimately, 43 states enacted eviction moratoria. For county-level information, we
used data from the Eviction Lab at Princeton University. Dynamic maps of our newly constructed
state and county eviction moratoria panels are available at https://covid19evictionmoratoria
.anderson.ucla.edu/map/.3
1
See the Census COVID-19 Household Pulse Survey.
2
In September 2020, the Centers for Disease Control and Prevention (CDC) broadened the federal eviction mora-
torium to effectively protect all of the nation’s 43 million rental households.
3
During our period of analysis, the federal eviction moratorium as specified by the CARES Act was limited only
to renters who received federal housing assistance or lived in a property with a federally backed mortgage. The
Federal Reserve Bank of Atlanta estimated that the CARES Act moratorium covered between 28 and 46 percent of
occupied rental units nationally, leaving as many as 31 million renter households without federal eviction protection.
1
II. Data and Research Design
A. Data Sources
A primary source of data for this study is the Federal Reserve Y-14M regulatory report. The report
contains detailed information on the asset portfolios of bank holding companies required to
participate in Federal Reserve stress testing. The monthly report at the account level contains
detailed information on borrower characteristics, credit card purchases, and payments.4 For the
purposes of our study, we aggregate the account-level credit card data to the zip code-level and form
a zip code by month panel. We focus on two outcomes, including credit card spending and payment.
To compute credit card spending, we include purchases using credit cards, cash withdrawals, and
convenience checks but exclude balance transfers to avoid double counting. To account for
seasonality, we calculate year-over-year changes in the outcome terms.5 Eviction moratoria and other
macroeconomic controls such as unemployment rate and house price index (HPI) are merged to
credit card data using geographic identifiers such as county Federal Information Processing
Standards (FIPS) code and state name.
We also use data from the real-time Opportunity Insights Economic Tracker (hereinafter Op-
portunity Insights) by Chetty et al. (2020) to assess consumer spending. The Opportunity Insights
data are available only at the state or county level. However, a distinct advantage is that
they contain measures of consumption by category of spending, including non-durable spending,
spending on grocery and food stores, and the like. The Opportunity Insights data also include
debit card purchases.
To assess the effect of eviction moratoria on food insecurity and mental health disorders, we
compiled information from the Census COVID-19 Household Pulse Survey (hereinafter Census
See “Housing Policy Impact: Federal Eviction Protection Coverage and the Need for Better Data,” by Sarah Stein
and Nisha Sutaria.
4
The data set contains about 500 million anonymized credit card accounts in the U.S. We work with a 1 percent
random sample of the data.
5
Zip codes with fewer than 100 accounts in 2020 are excluded to ensure that change statistics are not affected by
outliers.
2
Pulse Survey). That survey commenced on April 23, 2020, and included information on food
sufficiency and insecurity. We also use search query data from Google Trends to develop broad-
based real-time search indicators of food insecurity. The Census Pulse Survey partnered with
the National Center for Health Statistics (NCHS) to include three questions about anxiety or
depression. The mental health outcome terms include “feeling anxious,” “can’t stop worrying,”
and “feeling down.”
B. Empirical Strategy
We employ a panel data model with fixed effects to identify the relation between eviction morato-
rium and household well-being. Our observations are at zip code, county, or state level and our
outcome terms vary by month or week. Given sample structure, we estimate the following model:
Yit = α + βVit + X 0 it γ + τt + ζi + εit , (1)
where Yit represents the outcome in zip code/county/state i at time t, Vit is an indicator of the
treatment, eviction moratorium, in place i and period t; and Xit is a matrix of time- and place-
varying control terms such as unemployment rate and house price appreciation. τt and ζi are
time- and geographic-fixed effects. Finally, εit represents the error terms, which are assumed to be
clustered at the state or county level. The coefficient β is the treatment effect of eviction moratoria.
Eviction moratoria target renter populations, especially those having rental payment difficulties.
