The Impact of COVID-19 on Small Businesses in the United States, Thesis of History of film

Download The Impact of COVID-19 on Small Businesses in the United States and more Thesis History of film in PDF only on Docsity!

THE EFFECTS OF COVID-19 ON U.S. SMALL BUSINESSES:

EVIDENCE FROM OWNERS, MANAGERS, AND EMPLOYEES *

GEORGIJ ALEKSEEV†^ SAFAA AMER‡^ MANASA GOPAL§^ THERESA KUCHLER¶

JW SCHNEIDER||^ JOHANNES STROEBEL**^ NILS WERNERFELT††

November 2021

Abstract

We analyze a large-scale survey of small business owners, managers, and employees in the United States to understand the effects of the COVID-19 pandemic on those businesses. We explore two waves of the survey that were fielded on Facebook in April 2020 and December

2020. We document five facts about the impact of the pandemic on small businesses. (1) Larger firms, older firms, and male-owned firms were more likely to remain open during the early stages of the pandemic, with many of these heterogeneities persisting through the end of 2020. (2) At businesses that remained open, concerns about demand shocks outweighed concerns about supply shocks, though the relative importance of supply shocks grew over time. (3) In response to the pandemic, almost a quarter of the firms reduced their prices, with price reductions concentrated among businesses facing financial constraints and demand shocks; almost no firms raised prices. (4) Only a quarter of small businesses had access to formal sources of financing at the start of the pandemic, and access to formal financing affected how firms responded to the pandemic. (5) Increased household responsibilities affected the ability of managers and employees to focus on their work, while increased business responsibilities impacted their ability to take care of their household members. This effect persisted through December 2020 and was particularly strong for women and parents of school-aged children. We discuss how these facts inform our understanding of the economic effects of the COVID- pandemic and how they can help design policy responses to similar shocks.

JEL: E3, L26, M

Keywords: Small Businesses, COVID-19, Working from Home, Small Business Finance, Prod- uct Pricing

*We thank the Editor Tomasz Piskorski, Associate Editor, and three anonymous referees, as well as numerous sem- inar and conference participants, for helpful comments and feedback. Amer, Schneider, and Wernerfelt are employees at Facebook. Kuchler and Stroebel have a research consulting relationship with Facebook. We thank the NBIM for funding through a grant to the Volatility and Risk Institute at NYU Stern. †NYU Stern. Email: [email protected]. ‡Facebook. Email: [email protected]. §Scheller College of Business, Georgia Tech. Email: [email protected]. ¶NYU Stern. NBER, CEPR. Email: [email protected]. ||Facebook. Email: [email protected]. **NYU Stern. NBER, CEPR. Email: [email protected]. ††Facebook. Email: [email protected].

I INTRODUCTION

The COVID-19 pandemic and associated public health interventions have led to large changes in

business and work environments. Government orders forced many in-person establishments to

close suddenly, infection risks led to accelerated adoption of modern sales and communication

technologies, and school closures placed increased childcare duties on many working parents.

These forces have put particular pressures on the owners, managers, and employees of small busi-

nesses, where central responsibilities are often shared between only a few individuals. However,

while these small businesses are a key source of employment in the U.S. and an explicit focus of

many policy efforts, they are often underrepresented in traditional data sources (see Buffington,

Dennis, Dinlersoz, Foster, and Klimek, 2020). In addition, to the extent that official data is avail-

able at all, it is often only released after a considerable time lag, reducing its usefulness in guiding

policy responses in real time. This presents important challenges for researchers and policymakers

hoping to understand and respond to the aggregate and distributional effects of large economic

shocks such as COVID-19 on the performance of small businesses.

In this paper, we report findings from a large and comprehensive survey of owners, man-

agers, and employees of U.S. small businesses during the COVID-19 pandemic, and extract in-

sights into the effects of COVID-19 on those businesses. A first wave of the survey was conducted

in late April 2020, followed by a second, smaller survey wave in early December 2020. Having

information on firms from both points in time puts us in a unique position to compare the im-

mediate and medium-run effects of the pandemic. The survey waves were fielded through the

Facebook platform, targeting Facebook business page administrators, frequent sellers on its e-

commerce platform Marketplace, and the general user population. This sampling frame allowed

us to reach owners, managers, and employees of small businesses, as well as self-employed indi-

viduals, providing us with a unique ability to compare and contrast the effects of the pandemic

across numerous small business stakeholders.

The April survey, which is the main focus of this paper, contained 136 questions and obtained

complete responses from 28,188 businesses and 9,720 employees, making it one of the largest

undertakings to describe the early effects of the COVID-19 crisis on small businesses. The smaller

December wave obtained complete responses from 5,718 businesses. While no survey sample is

fully representative, due to selection both in the sample that can be reached as well as in who

responds, we nevertheless find that our respondents broadly match the characteristics of small

businesses in the U.S. as described by the Census Bureau (see Appendix Tables A.2 and A.3).

The survey first asked respondents if they are an employee or whether they own, manage,

or operate a business; owners/managers were then asked whether their business was currently

operational. These initial questions were used to direct respondents to different sections of the

survey, each focusing on a different aspect of COVID-19’s effects on businesses. Respondents also

answered a core set of demographic and business questions. Full details on the survey methodol-

ogy are provided in Section II.

From our rich data, we distill five facts on the effects of COVID-19 on U.S. small businesses,

and discuss how these facts inform our understanding of the economic effects of the pandemic, as

well as the operation of small businesses more broadly. These facts are:

1. Larger firms, older firms, male-owned firms, firms that relied less on in-person interactions,

and firms that advertised on Facebook were more likely to remain open during the early

stages of the pandemic. Older firms were also more likely to be open as of December 2020,

suggesting a large and persistent impact of the crisis on young small businesses.

2. Firms that remained open in April 2020 expected a variety of challenges over the subsequent

six months. Concerns related to demand shocks outweighed concerns related to supply

shocks, in particular for smaller firms and firms in regions with greater lockdown restric-

tions. Over time, the relative importance of concerns about supply shocks increased.

3. In response to the pandemic, almost a quarter of firms reduced prices, while only about 4%

of firms increased prices. Firms facing demand shocks and financial constraints were more

likely to reduce prices, while firms facing supply shocks were disproportionately likely to

increase prices.

4. Only around a quarter of small businesses had access to formal loans from financial institu-

tions when entering the pandemic, with many businesses relying largely on informal sources

of financing. Older and larger firms were more likely to have access to formal sources of fi-

nancing, and firms managed by men had more access to informal sources of credit.

5. Household responsibilities such as childcare led to large incremental burdens for small busi-

ness owners and employees. Similarly, increased business responsibilities substantially im-

pacted individuals’ abilities to focus on their household obligations during the pandemic.

These negative effects largely persisted through the end of 2020, particularly for women-led

businesses. A sizable share of employees reported household responsibilities led to un-

employment, with these responsibilities being the greatest burden for female employees at

mid-sized firms. Female employees were also disproportionately more likely to drop out of

the labor force during the pandemic.

