Abstract

The current coronavirus pandemic is an unprecedented public health challenge that has devastating economic
impacts for households. Using a sample of 230,540 respondents to online surveys in 17 countries in Latin America and
the Caribbean, we show that the economic impacts are large and unequal: 45% of respondents report that a household
member lost a job, and among households owning small businesses, 59% of respondents report that a household
member closed their business. Among households with the lowest income prior to the pandemic, 71% report that a
household member lost their job and 61% report that a household member closed their business. Declines in food
security and healthiness are among the disproportionate impacts. Our results provide evidence that the current public
health crisis will exacerbate economic inequality, and they are among the first estimates of the labor market and well-
being impacts of the pandemic in developing countries1 .

J E L Codes: I14,D14,I18,I32

KEYWORDS
Coronavirus pandemic, Public health, Labor markets, Inequality

1 We would like to thank Sebastián Espinoza and María Paula Medina for superb research assistance. We would like to thank Julián Cristia for his encourage-
ment and advise with the project. We also want to thank Tom Sarrazin, Sebastián Oliva and Pablo Bachelet for their tremendous support in the dissemination
of the survey. We also want to thank the support of IDB’s country offices, country representatives and country economists for their invaluable input in cus-
tomizing the questionnaires for each country. In particular, we want to thank Marta Ruiz Aranz and Diether Beuermann for their support in the rollout of
the surveys in Central America and the Caribbean. We would like to thank Ana María Ibáñez, Eric Parrado, Michael Lovenheim, and the team of researchers
in IDB’s Research Department for valuable support and feedback in different parts of the project. Opinions, findings, conclusions and recommendations
expressed here are those of the author and do not necessarily reflect the views of the Inter-American Development Bank.
2 Bottan, Hoffmann and Vera-Cossio

1 | AUTHOR SUMMARY

Using a household survey of 230,540 respondents in 17 developing countries, we document the devastating labor
market impacts of the coronavirus pandemic. We find that the relationship between job loss and business closure is
monotonically decreasing with income prior to the pandemic, exacerbating existing inequality. These unequal labor
market impacts lead to unequal impacts beyond income, including food security and nutrition. Cross-country com-
parisons suggest that the impacts of the pandemic are stronger in countries with higher rates of informality. Not only
are households from high-informality countries losing their livelihoods at higher rates, but they are also less resilient
to those negative shocks. We provide some evidence suggesting that differences in the strength of lockdown poli-
cies across countries and the ability to work from home can be related to these impacts. Furthermore, labor market
impacts affect policy preferences, as support for policies to contain the coronavirus is weaker among those experienc-
ing job loss or business closure. Our results are among the first and only estimates of the economic impacts of the
coronavirus pandemic on households outside of developed nations, and we find impacts that are larger than those
reported in developed countries.

2 | INTRODUCTION

Several prominent economists have posited that inequality is the principal economic issue of our era [1, 2, 3, 4]. Recent
economic slowdowns, such as the Great Recession of 2008-2009, increased economic inequality [5, 6, 7]. However,
compared to other economic downturns, the COVID-19 pandemic alters economic activity through different channels
and on an accelerated timeline.
To slow the spread of COVID-19, governments have implemented regulations that require social distancing, close
non-essential businesses, restrict travel, and in many cases require residents to remain in their homes [8]. The human
interactions that drive the economy, such as working together in enclosed areas and enjoying entertainment activities,
have been discouraged, restricted, or banned. Residents are complying with these measures, report strong support
for these regulations [9, 10, 11], and actively seek out information [12]. Although these measures are necessary for
public health, recent evidence from developed countries suggests that they have negative economic impacts in the
short run [13, 14], and that they can potentially deepen pre-existing gaps [15]. These impacts could be exacerbated
in developing countries with more vulnerable workers and firms, as well as lower levels of state capacity to face the
pandemic.
Using a large-scale online household survey, we show that the COVID-19 pandemic has unequal economic im-
pacts on households in Latin America and the Caribbean that will further exacerbate inequality. Research on economic
inequality is particularly relevant in Latin America and the Caribbean. Although inequality and poverty declined over
the past decade [16, 17], prior to the pandemic, the region still had the highest income inequality in the world [18],
and a large share of citizens were vulnerable to falling back into poverty due to economic shocks [19].

3 | DATA COLLECTION AND VALIDATION

We conducted online surveys of households in eight South American countries (Chile, Colombia, Bolivia, Ecuador,
Guyana, Peru, Suriname, and Uruguay), four North and Central American countries (Costa Rica, El Salvador, Mexico,
and Panama), and five Caribbean countries (Dominican Republic, Bahamas, Barbados, Jamaica, and Trinidad and To-
bago), resulting in a sample of 230,540 observations. The survey was first launched on March 27, 2020 and was
Bottan, Hoffmann and Vera-Cossio 3

launched in all countries in our sample by April 17, 2020. With the exception of Costa Rica, data collection continued
until April 30, 2020. Therefore, the majority of the data were collected during the last two weeks of April 2020. The
number of observations by country ranges from 565 to 35,556 (see Table S1 in the Supplementary Material section
for country-specific details).
Figure S1 in the Supplementary Material section depicts the geographic coverage of our sample, showing the
number of observations as a share of population (in %) by sub-national region for each country in our sample. Our
sample achieved broad geographic coverage with observations in 92% of the sub-national regions and in 61% of the
localities of countries in our sample (see Table S1).
We recruited participants 18 years old and over for the survey using paid advertisement posts on Facebook and
Instagram. We used keywords with broad appeal, such as fútbol (soccer) or names of local celebrities, to avoid selecting
participants based on COVID-19 knowledge or interest. The Supplementary Material section contains additional
information about the social media campaigns and shows the posts that were used for Uruguay.
The questionnaire was largely standardized across countries to allow pooling the data across countries. The pri-
mary objective of the surveys was to measure the negative economic and well-being impacts of the current pandemic
on households in Latin America and the Caribbean. For this reason, the questionnaire focused on collecting data on
labor market outcomes, financial situation, and social program enrollment. In addition, the survey collected informa-
tion on hunger, shortages of key goods, and agreement with different policies to slow the spread of COVID-19. A
copy of the questionnaire can be found in the Supplementary Material section.

3.1 | Data Validation

To validate our dataset, we compare demographic characteristics using data from the online and nationally repre-
sentative household surveys, collected through field visits. Columns (1) and (2) in Table S3 show that although the
respondents of the online surveys are more educated and more likely to be female, they do not differ substantially
in terms of household structure and income levels. Second, Columns (3) and (4) conduct an out-of-sample validation
exercise and show that by re-weighting the online responses by the inverse probability of being in the nationally rep-
resentative sample, the differences in demographic characteristics vanish. The Supplementary Material describes the
steps to estimate weights for the online survey and the validation exercise.
In most of our analysis, we re-weight observations in our sample to achieve national representativeness (see
Supplementary Material). These estimates also weight observations according to country population to account for
differences in sample size across countries. The exception is Figure 3, which re-weights observations to account for
temporal changes in the sample. See Supplementary Material for estimation details. All results are robust to not using
weights at all, as shown in Figures S2-S4 and Table S4 in the Supplementary Material.