Hence, we use the contrast between renters and homeowners to aid in identification. In that regard,
in addition to baseline models, we estimate the following treatment intensity difference-in-differences
(DID) regression:
Yit = α + β1 Vit + β2 Vit Ri + β3 Ri + X 0 it γ + τt + ζi + εit , (2)
3
where Ri is a treatment intensity indicator based on the local renter share and unemployment rate as
proxies for the share of local population in financial distress. More specifically, Ri is a dummy variable
for zip codes in the top two quartiles of renter share and unemployment rate in April 2020, the first
peak of the COVID-19 pandemic. Note that the impact of Ri is absorbed by the fixed effects in the
regression. In this DID setting, eviction moratorium is the treatment, and areas with high renter
share and high financial distress are more intensively “treated.” β2 is the lower-bound estimate of the
treatment effect. This augmented specification aids our inference of causality as renters (especially
those in financial distress) were the target beneficiaries of eviction moratoria.
III. Results
A. Credit Card Spending and Payment
We first present results based on the Federal Reserve Y-14M credit card data. Our focus variable in
Table I is an indicator of the presence of a state-level eviction moratoria in the zip code during a
particular month. We lag the focus variable by two weeks.6 Given the above model specification, we
show in column 1 the baseline model results for credit card spending. The positive coefficient of the
State Evic. Mor. term indicates that enactment of a state eviction moratorium is associated with
elevated zip code credit card spending.
Column 2 estimates treatment effects among targeted renter zip codes. The focus term is the
interaction of State Evic. Mor. with the above defined Target indicator. Results of this treatment
intensity difference-in-differences (DID) analysis indicate a positive and significant effect of state
eviction moratoria on credit card spending among targeted zip codes. The one-month target zip
code treatment effect is 1.356 percent, meaning that a 12-month treatment effect amounts to a 16
percent (1.356×12=16) increment in spending. To put this into perspective, the average year-over-
year decline in credit card spending in April 2020 was 25 percent.
6
To identify the effects of state eviction moratorium, we comprise a sample of only those states where no county-
level eviction moratorium was in place.
4
Table I: Effects of State-Level Eviction Moratoria on Credit Card Utilization and Consumption by
Category
Fed Y-14M zip code×month panel Opportunity Insights state×week panel
Spending Change Payment Change Food Grocery Retail Non
Service Spending with Durable
Spending Grocery Spending
(1) (2) (3) (4) (5) (6) (7) (8)
State Evic. Mor. 1.867∗ 1.458 0.249 -0.610 0.991* 0.917* 0.992* 1.373*
(0.988) (1.036) (1.353) (1.440) (0.530) (0.526) (0.590) (0.712)
State Evic. Mor.×Target 1.356∗∗ 1.199∗
(0.583) (0.707)
Constant -10.259∗∗∗ -9.527∗∗∗ 3.183 3.828 -9.570*** 11.887*** -7.640*** 3.235
(1.523) (1.482) (2.351) (2.298) (0.678) (0.722) (0.508) (2.525)
Zip Code/County FE X X X X X X X X
Month/Week FE X X X X X X X X
N 20,996 20,996 20,650 20,650 1,122 1,122 1,122 1,122
R2 0.5788 0.5790 0.4887 0.4888 0.849 0.873 0.966 0.606
Notes: Columns 1-4 present our estimates of the impact of state-level eviction moratorium on consumer credit card spending and
payment based on zip code by month panel data of YoY changes of the outcome variables. Columns 5-8 present our estimates
of the impact of a one-week state-level eviction moratorium on state spending seasonally adjusted by category, relative to its
level in January 2020. We follow Chetty et al. (2020) definition of durable and non-durable goods. Data sources include the
Federal Reserve Y-14M, the Bureau of Labor Statistics, the Bureau of Economic Analysis, the Census, the Eviction Lab, and
the Opportunity Insights. Robust standard errors in parentheses with error terms clustered at the state-level; ∗ p < 0.1, ∗∗
p < 0.05, ∗∗∗ p < 0.01.