Contributions to the Literature. Our findings expand upon a growing body of research study-

ing the economic implications of the COVID-19 pandemic, including several contemporaneous

efforts to use surveys to better understand the performance of U.S. small businesses during this

period.^1 For example, Bartik, Bertrand, Cullen, Glaeser, Luca, and Stanton (2020), Fairlie (2020),

and Humphries, Neilson, and Ulyssea (2020a) document business closures and mass layoffs early

in the pandemic (see also Bartlett, 2020; Campello, Kankanhalli, and Muthukrishnan, 2020). In ad-

dition to exploring the pandemic’s effect on general business performance, our paper focuses on

understanding the specific shocks hitting small businesses, the effects on their price-setting behav-

ior, and the importance of changes in the interaction of household and work responsibilities. Our

(^1) Other research papers in the large emerging literature studying the economic effects of COVID-19 include Cox,

Ganong, Noel, Vavra, Wong, Farrell, and Greig (2020); Giglio, Maggiori, Stroebel, and Utkus (2020); Howell, Kuch- ler, Snitkof, Stroebel, and Wong (2021); Chetty, Friedman, Hendren, Stepner, and Team (2020), and Coibion, Gorod- nichenko, and Weber (2020a,c).

survey is also unique in providing the perspectives of both small business owners/managers and

their employees. Our large sample size allows us to document important heterogeneities across af-

fected firms, with a particular focus on exploring differences across firm size, age, industry, ability

to operate remotely, the types of shocks they face, and local COVID case intensity, and lockdown

policies. We also study differential impact across genders of both owners and employees. Finally,

the availability of data from the December 2020 wave of the survey puts us in a unique position to

explore the medium-rum persistence of some of the observed patterns. For each of our five facts,

we will discuss how our findings relate to and expand upon the contemporaneous literature.

II SURVEY DETAILS

The survey was conducted alongside Facebook’s ongoing data collection efforts with the World

Bank and OECD on the Future of Business, and in partnership with the Small Business Roundtable.

In this section, we describe the structure of the survey and the sampling methodology.

Sampling and Screening – Wave 1. The first survey wave was fielded on the Facebook platform

between April 20 and April 28, 2020, to a stratified probability-based random sample of U.S. Face-

book users.^2 Every monthly active U.S. Facebook account was eligible for the survey,^3 though we

oversampled accounts of Facebook business page administrators and active sellers on Facebook

Marketplace.^4 This sampling frame led to a high chance of identifying individuals who own or

manage small businesses, many of which have a Facebook presence. Respondents from the gen-

eral Facebook population were more likely to be employees, both at small and large enterprises.

Sampled users received an invitation to participate in an online survey at the top of their

News Feed. This invitation was shown for three successive logins. After accepting the invitation,

users were shown an introductory text and screening questions to understand their possible roles

within the business (see Appendix Figure A.1 for more details). The introductory text described

that participation in the survey was voluntary and uncompensated, that responses would be kept

confidential, and that aggregated results from the survey might be shared publicly.

(^2) Facebook was created in 2004 and, by June 2020, had 2.7 billion active users around the world and 256 million

active users in the U.S. and Canada. An independent survey of Facebook users from 2019 found that more than 69% of the U.S. adult population used Facebook (Perrin and Anderson, 2019). That same survey shows that Facebook us- age rates among U.S.-based online adults were relatively constant across income groups, education levels, and race, and among urban, rural, and suburban residents; usage rates were slightly declining in age (from 79% of individuals aged 18 to 29, to 46% of individuals aged 65 and older). See Allen, Peng, and Shan (2020); Bailey, Cao, Kuchler, and Stroebel (2018); Bailey, Cao, Kuchler, Stroebel, and Wong (2018); Bailey, Dávila, Kuchler, and Stroebel (2019); Bailey, Farrell, Kuchler, and Stroebel (2020); Bailey, Gupta, Hillenbrand, Kuchler, Richmond, and Stroebel (2020); Bailey, John- ston, Kuchler, Russel, State, and Stroebel (2020); Bailey, Johnston, Koenen, Kuchler, Russel, and Stroebel (2020); Bali, Hirshleifer, Peng, and Tang (2018); Kuchler, Russel, and Stroebel (2020); Kuchler, Peng, Stroebel, Li, and Zhou (2020); Wilson (2019) and Rehbein and Rother (2020) for other economics and finance research using data from Facebook. (^3) Facebook generally does not allow accounts to receive multiple surveys in a short span of time. Since some of

these surveys followed different sampling regimes (e.g., simple random or potentially targeted sampling), the total pool for our survey was not drawn completely at random from the overall Facebook population. In practice, reweight- ing for sampling (and non-response) moves the point estimates minimally, and the observable characteristics of our respondents align well with those from nationwide, offline estimates (see Appendix Tables A.2 and A.3). (^4) Facebook pages are profiles on Facebook specifically for businesses, brands, communities, or public figures. Each

page must have an account tied to it as an administrator, and we oversampled those that were from business pages. A business page is required for small businesses to advertise on Facebook, and maintaining a business page is free of charge. Facebook Marketplace is an e-commerce platform where users can buy and sell different products.

The survey invitation was sent to about 1.9 million Facebook users, and 66,297 eligible in-

dividuals completed at least part of the survey. Our baseline sample consists of the 37,908 re-

spondents (28,188 from business owners/managers and 9,720 from employees) who completed

the entire survey, though we have verified that all facts are consistent within the broader sample

of respondents that completed the survey partially. There was no screening on firm size, though

the sampling frame ensured that most respondents were associated with small businesses. About

68% of owner/manager respondents were associated with businesses with fewer than 10 employ-

ees, and 93% were associated with businesses with fewer than 500 employees. On the other hand,

about 25% of the employee sample worked at businesses with more than 500 employees. We ver-

ify that our findings are robust to excluding responses pertaining to businesses with more than

500 employees, the threshold to be eligible for a Small Business Administration loan.

Survey Instrument – Wave 1. The survey questionnaire was designed with a complex flow to re-

duce the burden on respondents while addressing a wide range of important social and economic

issues (see Appendix Figure A.2). A respondent could skip any question in the survey with no

prompts to answer, and the questionnaire flow would take them to the next logical question. The

survey started with screening questions, followed by topic modules and a core set of questions

addressed to all respondents. Modules were assigned to respondents in a semi-random fashion

based on their business role and whether their business was operational. First, respondents were

asked preliminary questions to classify their employment status, role in the firm, and operational

status of the firm. Based on this, respondents were classified into: business owners and managers

(including operators of personal businesses) and employees (including both employed or recently

unemployed individuals).^5 Individuals not fitting any of these categories were not asked further

questions. Next, business owners and managers of firms that were operational, were randomly

assigned to one of five thematic modules. Employees (employed or unemployed) were surveyed

about the businesses they worked for and the impact on the interaction of their work and house-

hold responsibilities. Finally, all respondents answered questions about their demographics and

firm characteristics.