4 | RESULTS

We document that the COVID-19 pandemic has negative economic impacts for a broad segment of the population
and that these impacts are magnified for those with lower income prior to the pandemic. Overall, 45% of respondents
report that a household member lost a job and, among households owning small family businesses, 58% of respon-
dents reported that a household member closed their business. The recall period for these questions was randomized
between 1 week, 2 weeks, and 1 month. We find that the job-loss rates range between 42% for a recall period of
one week to 47% for a recall period of a month. We also find that the rates of business closures remain constant at
4 Bottan, Hoffmann and Vera-Cossio

58% across recall periods. Compared to similar statistics in the United States (43% in the case of small businesses
[14]), we find greater rates of business closure, implying that the economic impacts of the pandemic may be stronger
in developing countries.
These overall effects obscure highly unequal impacts across income levels prior to the pandemic. After accounting
for fixed factors by country, the percentage of households reporting job losses declines monotonically with income
in January 2020, prior to the onset of the pandemic (Figure 1). In the case of business closures, the decline is similar
though less dramatic. Households reporting income of less than the national monthly minimum wage for January 2020
experienced the largest impacts, with nearly 71% reporting that a household member lost their job and 61% reporting
that a household member closed their business. This contrasts sharply with the impacts reported by respondents with
the highest household incomes. Among the highest income respondents, only 14% reported that a household member
lost their job, and 54% reported that a household member closed their business.
100
80 60
Share %
4020
0
-20

<1 1-2 2-3 3-4 4-6 6-8 8-11 >11

Income in Jan. 2020 (in min. wages)

Job loss Business closure

FIGURE 1 Higher rates of job loss and business closure among households in the lowest income group. Point estimates
and 95% confidence intervals for regressing outcome on income bin indicators and country fixed effects. Data are weighted
using within- and cross-country weights. See Empirical Methods in Supplementary Material for details.

One explanation for these patterns is that high levels of informality in the region may limit the ability of the most-
vulnerable households to keep their sources of income. Using the share of self-employed workers as a proxy for the
share of informal workers in the labor market, Figure 2 shows that labor market informality is positively correlated
with loss of livelihood. Since informality rates are high in most developing countries, this result provides a novel
explanation of why labor markets in developing countries are particularly hard-hit during the crisis.
Our data suggest two important explanations for the higher vulnerability of households in developing countries
with high levels of informality. First, the type of policies to prevent the spread of the virus affects informal workers
more. Our survey captures data from countries without enforced mobility-restriction policies or curfews such as
Uruguay, but also more-stringent policies such as mandatory quarantines and closures of non-essential businesses,
as in the case of Bolivia or Peru. As most informal and self-employed workers tend to work in jobs that are prone to
contact with other people, such as those in the retail or services sectors, as opposed to office or industry jobs, the
Bottan, Hoffmann and Vera-Cossio 5

Loss of livelihood and self-employment

80
PER

COL
ECU 78

75
74 74

70
PAN BOL
69 SLV 68
65
MEX
BAH 66
% of respondents

64 DOM
BRB
62
60

61
60 TTO
GUY
CRC
57
55

CHL
56
55 JAM
53
52
50
45
40

URU

37
35
30

SUR
29

10 20 30 40 50 60 70
Share of self-employed workers (WDI)

FIGURE 2 Higher rates of livelihood loss in countries with higher informality. Each dot represents the share of
respondents who report that a household member lost a job or closed a business. Data are weighted using within- and
cross-country weights.

latter set of policies may lead to larger disruptions in labor markets. Indeed, our sample indicates that the share of
respondents reporting job losses in their household during April (69%) is substantially higher in countries with national
or local mandatory quarantines, relative to those of countries that did not implement mobility-restriction measures
(34%) or only implemented curfews (54%). See Supplementary Material section for a list of countries by type of policy.
Second, differences in the ability to telework could also be one reason that the negative impacts of the pandemic
are concentrated among households with lower incomes. Among respondents that are still employed, we find that the
share of respondents that report working from home during the past week is monotonically increasing with household
income in January 2020. Thirty percent of workers from households with incomes below the national minimum wage
reported working from home, while 76% of workers from the highest-income households report working from home.
Impacts on job loss and business closure translate into reductions in income. Table 1 reports within-locality
changes in outcomes as a response to job loss or business closure. Column 1 of Panel A shows that respondents
who reported a job loss or business closure are 24 percentage points (p-value<0.01) more likely to report a reduction
in income. Overall, 71% of respondents report that they expect their household income in April 2020 to be lower
than their household income in January 2020. Thirty-one percent of respondents reported household income less
than the national monthly minimum wage for January 2020, and 56% of households report that they expect their
household income to be less than the national minimum wage in April 2020. Figure 3 shows that the distribution of
household income expected in April 2020 is a leftward shift of the distribution of household income in January 2020.
In particular, the share of households with incomes marginally above the national minimum wage declines between
January and April 2020, suggesting that many vulnerable households expect to fall into poverty.
We find that job loss and business closure lead to reductions in health and food security. Columns 2-3 of Panel A
of Table 1 show that households with a job loss or business closure are 12 percentage points (p-value<0.01) more likely
to suffer from hunger and 8 percentage points (p-value<0.01) more likely to have a less healthy diet relative to their
6 Bottan, Hoffmann and Vera-Cossio

TA B L E 1 The loss of livelihood during the pandemic is linked to changes in nutrition and policy support

Panel A: Impacts on income, food security, and health

(1) (2) (3)

Decreased income Went hungry Eats less healthy

Lost job or closed business 0.241*** 0.127*** 0.085***

(0.008) (0.008) (0.008)

Observations 186,058 198,190 173,956

Adjusted R2 0.487 0.602 0.430

Panel B: Impacts on transfers and policy support

(1) (2) (3)

Gift/Loan Gov. Priority Lockdown (>=month)

Lost job or closed business 0.225*** -0.027*** -0.042***

(0.008) (0.008) (0.010)

Observations 198,017 196,076 125,359

Adjusted R2 0.479 0.482 0.540

∗p < 0.1, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

The table reports regression coefficients capturing the relationship between livelihood losses during the pandemic and
outcomes. Each column reports the results of a regression of the dependent variable on an indicator of whether any household
member either lost a job or closed a business and a vector of covariates. In addition, all regressions control for locality × day of
survey completion fixed effects (18,764), as well as economic-sector fixed effects. Standard errors are clustered at the locality
level (3,165). Data are weighted using within- and cross-country weights. See Empirical Methods in Supplementary Material
section for details.
Bottan, Hoffmann and Vera-Cossio 7

60
40
Share (%)
20
0

<1 1-2 2-3 3-4 4-6 6-8 8-11 >11

Income in minimum wages

Income in Jan. 2020 Income in Apr. 2020

FIGURE 3 The share of households in the bottom part of the income distribution is expected to increase. Shares of
households in each income bin for incomes reported for January 2020 and April 2020. Data counts weighted using within- and
cross-country weights. See Empirical Methods in Supplementary Material for details.

diet prior to the pandemic. The results suggest that the impacts of the pandemic may have long-lasting consequences
linked to declines in the stock of human capital. Consistent with the results from Figure 2, we also find that the
consequences of the loss of livelihoods on food security are stronger in countries with higher levels of informality
(See Table S5). This suggests that the structure of labor markets not only magnified the exposure to job losses and
business closures, but also magnified the impacts of the loss of livelihoods on household welfare as informal workers
may have lower access to formal safety nets.