Turning to credit card payments, our specification also includes lagged spending as households
typically increase debt paydown in the wake of an increase in the prior month’s spending. Column
3 shows results of the baseline model, while column 4 shows results of the difference-in-differences
(DID) analysis. The estimated coefficient for eviction moratoria in targeted areas is both statisti-
cally and economically significant. In that regard, a 12-month eviction moratorium is associated
with a 14 percent increase in credit card debt paydown in targeted zones.7
Certain counties implemented eviction moratoria even in the presence of similar statewide pol-
icy treatment. We thus seek to ascertain whether those county-level moratoria had an incremental
benefit. We select places that enacted state-level eviction moratoria and then reestimate our models
based on cross-county variations.8 The estimated treatment terms (not reported) are not statis-
7
During the pandemic study period, governments provided emergency income support including stimulus checks
and added unemployment benefits to households, many of whom are the credit card borrowers who we study in this
paper. To account for that, we included real disposable income as an additional control and reestimated all models.
Results (not shown) are robust are highly consistent with those in Table I.
8
Given that the treatment effect of interest is now at the county level, we use MSA by month-fixed effects to
account for variations in economic and other factors across MSAs and over time.
5
tically significant, suggesting little incremental advantage associated with additional county-level
treatment.
B. Consumer Spending by Category
The Opportunity Insights data allow us to test for the effects of eviction moratoria on specific
categories of household consumption, notably including basic food and nondurable retail consump-
tion. Table I shows that a one-week state eviction moratorium is associated with an annual increase
in food service spending (column 5) of 1 percent, an annual increase in grocery spending of 0.9
percent (column 6), an annual increase in retail with grocery of 1 percent (column 7), and an annual
increase in non-durable spending (column 8) of 1.4 percent.
C. Food Insecurity
In Table II, we use information from the Census Pulse Survey and report the results of regressions of
state eviction moratoria on food insecurity. We follow the same specification as in the previous tables.
We define “food insecurity” as the share of respondees who over the past 7 days declared that they
sometimes or often didn’t have enough food to eat. Column 1 of Table II reports the average
treatment effect (lagged in two weeks) on food insecurity. The estimated coefficient is negative but
not statistically significant. However, for African Americans (column 2), the estimated effect is
negative and statistically significant; an additional week of rental eviction moratorium treatment is
associated with a decline of 2 percent in the number of African American households that declared as
food insecured. On average, over the 10 weeks of the Census Pulse Survey, 21 percent of African
American households declared that in the prior seven days they “Sometimes do not have enough food
to eat” or “Often do not have enough food to eat.”
Using Google Trends data, we consider search keywords for food insecurity, such as “food” in
combination with the word “help.” This process leads to two key search terms, including “Food
Stamps” and “Food Banks Near Me.”9
9
See Chauvet et al. (2016) for a similar search algorithm.
6
In Table II, we report findings of the Google Trends analysis. Columns 3 and 4 of Table II show
that state eviction moratoria significantly reduce Google search for “Food Stamps” and “Food Banks
Near Me.” An additional week of a state eviction moratoria reduces the Google search query for
“Food Stamps” by 3.4, relative to an average SVI for that term of 40.5 between March and August
2020. Similarly, an additional week of state eviction moratoria reduce the Google search query for
“Food Banks Near Me” by 5.1, relative to an average Google search for that term of 20.5 between
March and August 2020.