Sampling, Screening, and Survey Instrument – Wave 2. Between November 20, 2020, and Decem-

ber 20, 2020, Facebook fielded a second wave of the survey using a similar sampling procedure as

the first wave. Many of the questions included in this second wave were similar to those of the

first wave, while other questions were dropped and new questions were introduced. The survey

sampling aimed to collect a smaller number of responses than the first wave, and we observe com-

plete responses from about 5,718 small business owners and managers. While this second wave

of the survey is not the primary focus of our analysis, we use it to compare survey responses from

April 2020 to those from December 2020, allowing us to explore the persistence of the patterns

under investigation.

(^5) “Personal” businesses were defined as respondents who reported that they were “Self-employed providing goods

or services” or that they “Produce goods sold for personal income” but did not otherwise self-identify as an owner or manager of a business. While there is no standard term for this category of businesses, they overlap a great deal with what are commonly called sole-proprietor or micro businesses.

III FACTS ABOUT THE EFFECTS OF COVID-19 ON SMALL BUSINESSES

In this section, we present five new facts about the effects of the COVID-19 pandemic on U.S. small

businesses and their employees. First, we document how business operations have been impacted

by studying the determinants of a business’s decision to cease operations during the pandemic.

For businesses that continued to operate during the pandemic, we discuss the types of shocks

these businesses faced, their financial conditions and access to external and internal sources of

finance, and their product pricing decisions. We also study the interaction between business and

household responsibilities of both small business owners and employees, a particularly important

dimension of the effects of the COVID-19 pandemic.

III.A Business Closures.

We first analyze the determinants of a business’s decision to shut down or remain operational

during the pandemic, and how these changed between the April 2020 and December 2020 waves

of the survey. Columns 1-3 of Table 2 focus on whether a business remained open at the end of

April 2020, providing us with information on the immediate short-term effects of the COVID-

crisis on small business openings. Different rows highlight heterogeneity across different firm

characteristics. Columns 4-6 explore whether a business was open in December 2020, a time when

many pandemic-related restrictions across the U.S. had been lifted (USA Today, 2020). This anal-

ysis allows us to assess the medium-run effects of the crisis on firm operations. Columns 1 and 4

present sample means for the full sample as well as across different firm characteristics. Columns

2 and 5 provide corresponding multivariate regression results when all firm characteristics are

included, while columns 3 and 6 present the associated heteroskedasticity robust standard errors.

We continue using heteroskedasticity robust standard errors in all regressions.

We start by looking at the April 2020 wave before comparing the results to the December

2020 wave. Only about 65% of small businesses in our sample remained open at the end of April

2020, a number that is comparable to the 43% of businesses that reported being closed in late

March 2020 in the survey studied by Bartik, Bertrand, Cullen, Glaeser, Luca, and Stanton (2020).

Older businesses, larger businesses, and businesses owned by men were more likely to remain

open during the early months of the pandemic. We show later in this paper that younger firms

faced greater demand shocks and women faced increased household responsibilities during the

COVID-19 pandemic, which could explain the differences in closure rates of businesses.

Furthermore, businesses that required more in-person interactions to operate and those op-

erating in regions with a higher COVID case intensity were less likely to be open in April 2020,

potentially due to government restrictions. However, we found no differential effect by the aver-

age decline in peoples’ mobility, which provides a proxy for the overall extent of and compliance

with social distancing restrictions. Businesses in the hospitality and service sectors had the largest

probability of closing by April 2020, while firms in the information and communications tech-

nology (ICT) sector were most likely to remain open. Still, 20% of ICT businesses had closed by

April 2020. We also see that firms that advertised via Facebook before the pandemic were more

likely to continue operating in the pandemic. This finding is consistent with the interpretation

that firms with more experience of online advertising had a comparative advantage in operating

during periods of sustained closures of physical businesses.^6

The multivariate regression reported in Column 2 suggests that most of these relationships

persist when controlling for the other observable firm characteristics. The exception is that when

controlling for both firm age and firm sales, which are highly correlated, differences in age largely

disappear, and it is differences in firm size that are associated with large differences in a business’s

propensity to remain open in April 2020.

By December 2020, almost nine months after the start of the pandemic, many of the busi-

nesses that were closed in April 2020 had reopened. However, nearly 18% of small businesses in

our sample, corresponding to about half of the businesses shut in April 2020, remained closed.

These results are quite surprising given earlier surveys on reopening expectations among closed

businesses: in the April 2020 wave of our survey, 67.9% of closed businesses reported that they

planned to reopen in the future, while 27.3% were unsure, and only 4.9% reported that they did

not plan to reopen. Similarly, Balla-Elliott, Cullen, Glaeser, Luca, and Stanton (2020) show that

in May 2020, most closed businesses reported plans to reopen within days of restrictions being

lifted, and only 18% of closed businesses expected they would delay their opening at least one

month after restrictions ended. In contrast to these expectations, we see a sizable share of closed

businesses as of December, despite there being few remaining restrictions on operations in most

of the United States.

By December 2020, we find no remaining correlation between either cumulative cases or mo-

bility declines and business openings. Similarly, the differences in the probability of being open

between firms in different industries had declined substantially, in large part due to a substantial

reopening of businesses in the hospitality and services sectors. However, younger businesses re-

mained substantially more likely to be shut, both unconditionally as well as conditional on other

firm characteristics, suggesting a persistent impact of the pandemic on the youngest small busi-

nesses. Note that we do not observe all covariates from the April 2020 wave in the December 2020

wave. Among others, we do not have information on the business size. Since business age and

size are highly correlated, the higher share of closed businesses among younger firms could reflect

the larger impact of the pandemic on smaller businesses.

III.B Supply or Demand Shocks

An important input to designing policy responses to an economic crisis is a good understanding of

whether the economy is largely facing a supply shock or a demand shock. A supply shock reduces

the economy’s ability to produce goods and services at a given price. On the other hand, a demand

shock corresponds to a reduction in consumers’ willingness to purchase these goods and services

at a given price. In particular, conventional fiscal policy tools, such as stimulus checks, can be

effective to counteract the effects of negative demand shocks, while these policies will be less

effective in the presence of supply shocks and may even lead to inflationary pressures. Similarly,

while supply and demand shocks have the same effect on quantities, they have opposing effects

on prices and, therefore, invite different monetary policy responses.

(^6) However, the observed patterns might also be explained by other unobservable characteristics. For example, a

larger online presence could be associated with a more flexible and modern business model.

During the COVID-19 pandemic, businesses were plausibly affected by both supply shocks

due to supply chain disruptions and demand shocks due to households’ unwillingness or inability

to purchase certain goods and services. An emerging literature has started to discuss the relative

importance and possible interaction between these types of shocks (see, for example, Guerrieri,

Lorenzoni, Straub, and Werning, 2020; Baqaee and Farhi, 2020, 2021). Our survey can advance

this literature by shedding light on whether small businesses perceived the challenges of COVID-

19 to be primarily a supply shock or a demand shock, how these perceptions changed over time,

and how they differed across industries and other firm characteristics.