We find that households cooperate across income levels to smooth the negative economic impacts of the pan-
demic. Seventy percent of respondents who reported January 2020 household income less than the national minimum
wage reported that a household member received a gift or loan from a friend or relative. In contrast, only 26% of re-
spondents with the highest incomes in January 2020 report that a household member received a gift or loan from a
friend or relative. This pattern reverses for providing a gift or loan. Thirty percent of households with January 2020
income less than the national minimum wage reported that a household member provided a gift or loan compared to
57% of households with the highest incomes January 2020. Across all income categories, the reception of a gift or
transfer is concentrated among households that reported losing a job or closing a business. Column 1 of Panel B of
Table 1 shows that households that lost their livelihoods during the crises are 22 percentage points (p-value<0.01)
more likely to receive transfers from relatives or friends. This highlights the importance of informal social protection
networks as a tool to cope with the negative impacts of the pandemic.

Despite these substantial economic impacts, there is strong support for measures to slow the spread of the coro-
navirus. Overall, 77% of respondents agree with a statement that the top priority of the national government should
be to stop the pandemic, and 54% of respondents think that non-essential businesses should be closed for one more
month. Although support for these policies is broad, it is likely to decrease as more households lose their livelihoods.
Column 2 of Panel B from Table 1 shows that the probability of agreeing with the idea that the Government’s priority
8 Bottan, Hoffmann and Vera-Cossio

70
60
Share (%)
50
40
30

30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
Days since first case

Did not lose job or close business

Lost job or closed business

FIGURE 4 Support for extending lockdown policies declines more among households that lost their livelihoods. Point
estimates and 95% confidence intervals for the share of respondents supporting extending lockdown policies according to days
since first COVID-19 case in the country. Data are weighted using within- and cross-country weights. See Empirical Methods in
Supplementary Material section for details.

should be fighting the pandemic is 2 percentage points (p-value<0.01) lower among households that experienced a
job loss or business closure. Column 3 of Panel B shows that the loss of livelihoods during the pandemic is linked to a
4 percentage-point decline (p-value<0.01) in the probability of agreeing with closing non-essential business for one
more month. Thus, the support for policies to slow the spread of the coronavirus is fragile. Indeed, Fig. 4 shows that,
as days go by, the decline in support for keeping businesses closed for one more month declines faster among house-
holds that lose their livelihoods. One important implication is that, without further assistance to impacted households,
compliance with mobility restriction policies is likely to decline.

5 | DATA ANALYSIS

| Figure 1

In Figure 1 we show how job loss and business closures vary throughout the income distribution. Job loss is a dummy
variable that equals one if the respondent indicates that they lost their job within the last week, two weeks or month
(these horizons were randomly assigned). Similarly, the variable for business closures equals one if subjects report
that their business was forced to close by the government or by lack of demand within the same randomly assigned
time horizon.
To explore differences in job loss and business closure across the income distribution, we first aggregate responses
at the country and income bracket level using within-country weights as described in the Supplementary Material
section. We then regress by OLS the country averaged outcome on indicators for income bin and control for country
fixed effects, weighting by country population. Because we only have 17 countries in our sample, using standard
Bottan, Hoffmann and Vera-Cossio 9

clustered standard errors would result in incorrect inference. Therefore, we follow Bertrand et al. [20] by aggregating
the data at the country level by income category. Figure 1 shows the point estimates and 95% confidence intervals
for each income bin average. As a robustness test, we show results for the raw pooled data, without applying weights,
in Figure S2 in the Supplementary Material.

| Figure 2

We present a cross-country comparison of the shares of households who lost their livelihoods as a function of the
share of self-employed workers in the economy. For this, we identified the respondents who reported that at least one
household member lost their job during or had to close their businesses during the survey reference period, and we
obtained shares using weights to correct for sampling issues (see Supplementary Material section). Data corresponding
to the share of self-employed workers was obtained from the World Bank’s World Development Indicators.

| Figure 3

We present the share of household income in each income bin for incomes reported by respondents corresponding to
January 2020 and April 2020 in Fig. 3. We restrict the sample to responses collected from April 13 to May 1, 2020 to
include all countries. Shares are calculated by re-weighting counts by within- and across-country weights as detailed
in the Supplementary Material section Estimation of Weights. As a robustness test, we show the results for the raw
pooled data without applying weights in Figure S3.

| Figure 4

Figure 4 shows the share of respondents who support extending the lockdown policies by at least one month over
time since the first COVID-19 case in the country. One challenge to exploring how support changes over time is that
the timing of responses varies between countries. Therefore, over time we have new countries entering the sample,
and the change in sample composition could bias our results. We address this issue in two steps. First, we construct
new weights and re-weight respondents after day 31 to match those responding at the beginning of the series (days
30 and 31) on observable demographic characteristics.

We estimated a model by fitting a logistic function (logit) and computed predicted probabilities of having re-
sponded to the survey on day 30 or 31 relative to first case. As explanatory variables we include demographic char-
acteristics such as household size, age indicators for: presence of children or elderly, having gone hungry, woman,
education primary or less, college education or higher, household income categories (<1, 1-2,2-3, 8-11,>11), and day
of week. We then used inverse probability weights based on the propensity score obtained.

Second, we separately estimate OLS regressions for respondents whose household member lost job or closed
business and for those who did not, where the dependent variable is an indicator variable that equals one if they
support extending the lockdown by at least a month, and zero otherwise. The independent variables are indicators
for days since first COVID-19 case, day of week indicators, and country fixed effects. The regression is weighted
by the inverse probability weight described in step one and we employ robust standard errors. Figure 4 presents
the point estimates and 95% confidence intervals for these estimates. We present similar estimates when using the
pooled raw data without re-weighting in Figure S4 in the Supplementary Material.
10 Bottan, Hoffmann and Vera-Cossio

| Table 1

In Table 1, we report comparisons of differences in the relevant outcomes among respondents whose households
experienced a loss of livelihood during the pandemic period. To control for time-varying characteristics we perform
comparisons focusing on respondents who are in the same locality and who completed the survey during the same
day. We exploit granular data from over 3,000 localities in 17 countries. Thus, we focus our analysis on the subset
of 18,000 locality-date-of-response cells that include more than one observation. This approach allows us to isolate
time-varying locality shocks and thus purge regional confounding factors. In addition, we control for industry sector
fixed effects to prevent differences in the sectors related to the household’s main economic activities from driving the
results, as exposure to the effects of the pandemic may vary across sectors.
We operationalize our approach by estimating the following specification, which is similar to that used in Gertler
and Gruber [21]:

Yi,l,c,t = β Lostlivelihoodi,l,c,t + Xi ,c,l ,t Σ + δ l ,c,t + θs + i ,l ,c,t (1)

Here, Yi ,l ,c,t denotes the outcome of interest corresponding to respondent i in locality l from country c collected
in date t . Lost livelihoodi ,l ,c,t is an indicator of whether any member of the respondent’s household lost her/his job
or closed her business during the past week, two weeks, or month. Xi ,c,l ,t is a vector of demographic characteristics
of the respondent (age, education level and gender) and of household characteristics (household size, presence of
children younger than 5, presence of school-age children, and people 60 years old or older). δl ,c,t denotes locality-
date fixed effects, and θs denotes industry fixed effects based on the main pre-pandemic source of income of the
respondent’s household. i ,c,l ,t denotes unobserved shocks. To account for possible serial correlation of outcomes
within localities, we cluster the standard errors at the locality-country level.
The parameter of interest, β , is reported in Table 1 and captures within-locality differences in outcomes between
respondents whose households experienced a loss of livelihood during the pandemic and those who did not. The
reported models are estimated using weights to achieve country-level representativeness, to correct for differences
in sample size across countries, and to provide weights based on country population size. All results are robust to
excluding these weights (see Robustness Section in Supplementary Material).
To investigate whether the correlates of job loss or business closures and household outcomes are stronger in
countries with high levels of informality, we also estimate the following specification:

Yi,l,c,t = β 1 Lostlivelihoodi,l,c,t + β 2 Lostlivelihoodi,l,c,t × Self − employmentc (2)

+Xi ,c,l ,t Σ + δ l ,c,t + θs + i ,l ,c,t

where Self-employment denotes the share of self-employed workers in country c and was obtained from the
World Bank’s World Development Indicators, using the most-recent observations for each country. We also report
results using the share of informal workers (as a share of non-agricultural workers) in each country. This information
was not available for Suriname, Jamaica, Trinidad and Tobago, Barbados, and Bahamas. See results in Table S4 in the
Supplementary Material.
Dependent variables: Went hungry is an indicator of whether any household member went hungry during the past
week due to lack of food. Eats less healthy takes the value of one if the respondent somewhat or totally agrees with the
Bottan, Hoffmann and Vera-Cossio 11

statement "I eat more unhealthy foods than normal." Gifts/Loans is an indicator of whether the respondent’s household
received a gift or transfer from either friends or relatives during the preceding week. Gov. Priority is an indicator of
whether the respondent somewhat agrees or totally agrees with the statement "The government’s priority should be
to stop the spread of the pandemic." Lockdown (>= month) is an indicator of whether the respondent reports agreeing
with closing non-essential business for one month or longer. As this question was asked only of people who reported
agreeing with policies that require non-essential businesses to close, Lock d ow n (>= mont h) takes the value of zero
when the respondent reported not supporting measures of keeping non-essential businesses closed at all, regardless
of the time.

6 | ROBUSTNESS

We demonstrate that all our results are robust to not re-weighting observations. Reassuringly, all results are both
qualitatively and quantitatively similar. See SI Appendix, Figures S2-S4 and Panel A of Table S4 in the Supplementary
Material for unweighted results.
In addition, Panels B and C of Table S4 in the Supplementary Material report robustness related to using only
country-date fixed effects instead of locality-date fixed effects as controls, both with and without weights. Reassur-
ingly, our results are both quantitatively and qualitatively similar.

7 | CONCLUSIONS

These results show that the negative economic impacts of the COVID-19 pandemic have been concentrated among
those with lower incomes prior to the pandemic. This finding is important from both social and economic perspectives.
Inequality is an important social outcome in itself and also has important economic implications. Although further
research is needed, several studies have found that current inequality is negatively correlated with future economic
growth [22, 23], and in particular, inequality driven by the lower tail of the income distribution stunts economic
growth [24]. This implies that the unequal economic impacts of this short-term public health pandemic could have
long-term implications for economic growth. In addition, our results indicate that country-level rates of informality in
labor markets are linked to stronger negative impacts and low resilience, suggesting that prompt policy responses are
needed to protect informal workers.
12 Bottan, Hoffmann and Vera-Cossio

8 | SUPPLEMENTARY MATERIAL

Figure S1
Sample has broad geographic coverage at the sub-national level. The sub-national regions of each country in the
sample are shaded according to number of observations as a share of population (in %). Sources of population data
for each country are shown in the Supplementary Material Section.

Figure S2
Higher rates of job loss and business closure among households in the lowest income group. Point estimates and
95% confidence intervals for regressing outcome on income bin indicators and country fixed effects. Pooled data (no
weights), robust standard errors. See the Empirical Methods section in the main text for more details.

Figure S3
The share of households in the bottom part of the income distribution is expected to increase. Shares of households
in each income bin for incomes reported for January 2020 and April 2020. Pooled data (no weights). See the Empirical
Methods section in the main text for more details.

Figure S4
Support for extending lockdown policies declines more among households that lost their livelihoods. Point estimates
and 95% confidence intervals for the share of respondents supporting extending lockdown policies according to days
since first COVID-19 case in the country. Pooled data (no weights). See the Empirical Methods section in the main
text for more details.

Table S1
Date of Launch and Number of Observations by Country.

Table S2
Parameters of logit models for the probability of being in the nationally representative survey.

Table S3
Differences between Online Survey Data and Household (field) Survey Data.

Table S4
Loss of livelihoods and changes in well-being, and policy support (unweighted results).

Table S5
Impacts of livelihood loss on income and food security by labor-market characteristics.
Bottan, Hoffmann and Vera-Cossio 13

Supplementary Information
| Data Collection and Construction
| Questionnaire
The questionnaire asks about monthly total household income in January 2020 and expected monthly total household
income in April 2020 in ranges constructed as multiples of the minimum wage. The questionnaire asked if total
household income was reduced during the past week. The questionnaire asked whether a household member lost
their job or closed their business. The recall period for these questions was randomized between 1 week, 2 weeks,
and 1 month. For respondents who remain in the labor market, the questionnaire asks whether the respondent
worked outside the home or from home during the past week, and we code a variable to represent teleworking if
the respondent reported working from home. Using the past week as the recall period, the questionnaire asks if
any member of the household went hungry due to lack of food and asks respondents to report the strength of their
agreement with a statement "I am eating less healthy than normal" on a scale from 1 (complete disagreement) to 5
(complete agreement). The questionnaire asks if any member of the household received (gave) a transfer or a loan
from (to) another household during the past week. The questionnaire also asks respondents to report the strength of
their agreement with a statement that COVID-19 should be the top priority of the national government on a scale
from 1 (complete disagreement) to 5 (complete agreement) and asked whether respondents think that non-essential
businesses should close.
Sample questionnaires in English, Spanish and Dutch can be downloaded here: https://www.dropbox.com/sh/
uuv17cfaz94kw4h/AAAWvgJUNYEBHhNxZPcQVv3Za?dl=0.