Table II: Effects of State-level Rental Eviction Moratoria on Food Insecurity and Mental Health
Insecurity Insecurity Food Food Banks Can’t Can’t Stop Feeling Feeling
All Black Stamps Near Me Stop Worrying Anxious Down
Worrying Black Black Black
(1) (2) (3) (4) (5) (6) (7) (8)
State Evic. Mor. -1.906 -1.958* -3.401* -5.124* -0.983* -0.985 -1.866*** -1.627***
(2.712) (1.053) (1.789) (2.697) (0.578) (2.451) (0.577) (0.576)
Constant 1.160*** 1.277*** 68.904*** 36.413*** 23.899*** 26.197*** 34.329*** 23.865***
(0.245) (0.233) (4.256) (3.724) (1.167) (5.183) (5.508) (5.119)
County FE X X X X X X X X
Week FE X X X X X X X X
N 612 612 1,122 1,122 612 612 612 612
R2 0.112 0.109 0.301 0.090 0.338 0.034 0.165 0.098
Notes: This table reports the results from regressions of state eviction moratoria on food insecurity and mental health. We
define “food insecurity” as the percentage of people who declared that they sometimes or often don’t have enough food to eat. In
columns 3 and 4, we use Google data to collect information from individuals seeking assistance via an Internet search on food
insecurity. In columns 5-8, we use three different mental health disorders, taken from the Census Pulse Survey, and include
feeling anxious, can’t stop worrying, and feeling down. For each of the three indicators, we define the percentage of people who
replied that they experienced this feeling more than half the days or nearly every day over the last seven days. Data sources
include the Eviction Lab, Google Trends, and the Census Pulse Survey. Robust standard errors in parentheses with error terms
clustered at the state-level; ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01.
D. Mental Health
The Census Pulse Survey indicated that some 30 percent of households felt depressed or down during
the pandemic survey period. Brodeur et al. (2021) use Google Trends data to show adverse effects of
the pandemic on measures of boredom, loneliness, worry, and sadness. We test if some of those
symptoms have been relieved by a temporary stay in eviction.
Columns 5-6 of Table II estimate the average rental eviction moratorium treatment effect (lagged
two weeks) on the share of households reported that they “can’t stop worrying” over the prior seven
7
days. Column 5 of Table II shows that eviction moratoria significantly reduced the number of house-
holds that reported they “can’t stop worrying.” As indicated in column 7 of Table II, an additional
week of rental eviction moratoria is associated with a significant decline of 1.9 percent in the share of
African American households that reported “feeling anxious.” The Census Pulse Survey indicated an
increase by roughly one-third in the share of African American households that reported “feeling
anxious” during the April to August 2020 pandemic period. Also, as shown in column 8, an ad-
ditional week of eviction moratoria policy treatment is associated with a reduction by 1.6 percent in
the share of African American households that reported “feeling down.” As suggested by the survey,
the pandemic study period witnessed a roughly one-quarter increase in share of African
American households that reported “feeling down.”
IV. Conclusions and Discussion
This paper provides evidence of broad salutary impacts of COVID-19 rental eviction moratoria
during a period of widespread virus and economic distress. Analysis of both Federal Reserve and
Opportunity Insights data indicates that the imposition of rental eviction moratoria served to boost
food and grocery spending, especially among policy-targeted neighborhoods. Eviction moratoria
also reduced Census Pulse Survey-measured food insecurity and mental stress, especially among
African American households. Results are corroborated in analysis of search query data from
Google.
Federal eviction moratoria expired on October 3, 2021, leaving upward of 2 million U.S. house-
holds at risk of eviction. Yet data from the Eviction Lab show that evictions were lower than
anticipated through year-end 2021, likely owing to substantial distribution of $47 billion in federal
stimulus emergency rental assistance funds. While early outcomes were encouraging, large numbers
of households remained at risk of eviction and related adverse housing stability, employment, and
health outcomes.
8
References
Brodeur, A., A. E. Clark, S. Fleche, and N. Powdthavee (2021). Covid-19, lockdowns and well-being:
Evidence from Google trends. Journal of Public Economics 193, 104346.
Chauvet, M., S. Gabriel, and C. Lutz (2016). Mortgage default risk: New evidence from Internet search
queries. Journal of Urban Economics 96, 91–111.
Chetty, R., J. Friedman, N. Hendren, M. Stepner, et al. (2020). How did Covid-19 and stabilization policies
affect spending and employment? A new real-time economic tracker based on private sector data. NBER
Working Paper 27431.
9
File and source
- File
- philfed-wp22-10-an-gabriel-tzurilan.pdf
- Size
- 349,533 bytes
- SHA-256
- 49be7d4149c48ed8c5355eedecbc6f938b1ac07d26ecbdc9949c3216b553c16c
- Original
- www.philadelphiafed.org