In particular, in both waves of the survey, operational businesses were asked “What do you

expect this business’s biggest challenge to be in the next few months?”. We take the responses to this

question and classify them to either correspond to a demand shock or a supply shock, and then

study the distribution of these shocks across firm characteristics.^7

Table 3 Panel A shows that, in April 2020, 54.5% of businesses were primarily facing a de-

mand shock, and 30.0% of businesses were facing a supply shock. The remainder either reported

not facing any challenges or an “other” challenge that we could naturally not classify. This is con-

sistent with the findings of Balleer, Link, Menkhoff, and Zorn (2020), who studied the price-setting

behavior of a sample of (relatively large) German firms to infer that demand shocks dominated

in the German context. Larger businesses were disproportionately likely to be concerned about

supply shocks, while at smaller businesses, concerns about demand shocks dominated. Firms

with access to formal (or both formal and informal) financing were more likely to report facing

demand shocks. All else equal, businesses in areas with more COVID-19 cases and larger mobility

declines were more likely to be concerned about demand shocks, consistent with households in

those regions being less willing to consume goods or services that required in-person interaction.

Supply shocks were disproportionately felt by firms in the retail/wholesale sector, while demand

shocks dominated for firms in the ICT sector.

In December 2020, the share of businesses that were primarily worried about demand shocks

had declined to 45.7%, while the share of firms facing a supply shock had increased somewhat to

33.0% (Table 3, Panel B ). In the multivariate regressions for December, businesses operated by men

were less likely to report facing demand shocks and more likely to report facing a supply shock,

after conditioning for industry and location. Younger firms continued reporting higher levels of

demand shocks. These differences could explain the prolonged closure of younger businesses in

the pandemic, as documented in the previous section. The relative importance of supply and

demand shocks across different industries were broadly the same as in April 2020, though the

relative importance of supply shocks had increased disproportionately in the construction sector.

(^7) We classified the business as primarily facing a demand shock if they responded that their biggest struggle would

be “Lack of demand”, “Cashflow”, or “Repaying Loans”. We classified the businesses as primarily facing a supply shock if they responded to be primarily struggling due to “Inventory”, “Logistics (e.g., shipping, delivering services or goods)”, “Finding supplies”, “Lack of staff”, or “Government/Health Authority Orders”. These classifications are necessarily imperfect and involve a degree of judgment. In particular, “Government/Health Authority Orders” could also be considered as a demand shock, and there are interpretations of “Repaying Loans” that might correspond more closely to a supply shock. We verify that none of our conclusions are sensitive to how we classify these two responses. In addition, we cross-validate their benchmark classification against their correlation with price-setting responses; see Section III.C. For example, we verify that firms reporting “Government/Health Authority Orders” as the primary challenge were disproportionately likely to raise prices (rather than reduce prices).

The types of shocks that businesses face are closely tied to the optimal policy responses desired

by these firms. In Appendix Table A.1, we explore responses to a question in the April 2020

wave that asked businesses which policies would be most useful in supporting them through the

pandemic. Businesses faced with a demand shock were more likely to ask for support through

wage/utility subsidies, loan guarantees or deferrals, and tax/rent deferrals. Businesses facing

supply shocks, on the other hand, were more likely to request support for taking care of their

household.

III.C Price-Setting Response

A key mechanism used by firms to respond to economic shocks is to adjust their product pric-

ing. Price changes can reflect shifts in demand and supply and can provide us with additional

information on their relative importance (a supply shock would generally lead to an increase in

prices, while a reduction in demand would motivate price reductions). In addition, these pric-

ing responses are a direct determinant of aggregate changes in the price level and are, therefore,

important to understand, in particular for monetary policymakers.

In the April 2020 wave of the survey, firms were asked whether they had reduced or increased

their prices in response to the pandemic. Table 4 shows that only 3.9% of firms had increased

prices on their products or services, while 24.2% had reduced prices, consistent with the relative

dominance of demand shocks documented in the prior section. Businesses that reported primarily

facing demand shocks were more likely to reduce prices, while businesses reporting to face supply

shocks were more likely to increase prices. On average, hotels and restaurants were more likely to

increase prices, while businesses in the transportation and logistic sector were more likely to re-

duce prices; these relationships are true both unconditionally and conditional on other observable

firm characteristics. Older firms were the least likely to reduce their prices. This is consistent with

older firms reporting lower levels of demand shocks.

Table 4 also shows how firms’ decisions to adjust prices varied with their cashflow situations.

In April 2020, cash outflows were higher than cash inflows for 41.8% of businesses. Outflows

equaled inflows for 19.8% of businesses and inflows exceeded outflows for 23.2% of businesses.

The remaining businesses reported that they did not know if their inflows exceeded their outflows.

Of the surveyed firms, 78.2% reported being at least somewhat concerned about their cashflow

situation over the next 3 months. Table 4 highlights that businesses facing cashflow concerns

(either reporting a high concern for future cashflows or reporting current outflows greater than

inflows) were more likely to reduce their product prices. Furthermore, businesses struggling with

making their payments also reduced their prices.^8

These results on the relationship between financial constraints and price changes directly

speak to an ongoing debate in the academic literature. In particular, our findings are in line

with evidence in Kim (2020), who shows that even outside a pandemic, businesses facing a credit

crunch often temporarily decrease their prices. In contrast, Gilchrist, Schoenle, Sim, and Zakrajšek

(^8) Our direct measure of demand shocks is highly correlated with the presence of cashflow concerns and firms

struggeling to make payments, explaining why our direct measure of demand shock does not survive the multivariate regression. However, the results are consistent with various proxies for a drop in demand leading to price reductions. Also, demand shocks significantly decrease the probability of increasing prices.

(2017) conclude that financial constraints lead to an increase in product prices. This difference in

results is potentially driven by differences in the sample of firms: Gilchrist, Schoenle, Sim, and

Zakrajšek (2017) focus on firms in the S&P500 while our sample focuses on small businesses.

III.D Access to External Finance and Applications for Aid

We next explore the distribution of financing sources across small businesses and analyze how

access to formal and informal sources of financing affected firm behavior during the pandemic.

Such information on the financing of small private businesses in the U.S. has been largely un-

available.^9 The findings from our survey can thus contribute to a better understanding of small

business finances even beyond the immediate setting of the pandemic.

We start by exploring the funding sources of small businesses before the pandemic. In partic-

ular, we asked businesses whether they had access to a loan or line of credit from a bank or other

financial institution. Column 1 of Table 5 shows that only 26.5% of businesses had access to such

formal sources of credit.^10 Older firms, larger firms, and firms in the capital-intensive manufac-

turing, construction, and agricultural sectors were more likely to have access to formal financing

before the pandemic, both unconditionally and conditional on other firm characteristics.^11

In Columns 4-6 of Table 5, we explore access to informal sources of funding for small busi-

nesses, which include personal savings, funds from family and friends, retirement funds, unem-

ployment benefits, and community donations. About 63.1% of businesses had access to at least

some source of informal financing. Conditional on other firm characteristics, businesses in con-

struction, manufacturing, and transportation industries were least likely to have access to informal

financing. The largest firms (by sales) were also less likely to rely on informal sources of funding.