| Data Construction
The questionnaire was implemented online using Qualtrics. Column (1) of Table S1 displays the launch date of the
survey in each country. We construct the data set in several steps and the resulting number of observations that
comprise the sample for each country is shown in column (2) of Table S1. First, we restrict the sample to completed
surveys. Overall, approximately 59% of surveys that are started are completed. Second, we restrict the sample to
surveys associated with IP addresses within the borders of the country for which the respondent is completing the
survey. Across all countries in the sample, over 99% of completed surveys comply with this criteria. Third, Qualtrics
flags surveys that are completed on the same device and likely to be repeat surveys completed by the same individual
or household based on cookies. This is an imperfect filter. For example, it will not recognize repeated surveys by the
same individual or household that are completed on different devices. Surveys flagged as repeats comprise less than
2.3% of completed surveys and we drop these surveys from the sample.
14 Bottan, Hoffmann and Vera-Cossio

| Recruitment Materials

| Recruitment through Social Media Posts

We recruited participants for the survey using paid advertisement posts on Facebook and Instagram. To be eligible
to participate in the survey, participants had to be at least 18 years old. In each country, we targeted the social
media campaign to people ages 19 and up. We utilized keywords that have broad appeal in the country, such as
futbol (soccer) and the names of futbol stars or local celebrities, and that are unrelated to COVID-19 in order to avoid
selecting respondents based on their knowledge or experience regarding the current pandemic. In countries in which
the average age of our sample was high or expected to be high, we utilized a second, simultaneous social media
campaign with the same images and keywords, but specifically targeted to millennials aged 24-35.
In each country, we used three posts, an image of a male wearing a face mask, an image of a female wearing a
face mask, and an image of an iconic landmark in that country. The posts were adapted to each country by changing
the image of the landmark, changing the image of the male and female if appropriate, changing the country name in
the text, and changing the language (Spanish, English, or Dutch) of the posts. Facebook algorithms select which post
to show each user. As an example, we show the posts that we used for Uruguay below.
Bottan, Hoffmann and Vera-Cossio 15

| Re-weighting and external validity

| Estimation of weights
For each country, we modeled the probability of being in the nationally representative sample as a function of de-
mographic, time-invariant characteristics such as gender, age and education of the respondent, indicators for the
presence of children of 5 years old or younger in the household and elderly (60 years old or older) in the households,
as well as the number of household members and the number of children enrolled in school. Finally, we also include
region-specific indicators.
We conducted this process country by country. For each country, we used the most recent nationally represen-
tative survey available in the Inter-American Development Bank harmonized survey data repository. We estimated
the model by fitting a logistic function (logit) and computed predicted probabilities of being in the nationally repre-
sentative data set (p̂ i ,c ) for each respondent i in country c (see Table S2). We then used inverse probability weights
(i pwi ,c = 1/(1 − p̂ i ,c )) to, at least in terms of observable characteristics, resemble those from nationally representative
surveys.
To prevent differences in response rates from driving the results, we re-scale the within-country weights i pwi ,c
by the inverse share of re-weighted number of responses per country, relative to the country’s population size
(P opul at i on c /( iN c i pwi ,c )).
Í

| Out-of-sample validation
We validate the country-specific weights by looking at their out-of-sample performance. For each country we ran-
domly selected 60% of the observations in the online survey and the household (field) surveys. We used this sample
to estimate logit models. We then use the resulting models to predict the probability of being observed in the house-
hold survey over the observations in the remaining 40% of the sample, which we call the testing sample. We then
re-weight the observations in the testing sample and compare the adjusted means of demographic characteristics
to those obtained from household (field) survey data. Table S3 shows this exercise in Columns 3 and 4. We were
able to reduce the differences in respondent characteristics between the online and household surveys. This out-of-
sample exercise suggests that the improvement in balance is not driven by overfitting. We also show that our online
data do not differ substantially from the household survey data in non-targeted moments such as the proportion of
respondents by income category.

| Countries by type of mobility-restriction policies

The following list details the type of mobility restrictions implemented by each country. Data were collected from
official government websites and press articles. The information was collected on April 23, 2020.
No mandatory policies: Bahamas, Suriname, Trinidad and Tobago, Uruguay. Curfews: Dominican Republic, Guyana.
Local Quarantines: Chile, Jamaica, Mexico. National Quarantines Barbados, Bolivia, Colombia, Costa Rica, Ecuador,
El Salvador, Panama, Peru.
16 Bottan, Hoffmann and Vera-Cossio

| Datasets
| Population and Household Survey Data

Year and Source of Household Survey Data by Country

Country Survey Name Year Link to source

Bahamas LFS 2014 http://www.bahamas.gov.bs/wps/portal/public

Barbados BSLC 2016 http://sistemasintegrales.cl/project/barbados-survey-of-living-conditions/

Bolivia ECH 2018 https://www.ine.gob.bo/index.php/herramientas/bases-de-datos-catalogo-anda/bases-de-datos-encuestas-sociales/

Chile Casen 2017 http://observatorio.ministeriodesarrollosocial.gob.cl/casen-multidimensional/casen/basedatos.php

Colombia GEIH 2018 http://microdatos.dane.gov.co/index.php/catalog/659/get_microdata

Costa Rica ENAHO 2018 https://www.inec.cr/noticias/enaho

Dominican Republic ENCFT 2018 https://www.bancentral.gov.do/a/d/2539

Ecuador ENEMDU 2018 https://www.ecuadorencifras.gob.ec/estadisticas/

El Salvador EHPM 2018 http://www.digestyc.gob.sv/index.php/temas/des/ehpm.html

Guyana LFS 2018 https://statisticsguyana.gov.gy/data/databases/

Jamaica SLC 2014 https://statinja.gov.jm/living_conditions_poverty.aspx

Mexico ENIGH 2018 https://www.inegi.org.mx/programas/enh

Panama EHPM 2018 https://www.inec.gob.pa/publicaciones/Default2.aspx?ID_CATEGORIA=5&ID_SUBCATEGORIA=38

Peru ENAHO 2018 https://webinei.inei.gob.pe/anda_inei/index.php/catalog/672

Suriname SLC 2017 https://statistics-suriname.org/en/

Trinidad & Tobago CSSP 2015 https://cso.gov.tt/methods/classifications/

Uruguay ECH 2018 http://www.ine.gub.uy/encuesta-continua-de-hogares1

We used the most recent household survey available at the IDB’s harmonized database repository. Each survey can be accessed on the websites of the specific institutions in
each country. The table provides links to original sources for data access. In some instances, the microdata must be requested directly from the competent body due to privacy
restrictions.