These findings confirm conventional wisdom suggesting that smaller and younger firms largely

rely on funds from owners, and that financing sources vary with firm size and age, with larger

and older firms shifting to formal debt (Berger and Udell, 1998). Our survey allows researchers

to quantify such relationships and uncover novel insights on how access to internal and external

financing varies across other firm characteristics.

We also explore how businesses tackled their financing needs during the pandemic by asking

them whether they applied for any government-provided loans or grants, or for a new bank loan.

(^9) The best existing sources are the Survey of Small Business Finance (SSBF) and the Fed Small Business Credit

Survey. However, the SSBF has not been conducted since 2003, and the financing sources of small businesses have significantly changed since that time, with a decline in bank lending and a rise in nonbank sources of financing such as finance companies and FinTech lenders (Gopal and Schnabl, 2020). The Fed Small Business Credit Survey has been conducted annually since 2016. However, the SBCS only shows how outcome variables vary with a limited number of demographics. On the other hand, our results include many additional firmographic splits and allow us to conduct multivariate regressions. Beyond that, our survey elicits responses from a much larger sample and provides detailed information on access to financing and responses of non-employer firms. (^10) The 2018 Small Business Credit Survey (SBCS) shows that 28% of non-employer firms and 55% of employer firms

had a loan or line of credit outstanding. These numbers are larger than the 26.5% of businesses that have a loan outstanding in our sample. We think these differences may be driven by the fact that the set of firms in our sample are significantly smaller (with 68% of businesses in our sample having fewer than 10 employees). The industry composition across the samples is also different, with a much larger share of retail firms (who are less likely to have access to formal financing) in our sample than in the SBCS sample. Nevertheless, the SBCS also shows that only 14% of small employer businesses use external sources as primary financing—a number that better aligns with our results. (^11) These findings align with the 2018 SBCS, which confirms that the share of firms relying on external financing

increases in firm size and is particularly high for capital-intensive industries. However, the released data does not allow to draw conclusions in a multivariate setting.

By the end of April 2020, about 42.1% of firms had applied for a government grant, while 18.6%

of firms had applied for a bank loan. We find broadly similar heterogeneities across firm charac-

teristics in the firms’ propensities to apply for government loans/grants and bank loans. Hotels,

restaurants, and cafes were, by far, the most common applicants. Older firms, larger firms, firms

that relied more on in-person interactions, and firms facing demand shocks were also more likely

to make these applications. The observed differences are large: 52.9% of firms with sales above

$1m applied to government funding, whereas only 22.3% of firms with sales below $50k did.

Our results in Table 5 also suggest that demand for new external financing, either through

bank loans or government loans/grants, was concentrated among firms that already had pre-

existing bank loans. This is consistent with evidence that larger businesses had more informa-

tion and were more likely to apply for government programs, and that banks were prioritizing

customers with pre-existing relationships, even for government programs such as the Paycheck

Protection Program (see Granja, Makridis, Yannelis, and Zwick, 2020; Howell, Kuchler, Snitkof,

Stroebel, and Wong, 2021; Humphries, Neilson, and Ulyssea, 2020b; Joaquim and Netto, 2020).^12

III.E Business and Household Interactions

One central distinguishing feature between small businesses and larger corporations is that the

operation of small businesses is often concentrated among relatively few individuals, often from

within the same family unit. As a result, shocks to the households’ domestic situations—such

as the loss of a childcare provider—can have potentially large effects on the operation of SMEs.

Similarly, the business’s economic situation can have large effects on the owners’ and managers’

ability to focus on their household responsibilities. These interactions between business and work

responsibilities were particularly important during the COVID-19 crisis, which featured dramati-

cally increased household burdens for many individuals due to a combination of school closures,

requirements to quarantine, and the unavailability of hired domestic help. Our survey is unique

in its ability to quantify this important dimension of the operation of small businesses, which has

so far not received much attention in the literature, either before or during the pandemic.

Owner and Manager Responses. We first explore how the pandemic affected the interaction

of business and household responsibilities for owners and managers of small businesses, both in

April 2020 (Table 6, Panel A) and December 2020 (Table 6, Panel B).

In both panels of of Table 6, the dependent variable in columns 1-3 takes a value of one if the

owner/manager reported that their household responsibilities were affecting their businesses “a

lot” or “a great deal.” In April 2020, 31.9% of respondents stated that their household respon-

sibilities substantially affected their ability to focus on work during the crisis. These numbers

declined only slightly, to 28.7%, by December 2020, highlighting the persistent nature of the effect

of increasing household responsibilities on small businesses during the COVID-19 pandemic.

In April 2020, managers responsible for caring for children or dependent adults, managers

who spent more time on household work, and managers who were struggling with household

(^12) One important, alternate form of financial support provided to small businesses in the pandemic was through

debt forbearance. While our survey did not explicitly ask owners about forbearance, Cherry, Jiang, Matvos, Piskorski, and Seru (2021) find that small business owners took mortgage forbearance at a higher rate compared to non-business homeowners.

expenses were the most likely to report an adverse effect of household responsibilities on the

business. In the unconditional comparisons, female managers were more likely to report an im-

pact of household responsibilities on their ability to focus on their business. These relationships

disappear when conditioning on the extent of household-related duties, consistent with women

being disproportionately affected by such duties. By December 2020, the unconditional gap be-

tween men and women in the extent to which household responsibilities affected their abilities to

focus on their businesses had increased from 6.3 percentage points to 10.2 percentage points.

In columns 4-6 of Table 6, the dependent variable takes a value of one if the manager reported

that their businesses responsibilities were affecting their households “a lot” or “a great deal.” On

average, 29.6% of the respondents in April 2020 and 24.7% of respondents in December 2020 re-

ported a substantial adverse impact of their business responsibilities on their household. Larger

businesses in April and older businesses in December were more likely to report their work was af-

fecting their household. Moreover, childcare duties and difficulties in paying household expenses

had an adverse impact on the manager’s household. These effects persisted through December

2020. However, while in April 2020, men and women reported that their job affected their house-

hold responsibilities by roughly equal amounts, by December 2020, men were less likely than

women to continue reporting a large negative impact of their business on their household respon-

sibilities. This reinforces our earlier finding that the gender gap in the effect of the pandemic on

the interaction of work and household responsibilities increased as the pandemic progressed.

Employee Responses. While the previous section focused on the perspectives of business own-

ers and managers, we now analyze the relationships between business and household responsi-

bilities for employees. The results presented in Table 7 are based on all employee respondents in

our survey, including the approximately 25% of employees working for firms with more than 500

employees. We find that excluding employees working for firms with more than 500 employees

does not alter our conclusions, suggesting similar effects for employees of large and small firms.

Employed individuals were asked questions similar to those asked of business owners and

managers. Panel A of Table 7 shows that 24% of employees reported that their household re-

sponsibilities had affected their ability to focus on work “a lot” or “a great deal”. This is particu-

larly true for female employees, employees aged 25-45 years, employees that spend more time on

household activities, and employees in regions with high a COVID prevalence. Similarly, younger

employees, female employees, and employees working at larger firms report that their work re-

sponsibilities had affected their ability to take care of their household “a lot” or “a great deal.”

In Panel B of Table 7, we look at unemployed workers. About 6% of employees report that

their primary reason for unemployment was that they need to take care of children or dependent

adults. Once again, women and respondents who had to spend a lot of time on household activ-

ities were the most affected. Less than a quarter of our unemployed respondents were actively

looking for a job (Column 4-6), consistent with Coibion, Gorodnichenko, and Weber (2020b), who

show that many who lost jobs during the pandemic were not actively looking for a new one. Older

respondents and women were less likely to be looking for a new job. These results are consistent

with a significant drop in labor force participation in the pandemic, particularly for women.

Take-Aways. Overall, these responses highlight that shocks to household responsibilities, such

as those induced by COVID-19, have profound effects on individuals’ abilities to focus on their

work responsibilities. Since, at small businesses, many of these responsibilities are split between

only a small number of individuals, shocks to the household responsibilities of their owners and

managers can constitute a critical threat to those small businesses. This suggests that policies

aimed at increasing households’ access to support with household tasks—such as the universal

access to reliable child care—have the potential to protect SMEs from the destabilizing effects of

household-level shocks to their owners, managers, and employees, even in a post-COVID world.

IV CONCLUSION

This paper presents a large-scale survey of the impact of the COVID-19 pandemic on small busi-

ness owners, managers, and employees. We study the short and medium-term impact of the

pandemic on U.S. businesses through two waves of the survey, conducted through Facebook in

April and December 2020. We present five new facts on the impact of COVID-19 on small business

owners and their employees. These facts inform our understanding of the economic effects of the

COVID-19 pandemic and help guide policy responses to similar shocks. The patterns documented

in our paper also expand our understanding of the financing and operations of small businesses

more broadly, and highlight the power of using large-scale surveys to overcome the challenges

from the lack of representative administrative data on small businesses.

REFERENCES

Allen, Linda, Lin Peng, and Yu Shan, 2020. Social Networks and Credit Allocation on FinTech Lending

Platforms. Available at SSRN.

Bailey, Michael, Rachel Cao, Theresa Kuchler, and Johannes Stroebel, 2018. The Economic Effects of Social

Networks: Evidence from the Housing Market. Journal of Political Economy 126, 2224–2276.

, and Arlene Wong, 2018. Social Connectedness: Measurements, Determinants, and Effects. Journal of Economic Perspectives 32, 259–80.

Bailey, Michael, Eduardo Dávila, Theresa Kuchler, and Johannes Stroebel, 2019. House Price Beliefs and

Mortgage Leverage Choice. The Review of Economic Studies 86, 2403–2452.

Bailey, Michael, Patrick Farrell, Theresa Kuchler, and Johannes Stroebel, 2020. Social Connectedness in

Urban Areas. Journal of Urban Economics p. 103264.

Bailey, Michael, Abhinav Gupta, Sebastian Hillenbrand, Theresa Kuchler, Robert Richmond, and Johannes

Stroebel, 2020. International Trade and Social Connectedness. Working Paper 26960 National Bureau of Economic Research.

Bailey, Michael, Drew Johnston, Martin Koenen, Theresa Kuchler, Dominic Russel, and Johannes Stroebel,

2020. Social Distancing During a Pandemic: The Role of Friends. NBER Working Paper.

Bailey, Michael, Drew Johnston, Theresa Kuchler, Dominic Russel, Bogdan State, and Johannes Stroebel,

2020. The Determinants of Social Connectedness in Europe. Social Informatics.

Bali, Turan G, David A Hirshleifer, Lin Peng, and Yi Tang, 2018. Attention, Social Interaction, and Investor

Attraction to Lottery Stocks. 9th Miami Behavioral Finance Conference.

Balla-Elliott, Dylan, Zoe B Cullen, Edward L Glaeser, Michael Luca, and Christopher T Stanton, 2020. Busi-

ness Reopening Decisions and Demand Forecasts During the COVID-19 Pandemic. Discussion paper, National Bureau of Economic Research.

Balleer, Almut, Sebastian Link, Manuel Menkhoff, and Peter Zorn, 2020. Demand or Supply? Price Adjust-

ment during the COVID-19 Pandemic. Working Paper.

Baqaee, David, and Emmanuel Farhi, 2020. Supply and Demand in Disaggregated Keynesian Economies

with an Application to the Covid-19 Crisis. Discussion paper, National Bureau of Economic Research.

, 2021. Keynesian Production Networks and the Covid-19 Crisis: A Simple Benchmark. in AEA Papers and Proceedings vol. 111 pp. 272–76.

Bartik, Alexander W, Marianne Bertrand, Zoe Cullen, Edward L Glaeser, Michael Luca, and Christopher

Stanton, 2020. The Impact of COVID-19 on Small Business Outcomes and Expectations. Proceedings of the National Academy of Sciences.

Bartlett, Morse, 2020. Small Business Survival Capabilities and Policy Effectiveness: Evidence from Oak-

land. Discussion paper, National Bureau of Economic Research.

Berger, Allen, and Gregory Udell, 1998. The Economics of Small Business Finance: The Roles of Private

Equity and Debt Markets in the Financial Growth Cycle. Journal of Banking & Finance 22, 613 – 673.

Buffington, Catherine, Carrie Dennis, Emin Dinlersoz, Lucia Foster, and Shawn Klimek, 2020. Measuring

the Effect of COVID-19 on US Small Businesses: The Small Business Pulse Survey. Discussion paper, Working Paper.

Campello, Murillo, Gaurav Kankanhalli, and Pradeep Muthukrishnan, 2020. Corporate Hiring under

COVID-19: Labor Market Concentration, Downskilling, and Income Inequality. Discussion paper, Na- tional Bureau of Economic Research.

Cherry, Susan, Erica Jiang, Gregor Matvos, Tomasz Piskorski, and Amit Seru, 2021. Government and Private

Household Debt Relief during COVID-19. Working Paper.

Chetty, Raj, John N Friedman, Nathaniel Hendren, Michael Stepner, and The Opportunity Insights Team,

2020. How did COVID-19 and stabilization policies affect spending and employment? A new real-time economic tracker based on private sector data National Bureau of Economic Research Cambridge, MA.

Coibion, Olivier, Yuriy Gorodnichenko, and Michael Weber, 2020a. How did US Consumers Use Their

Stimulus Payments?. NBER Working Paper.

, 2020b. Labor Markets During the COVID-19 Crisis: A Preliminary View. NBER Working Paper.

, 2020c. The Cost of the COVID-19 Crisis: Lockdowns, Macroeconomic Expectations, and Consumer Spending. NBER Working Paper.

Cox, Natalie, Peter Ganong, Pascal Noel, Joseph Vavra, Arlene Wong, Diana Farrell, and Fiona Greig, 2020.