Year and Source of Population Data by Country

Country Year Link to Source

Bahamas 2010 https://www.bahamas.gov.bs/wps/wcm/connect/22f9b2b0- 68fa- 4a26- 8bd8- 474952e42dc2/Population+Projection+Report+2010-

2040.pdf?MOD=AJPERES

Barbados 2010 https://web.archive.org/web/20170118220332/http://www.barstats.gov.bb/files/documents/PHC_2010_Census_Volume_1.pdf

Bolivia 2020 https://www.ine.gob.bo/subtemas_cuadros/demografia_html/PC20106.htm

Chile 2020 https://www.ine.cl/estadisticas/sociales/demografia-y-vitales/proyecciones-de-poblacion

Colombia 2020 https://www.dane.gov.co/index.php/estadisticas-por-tema/demografia-y-poblacion/proyecciones-de-poblacion

Costa Rica 2020 https://www.inec.cr/poblacion/estimaciones-y-proyecciones-de-poblacion

Dominican Republic 2020 https://www.one.gob.do/demograficas/proyecciones-de-poblacion

Ecuador 2020 https://sni.gob.ec/proyecciones-y-estudios-demograficos

El Salvador 2020 https://www.transparencia.gob.sv/search?utf8=%E2%9C%93&ft=Proyecciones+municipales

Guyana 2012 https://statisticsguyana.gov.gy/publications/#elementor-tab-content- 1465%20%3E%20ul:nth-child(3)%20%3E%20li:nth-

child(1)%20%3E%20span

Jamaica 2018 https://statinja.gov.jm/Demo_SocialStats/PopulationStats.aspx

Mexico 2020 https://www.gob.mx/cms/uploads/attachment/file/63977/Documento_Metodologico_Proyecciones_Mexico_2010_2050.pdf

Panama 2020 https://www.inec.gob.pa/publicaciones/Default3.aspx?ID_PUBLICACION=499&ID_CATEGORIA=3&ID_SUBCATEGORIA=10

Peru 2020 http://proyectos.inei.gob.pe/web/biblioineipub/bancopub/Est/Lib0846/libro.pdf

Suriname 2012 https://statistics-suriname.org/en/census-statistics- 2012/

Trinidad and Tobago 2011 https://cso.gov.tt/census/2011-census-data/

Uruguay 2020 http://www.ine.gub.uy/estimaciones-y-proyecciones

Bottan, Hoffmann and Vera-Cossio 17

| Supplementary Figures

F I G U R E S 1 Sample has broad geographic coverage at the sub-national level. The sub-national regions of each
country in the sample are shaded according to number of observations as a share of population (in %). Sources of
population data for each country are shown in the Supplementary Text Information Section.
18 Bottan, Hoffmann and Vera-Cossio

100
80 60
Share %
4020
0
-20

<1 1-2 2-3 3-4 4-6 6-8 8-11 >11

Income in Jan. 2020 (in min. wages)

Job loss Business closure

F I G U R E S 2 Higher rates of job loss and business closure among households in the lowest income group. Point
estimates and 95% confidence intervals for regressing outcome on income bin indicators and country fixed effects.
Pooled data (no weights), robust standard errors. See the Empirical Methods section in the main text for more details.
Bottan, Hoffmann and Vera-Cossio 19

60
40
Share (%)
20
0

1 1-2 2-3 3-4 4-6 6-8 8-11 +11

Income in minimum wages

Income in Jan. 2020 Income in Apr. 2020

F I G U R E S 3 The share of households in the bottom part of the income distribution is expected to increase.
Shares of households in each income bin for incomes reported for January 2020 and April 2020. Pooled data (no
weights). See the Empirical Methods section in the main text for more details.
20 Bottan, Hoffmann and Vera-Cossio

70
60
Share (%)
50
40
30

30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51
Days since first case

Did not lose job or close business

Lost job or closed business

F I G U R E S 4 Support for extending lockdown policies declines more among households that lost their
livelihoods. Point estimates and 95% confidence intervals for the share of respondents supporting extending
lockdown policies according to days since first COVID-19 case in the country. Pooled data (no weights). See the
Empirical Methods section in the main text for more details.
Bottan, Hoffmann and Vera-Cossio 21

| Supplementary Tables

TA B L E S 1 Date of Launch and Number of Observations by Country

(1) (2) (3)

Country Launch No. of % of Localities

Date Observations with Observations

Chile 3/27/2020 35,556 97%

Bolivia 4/1/2020 25,970 83%

Panama 4/3/2020 15,521 100%

Uruguay 4/3/2020 21,191 64%

Peru 4/7/2020 25,452 47%

Mexico 4/13/2020 19,483 51%

Costa Rica 4/13/2020 9,151 90%

Colombia 4/15/2020 23,458 71%

El Salvador 4/16/2020 13,456 50%

Dominican Republic 4/16/2020 7,965 94%

Barbados 4/16/2020 2,072 100%

Jamaica 4/16/2020 2,547 91%

Guyana 4/16/2020 1,670 96%

Bahamas 4/16/2020 896 81%

Trinidad and Tobago 4/16/2020 4,683 100%

Ecuador 4/16/2020 18,688 68%

Suriname 4/17/2020 565 100%

Date of launch is the date on which the social media posts began. The date that the survey was rolled out in each country was largely
determined by bureaucratic processes and approvals. With the except of Costa Rica, data collection in each country continued until April
30, 2020. The number of observations for each country reflected the number in the sample after data cleaning. The percent of localities is
the percent of localities of each country for which we have one or more observations in the sample.
 22

TA B L E S 2 Parameters of logit models for the probability of being in the nationally representative survey
Bahamas Barbados Bolivia Chile Colombia Costa Rica Dominican Rep. Ecuador El Salvador Guyana Jamaica Mexico Peru Panama Suriname Trinidad & Tobago Uruguay

Household size -0.002 -0.575*** -0.206*** -0.123*** -0.073*** -0.061*** -0.084*** -0.047*** -0.195*** -0.326*** -0.048*** -0.085*** -0.097*** -0.197*** -0.067*** -0.125***

(0.034) (0.020) (0.008) (0.007) (0.013) (0.020) (0.009) (0.012) (0.031) (0.044) (0.008) (0.006) (0.011) (0.032) (0.009) (0.009)

# of children in school age -0.814*** 0.389*** 0.225*** 0.168*** 0.124 0.138** 0.563*** 0.153* 0.973*** 2.720*** 0.583*** 0.967*** 0.080 0.665***

(0.135) (0.033) (0.008) (0.040) (0.096) (0.066) (0.027) (0.089) (0.108) (0.201) (0.039) (0.033) (0.059) (0.051)

Children under 5 years old in household (%) -0.238** -0.455*** -0.348*** -0.264*** -0.257*** -0.835*** -0.299*** -0.891*** -3.600*** -0.689*** -1.104*** -0.097* -0.701***

(0.112) (0.030) (0.037) (0.093) (0.061) (0.023) (0.087) (0.082) (0.203) (0.035) (0.030) (0.056) (0.047)

Elderly in household (%) -0.520*** 0.702*** -1.010*** 0.038** -0.167*** 0.073* -0.329*** -0.072** -0.183*** -0.013 0.157 -0.103*** -0.280*** 0.028 0.138 0.066 0.183***