Initial Impacts of the Pandemic on Consumer Behavior: Evidence from Linked Income, Spending, and Savings Data. University of Chicago, Becker Friedman Institute for Economics Working Paper.

Fairlie, Robert W, 2020. The Impact of COVID-19 on Small Business Owners: Continued Losses and the

Partial Rebound in May 2020. Discussion paper, National Bureau of Economic Research.

Giglio, Stefano, Matteo Maggiori, Johannes Stroebel, and Stephen Utkus, 2020. Inside the Mind of a Stock

Market Crash. Discussion paper, National Bureau of Economic Research.

Gilchrist, Simon, Raphael Schoenle, Jae Sim, and Egon Zakrajšek, 2017. Inflation Dynamics during the

Financial Crisis. American Economic Review 107, 785–823.

Gopal, Manasa, and Philipp Schnabl, 2020. The Rise of Finance Companies and FinTech Lenders in Small

Business Lending. Working Paper.

Granja, João, Christos Makridis, Constantine Yannelis, and Eric Zwick, 2020. Did the Paycheck Protection

Program Hit the Target?. Working Paper 27095 National Bureau of Economic Research.

Guerrieri, Veronica, Guido Lorenzoni, Ludwig Straub, and Iván Werning, 2020. Macroeconomic implica-

tions of COVID-19: Can negative supply shocks cause demand shortages?. Discussion paper, National Bureau of Economic Research.

Howell, Sabrina, Theresa Kuchler, David Snitkof, Johannes Stroebel, and Jun Wong, 2021. Racial Disparities

in Access to Small Business Credit: Evidence from the Paycheck Protection Program. Discussion paper, Working Paper.

Humphries, John Eric, Christopher Neilson, and Gabriel Ulyssea, 2020a. The Evolving Impacts of COVID-

19 on Small Businesses Since the CARES Act. Cowles Foundation Discussion Paper.

Humphries, John Eric, Christopher A. Neilson, and Gabriel Ulyssea, 2020b. Information Frictions and Ac-

cess to the Paycheck Protection Program. Journal of Public Economics 190, 104244.

Joaquim, Gustavo, and Felipe Netto, 2020. Bank Incentives and the Impact of the Paycheck Protection Pro-

gram. Working Paper.

Kim, Ryan, 2020. The Effect of the Credit Crunch on Output Price Dynamics: The Corporate Inventory and

Liquidity Management Channel. Quarterly Journal of Economics.

Kuchler, Theresa, Lin Peng, Johannes Stroebel, Yan Li, and Dexin Zhou, 2020. Social Proximity to Capital:

Implications for Investors and Firms. Discussion paper, Working paper.

Kuchler, Theresa, Dominic Russel, and Johannes Stroebel, 2020. The Geographic Spread of COVID-19 Corre-

lates with Structure of Social Networks as Measured by Facebook. Working Paper 26990 National Bureau of Economic Research.

Perrin, Andrew, and Monica Anderson, 2019. Share of U.S. Adults Using Social Media, Including Facebook,

is Mostly Unchanged Since 2018. Pew Research Center.

Rehbein, Oliver, and Simon Rother, 2020. Distance in Bank Lending: The Role of Social Networks. Discus-

sion paper, University of Bonn and University of Mannheim, Germany.

USA Today, 2020. COVID-19 Restrictions: Map of COVID-19 Case Trends, Restrictions and Mobility. Ac-

cessed: July 2th 2021.

Wilson, Riley, 2019. The Impact of Social Networks on EITC Claiming Behavior. Brigham Young University,

Department of Economics.

Table 1: Variable Definitions

Variable Definition

Age Number of years since business started.

Industry Self-reported industry of the owner or manager.

Gender of Owner Gender of owner or manager that responded to the survey.

Sales Value of total revenues (sales) of the business in 2019.

In-person Interactions Share of business’s interactions between clients/customers and employ- ees/workers that need to be conducted in the same physical location.

Advertising Dummy that takes a value of one if the business advertised on Facebook at any point prior to April 2020. Respondents are classified into ones that have no matched Facebook page ("no matched firm"), businesses that have a matched firm page on Facebook but have not advertised previously ("matched firm, no advertising"), or businesses that have a matched firm page on Facebook and have advertised before April 2020 ("matched firm, advertising").

Access to Financing Source of capital or funds the business has access to. Formal financing is a dummy that takes a value of one if the business has a line of credit or loan from a financial institution. Informal finance is a dummy that takes a value of one if the business has access to community donations, personal savings, funds from family and friends, loans from retirement funds, or unemployment benefits. Firms that have access to both formal and infor- mal sources of finances or neither of the two are classified as "both" and "none" respectively.

Type of Shock Based on the business’s expectation of their biggest challenge in the next few months. Demand shock is a dummy that takes a value of one if the business reports struggling with “lack of demand”, “repaying loans”, or “cashflow”. Supply shock is a dummy that takes a value of one if the busi- ness reports struggling with “inventory”, “finding supplies”, “logistics”, “lack of staff”, or “government health authority orders”. We classify the business shock as Other/None if they report the shock as “other” or “none of the above”.

COVID Cashflows Based on the cashflow of the business in the past 30 days. “Outflow <= Inflow” is a dummy that takes a value of one if the business reported the cash outflow was less than inflow or that the outflow was about the same as inflow. “Outflow > Inflow” is a dummy that takes a value of one if the business reported its cash outflow was greater than inflow.

Cashflow Concerns Cashflow concerns are “low” if the business reports they are somewhat concerned or not concerned about the business’ cashflow situtation over the next three months, and “high” if the respondent says they are very concerned about the cashflow situation.

Variable Definitions - Continued

Variable Definition

Payment Struggles A business is classified as facing “some” payment struggles if they report struggling with employee/worker salaries and wages, bills or accounts payable, debt or loans, rent or lease, taxes, employee/worker benefits or hazard pay. If they do not face any payment struggles, the dummy vari- able “none” takes a value of one.

HH Responsibilities Household responsibilities are based on activities that the respondent had to spend more time on since the beginning of the pandemic. “Child-care” takes a value of one if the respondent had to provide daycare for children in their household or education for school-aged children. “Adult-care” takes a value of one if the respondent had to care for a dependent adult or household members who were self-isolating. “Other” takes a value of one if the respondent spent more time on housework.

HH Expenses Based on the respondent’s reply to how easy or difficult has it been to pay the household’s usual expenses. Respondents that reply "very easy" or "easy" are classified as easy, "neither easy nor difficult" are classified as neutral, and "difficult" or "very difficult" are classified as difficult.

Time on HH Activities Number of hours spent per day on domestic or household care activities. “Low” = less than 3 hours, “Medium” = between 3-6 hours, “High” = greater than 6 hours.

COVID Case Intensity Cumulative confirmed cases per capita obtained from Johns Hopkins Uni- versity (source). Case intensity is divided into terciles (low, medium, and high) of business respondents.

Decline Mobility Terciles of changes in median distance (in meters) traveled from the geohash-7 of the home obtained from Safegraph (source). We first cal- culate the median for each device and then find the median across all de- vices. The decline in mobility for the April survey is measured between mid-February and the start of May, and for the December survey, it is mea- sured between mid-February and the start of December.