(0.133) (0.066) (0.030) (0.016) (0.024) (0.040) (0.058) (0.032) (0.037) (0.110) (0.155) (0.025) (0.022) (0.039) (0.177) (0.041) (0.026)

# of children enrolled in school -0.831*** -1.060*** -0.325*** -1.366*** -0.737*** -0.699*** -0.460*** -0.360*** -0.234*** -0.964*** -1.879*** -0.564*** -0.645*** -0.338*** -0.588*** -1.100*** -0.983***

(0.122) (0.065) (0.027) (0.016) (0.021) (0.033) (0.048) (0.028) (0.032) (0.100) (0.168) (0.021) (0.019) (0.034) (0.166) (0.040) (0.022)

Respondent is a female -0.071 0.000 -1.108*** -2.543*** -2.018*** -2.227*** -0.996*** -3.228*** -2.272*** -1.208** 1.027* -1.843*** -1.773*** -2.076*** -1.642 -1.671***

(1.122) (.) (0.186) (0.127) (0.182) (0.197) (0.264) (0.149) (0.143) (0.532) (0.580) (0.167) (0.192) (0.587) (1.061) (0.191)

Completed Primary -2.205** 0.110 -3.632*** -4.496*** -3.329*** -3.297*** -3.660*** -3.478*** -3.874*** -4.066*** -1.585*** -5.006*** -4.477*** -4.905*** -5.261*** -4.638***

(1.029) (0.349) (0.167) (0.126) (0.178) (0.197) (0.218) (0.145) (0.141) (0.505) (0.545) (0.163) (0.180) (0.580) (1.018) (0.190)

Completed Secondary -5.185*** -1.623*** -6.316*** -6.035*** -5.904*** -5.698*** -6.427*** -6.900*** -6.614*** -6.726*** 0.000 -6.797*** -6.041*** -8.058*** -10.927*** -5.377***

(1.021) (0.346) (0.165) (0.125) (0.176) (0.196) (0.217) (0.143) (0.139) (0.508) (.) (0.162) (0.180) (0.580) (1.070) (0.190)

University/Vocational Training or Higher 0.010** -0.010*** 0.039*** -0.017*** 0.004*** 0.011*** 0.016*** 0.007*** 0.006*** 0.009*** 0.018*** -0.017*** 0.015*** 0.019*** -0.003 0.017*** 0.002***

(0.004) (0.002) (0.001) (0.001) (0.001) (0.001) (0.002) (0.001) (0.001) (0.003) (0.005) (0.001) (0.001) (0.001) (0.006) (0.001) (0.001)

Age of respondent -0.385*** -0.181*** -0.398*** -0.452*** -0.611*** -0.500*** 0.044 0.214 -0.221*** -0.282*** -0.082* -0.011 -0.065**

(0.032) (0.026) (0.045) (0.055) (0.034) (0.040) (0.113) (0.151) (0.027) (0.023) (0.044) (0.048) (0.029)

N 6413 11314 48150 201380 159692 34367 19722 58403 61294 10323 13363 199884 115388 41597 5376 29008 103015

∗p < 0.1, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

The table presents estimates of coefficients from a logit model of the probability of being the nationally representative household survey as a function of demographic characteristics. The models were estimated separately for each country. All
models also included region-specific indicators, except in the case of Ecuador,in which data regarding regions were not available. Empty cells in the table imply that the relevant variable was not available in the household surveys. Robust
standard errors are presented in parenthesis.
Bottan, Hoffmann and Vera-Cossio
Bottan, Hoffmann and Vera-Cossio 23

TA B L E S 3 Differences between Online Survey Data and Household (field) Survey Data

Full Sample Testing sample

(1) (2) (3) (4)

Online HH Survey Online HH Survey

(Raw) (Sampling weights) (Re-weighted) (Sampling weights)

Household size 4.27 3.87 4.30 3.87

Children under 5 years old in household (%) 0.28 0.24 0.33 0.24

Elderly in household (%) 0.35 0.36 0.36 0.36

# of children enrolled in school 1.04 0.81 0.89 0.81

Respondent is a female 0.70 0.52 0.58 0.52

Completed Primary 0.03 0.36 0.31 0.36

Completed Secondary 0.22 0.32 0.30 0.32

University/Vocational Training or Higher 0.74 0.12 0.24 0.12

Age of respondent 38.13 43.07 39.62 43.04

Income categories

0-0.5 MW 0.10 0.14 0.17 0.14

0.5-1 MW 0.16 0.09 0.22 0.09

1-2 MW 0.21 0.20 0.22 0.20

2-3 MW 0.14 0.15 0.11 0.16

3-4 MW 0.11 0.10 0.08 0.10

4-6 MW 0.10 0.12 0.07 0.12

6-8 MW 0.06 0.06 0.04 0.06

8-11 MW 0.05 0.04 0.02 0.04

11+ MW 0.07 0.09 0.06 0.09

Countries 17 17 17 17

Observations 230,540 913,694 92,715 364,972

The table presents means of household and survey respondent demographic characteristics using data from the online survey and
nationally representative surveys, and pooling observations from all study countries (weighting by country size). Column (1) reports raw
means using all the observations from the online surveys. Column (2) reports means using all available observations in the household (field)
surveys using sampling weights. Column (3) reports means from the online survey data using only data from the testing sample (i.e., the
sample not used for the estimation of the inverse probability weights). Column (4) reports means using data from the household (field)
surveys corresponding to the testing sample. The testing sample corresponds to a randomly selected subsample corresponding to 40% of
all the observations in the online and household (field) surveys. MW stands for national minimum wage. The inverse probability weights are
computed based on logit models of the probability of being observed in the household survey which are estimated country by country. The
models include including age, gender, and education categories of the respondent as well as household-level demographic characteristics
such as the presence of children younger than 5 years old, the presence of elderly children in the household, # of children enrolled in school,
household size, as well as region fixed effects.
24 Bottan, Hoffmann and Vera-Cossio

TA B L E S 4 Loss of livelihoods and changes in well-being, and policy support (unweighted results)
Panel A: Locality X date Fixed effects - unweighted

(1) (2) (3) (4) (5) (6)

Decreased income Went hungry Eats less healthy Gift/Loan Gov. Priority Lockdown (>= mont h)

Lost job or closed business 0.293*** 0.096*** 0.073*** 0.213*** -0.026*** -0.030***

(0.007) (0.003) (0.003) (0.003) (0.002) (0.003)

Observations 188152 200594 176198 200421 198410 125359

Adjusted R2 0.235 0.140 0.036 0.157 0.035 0.430

Panel B: Country X date fixed effects (weighted)

(1) (2) (3) (4) (5) (6)

Decreased income Went hungry Eats less healthy Gift/Loan Gov. Priority Lockdown (>= mont h)

Lost job or closed business 0.245*** 0.162*** 0.069*** 0.216*** -0.036*** -0.008

(0.021) (0.025) (0.016) (0.018) (0.008) (0.018)

Observations 202832 215410 190065 215236 213188 139800

Adjusted R2 0.266 0.244 0.097 0.192 0.110 0.111

Panel C: Country X date fixed effects (unweighted)

(1) (2) (3) (4) (5) (6)

Decreased income Went hungry Eats less healthy Gift/Loan Gov. Priority Lockdown (>= mont h)

Lost job or closed business 0.291*** 0.100*** 0.074*** 0.212*** -0.026*** -0.031***

(0.023) (0.010) (0.005) (0.009) (0.003) (0.008)

Observations 204464 217249 191801 217075 214980 141617

Adjusted R2 0.236 0.124 0.031 0.149 0.026 0.026

∗p < 0.1, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Panel A reports unweighted regression coefficients capturing the relationship between livelihood losses during the pandemic and outcomes.
Each column reports results of a regression of the dependent variable on an indicator of whether any household member either lost a job or
closed a business and a vector of covariates. In addition, all regressions control for locality × day of survey completion fixed effects (18,764),
as well as economic-sector fixed effects. Standard errors are clustered at the locality level (3,165). Panels B and C replicate the results using
country-date fixed effects (300) and standard errors clustered at the country level (17). See the Empirical Methods section in the main text
for more details.
Bottan, Hoffmann and Vera-Cossio 25

TA B L E S 5 Impacts of livelihood loss on income and food security by labor market characteristics

Panel A: Impacts on food security by country-level rates of self-employment

(1) (2) (3)

Decreased income Went hungry Eats less healthy

Lost job or closed business × % Self-Employed -0.006*** 0.002*** 0.000

(0.001) (0.000) (0.000)

Lost job or closed business 0.488*** 0.019 0.066***

(0.022) (0.020) (0.022)

Observations 186,521 198,726 174,452

Adjusted R-squared 0.437 0.545 0.350

Panel B: Impacts on food security by country-level rates of informality

(1) (2) (3)

Decreased income Went hungry Eats less healthy

Lost job or closed business × % Informal workers -0.003*** 0.002*** 0.001**

(0.000) (0.000) (0.000)

Lost job or closed business 0.421*** 0.017 0.038**

(0.013) (0.015) (0.016)

Observations 179,040 190,453 166,581

Adjusted R-squared 0.416 0.548 0.355

∗p < 0.1, ∗ ∗ p < 0.05, ∗ ∗ ∗p < 0.01

Panel A reports regression coefficients capturing the relationship between livelihood losses during the pandemic and outcomes, as a
function of the share of self-employed workers in each country. Each column reports results of a regression of the dependent variable on an
indicator of whether any household member either lost a job or closed a business, and an interaction term of job loss or business closure
with the share of self-employed workers in each country, and a vector of covariates. In addition, all regressions control for locality × day of
survey completion fixed effects (18,764), as well as economic-sector fixed effects. Standard errors are clustered at the locality level (3,165).
Panels B replicates the results using the percentage of informal workers as in each country instead of the share of self-employed workers.
See the Empirical Methods section in the main text for more details.
26 Bottan, Hoffmann and Vera-Cossio

References
[1] Stiglitz JE. Why It Matters. In: The priceof inequality: How today’s divided society endangers our future W.W. Norton
and Company; 2013.

[2] Milanovic B. Global Inequality in This Century and the Next. In: Global inequality: A new approach for the age of
globalization Harvard University Press; 2016.

[3] Piketty T, Yang L, Zucman G. Capital accumulation, private property, and rising inequality in China, 1978-2015. American
Economic Review 2019;109(7):2469–2496.

[4] Zucman G. Global wealth inequality; 2020.

[5] Smeeding T. Income, wealth, and debt and the Great Recession; 2012.

[6] Heathcote J, Perri F, Violante GI. Unequal we stand: An empirical analysis of economic inequality in the United States,
1967-2006. Review of Economics Dynamics 2010;13(1):15–51.

[7] Brzezinski M. Income inequality and the Great Recession in Central and Eastern Europe. Economic Systems
2018;42(2):219–247.

[8] Blackman A, Ibáñez AM, Izquierdo A, Keefer P, Moreira MM, Schady N, et al. La política publica frente al Covid-10; 2020,
iDB Policy Note.

[9] Fetzer T, Witte M, Hensel L, Jachimowicz JM, Haushofer J, Ivchenko A, et al. Global behaviors and perceptions at the
onset of the COVID-19 pandemic; 2020.

[10] Allcott H, Boxell L, Conway JC, Gentzkow M, Thaler M, Yang DY. Polarization and public health: Partisan differences in
social distancing during the coronavirus pandemic; 2020, nBER Working Paper No. 26946.

[11] Briscese G, Lacetera N, Macis M, Tonin M. Compliance with COVID-19 social-distancing measure in Italy: The role of
expectations and duration; 2020, nBER Working Paper No. 26916.

[12] Bento AI, Nguyen T, Wing C, Lozano-Rojas F, Ahn YY, Simon K. Evidence from internet search data shows information-
seeking responses to news of local COVID-19 cases. Proceedings of the National Academy of Sciences 2020;https:
//www.pnas.org/content/early/2020/05/01/2005335117.

[13] Adams-Prassl A, Boneva T, Golin M, Rauh C. Inequality in the impact of the coronavirus shock: Evidence from real time
surveys; 2020.

[14] Bartik AW, Bertrand M, Cullen ZB, Glaeser EL, Luca M, Stanton CT. How Are Small Businesses Adjusting to COVID-19?
Early Evidence from a Survey. National Bureau of Economic Research; 2020.

[15] Alon TM, Doepke M, Olmstead-Rumsey J, Tertilt M. The Impact of COVID-19 on Gender Equality. National Bureau of
Economic Research; 2020.

[16] Messina J, Silva J. Wage inequaity in Latin America. Understanding the past to prepare the future; 2018.

[17] Levy S, Schady N. Latin America’s social policy challenge: Education, social insurance, and redistribution. Journal of
Economic Perspectives 2013;27(2):193–218.

[18] Amarante V, Galván M, Mancero X. Inequality in Latin America: A global measurement; 2016.

[19] Báez JE, Fuchs A, Rodríguez-Castelán C. Shaking up economic progress. Aggregate shocks in Latin America and the
Caribbean; 2017.

[20] Bertrand M, Duflo E, Mullainathan S. How Much Should We Trust Differences-In-Differences Estimates?*. The Quarterly
Journal of Economics 2004 02;119(1):249–275.
Bottan, Hoffmann and Vera-Cossio 27

[21] Gertler P, Gruber J. Insuring Consumption Against Illness. American Economic Review 2002 March;92(1):51–70.

[22] Berg AG, Ostry JD. Inequality and Unsustainable growth: Two sides of the same coin?; 2011.

[23] Ostry J, Berg A, Tsangarides CG. Redistribution, inequality, and growth; 2014.

[24] Cingano F. Trends in income inequality and its impact on economic growth; 2014.