Table 2: Share of Businesses Open

Note. The dependent variable is a dummy that takes a value of one if the business is operational or engag- ing in any revenue-generating activities at the time of the survey (April 2020 or December 2020). The table presents both univariate means as well as coefficients and standard errors from multivariate regressions. All variables are defined in Table 1.

April 2020 December 2020 Mean Reg Coeff SE Mean Reg Coeff SE

All 0.648 0.

By Age < 2 years 0.575 - - 0.737 - - 2-5 years 0.623 0.005 (0.010) 0.826 0.089*** (0.019)

5 years 0.681 0.013 (0.008) 0.883 0.153*** (0.015)

By Industry Agriculture or Mining 0.762 0.169*** (0.018) 0.827 0.005 (0.030) Construction 0.733 0.068*** (0.013) 0.904 0.074*** (0.020) Hotel/Café/Restaurant 0.560 −0.017 (0.013) 0.814 0.027 (0.023) Information/Communications 0.795 0.120*** (0.012) 0.857 0.035* (0.021) Manufacturing 0.796 0.103*** (0.016) 0.883 0.044* (0.025) Retail and Wholesale Trade 0.717 0.119*** (0.009) 0.874 0.066*** (0.016) Services 0.587 −0.001 (0.008) 0.844 0.040*** (0.015) Transportation and Logistics 0.694 0.051*** (0.018) 0.854 0.004 (0.036) Other 0.614 - - 0.796 - -

By Gender of Owner Female 0.620 - - 0.825 - - Male 0.685 0.023*** (0.006) 0.856 0.014 (0.010)

By Sales < $50k 0.544 - - $50k-$1m 0.669 0.155*** (0.007)

$1m 0.812 0.288*** (0.008)

By In-Person Interaction More than half 0.538 - - Half or less 0.757 0.230*** (0.006)

By Facebook Advertising No matched firm 0.620 - - Matched firm, no advertising 0.580 −0.003 (0.012) Matched firm, advertising 0.664 0.070*** (0.009)

By Covid Case Intensity Low 0.666 - - 0.844 - - Medium 0.665 −0.011 (0.007) 0.846 0.016 (0.013) High 0.615 −0.058*** (0.007) 0.824 −0.005 (0.013)

By Decline in Mobility Low 0.644 - - 0.832 - - Medium 0.650 −0.002 (0.007) 0.835 0.006 (0.013) High 0.653 −0.001 (0.007) 0.846 0.015 (0.013)

Table 3: Type of Shock Faced by Business

Note. The dependent variable is based on the business’s expectation of its biggest challenge in the next few months. We classified the business as primarily facing a demand shock if they responded that their biggest struggle would be “Lack of demand”, “Cashflow”, or “Repaying Loans”. We classified the businesses as primarily facing a supply shock if they responded to be primarily struggling due to “Inventory”, “Logistics (e.g., shipping, delivering services or goods)”, “Finding supplies”, “Lack of staff”, or “Government/Health Authority Orders”. Panel A uses the April 2020 wave of the survey, Panel B the December 2020 wave. Not all covariates are available in the December wave. The table presents both univariate means as well as coefficients and standard errors from multivariate regressions. All variables are defined in Table 1.

Panel A — April 2020

Type of Shock = Demand Type of Shock = Supply Mean Reg Coeff SE Mean Reg Coeff SE All 0.545 0. By Age < 2 years 0.571 - - 0.271 - - 2-5 years 0.603 0.028 (0.028) 0.261 −0.003 (0.025)

5 years 0.526 −0.033 (0.024) 0.317 0.017 (0.022)

By Industry Construction 0.529 0.075* (0.041) 0.327 −0.020 (0.037) Hotel/Café/Restaurant 0.578 0.089** (0.041) 0.346 0.028 (0.039) Information/Communications 0.613 0.074** (0.035) 0.203 −0.063** (0.029) Manufacturing 0.511 0.126** (0.055) 0.365 −0.072 (0.048) Retail and Wholesale Trade 0.519 −0.023 (0.028) 0.379 0.145*** (0.026) Services 0.598 0.089*** (0.024) 0.250 −0.004 (0.021) Transportation and Logistics 0.478 0.034 (0.054) 0.355 0.012 (0.050) Other 0.501 - - 0.291 - - By Gender of Owner Female 0.554 - - 0.287 - - Male 0.539 −0.018 (0.018) 0.313 0.027* (0.016) By Sales < $50k 0.577 - - 0.251 - - $50k-$1m 0.599 −0.007 (0.021) 0.277 0.039** (0.019)

$1m 0.460 −0.123*** (0.028) 0.385 0.131*** (0.026) By In-Person Interaction More than half 0.529 - - 0.344 - - Half or less 0.559 0.024 (0.018) 0.269 −0.033* (0.017) By Access to Financing None 0.547 - - 0.276 - - Formal 0.577 0.064* (0.033) 0.309 −0.002 (0.030) Informal 0.568 −0.013 (0.021) 0.278 0.025 (0.018) Both 0.634 0.101*** (0.026) 0.267 −0.030 (0.024) By Covid Case Intensity Low 0.509 - - 0.334 - - Medium 0.555 0.035* (0.021) 0.288 −0.022 (0.019) High 0.571 0.041* (0.021) 0.277 −0.047** (0.019) By Decline in Mobility

Low 0.548 - - 0.297 - - Medium 0.521 −0.033 (0.021) 0.317 0.033* (0.019) High 0.565 0.040* (0.021) 0.286 0.006 (0.019)

Type of Shock Faced by Business — Continued

Panel B — December 2020

Type of Shock = Demand Type of Shock = Supply Mean Reg Coeff SE Mean Reg Coeff SE All 0.457 0.

By Age < 2 years 0.559 - - 0.236 - - 2-5 years 0.516 −0.028 (0.023) 0.291 0.056*** (0.020)

5 years 0.418 −0.133*** (0.018) 0.382 0.147*** (0.016)

By Industry Agriculture or Mining 0.326 −0.122*** (0.039) 0.424 0.104*** (0.040) Construction 0.334 −0.104*** (0.031) 0.493 0.164*** (0.032) Hotel/Café/Restaurant 0.449 −0.031 (0.030) 0.389 0.104*** (0.029) Information/Communications 0.585 0.119*** (0.029) 0.229 −0.089*** (0.025) Manufacturing 0.443 −0.005 (0.039) 0.412 0.093** (0.038) Retail and Wholesale Trade 0.488 −0.002 (0.023) 0.378 0.091*** (0.022) Services 0.486 0.009 (0.021) 0.307 0.015 (0.019) Transportation and Logistics 0.360 −0.067 (0.048) 0.424 0.095* (0.049) Other 0.454 - - 0.297 - -

By Gender of Owner Female 0.493 - - 0.299 - - Male 0.433 −0.045*** (0.014) 0.378 0.061*** (0.014)

By Covid Case Intensity Low 0.461 - - 0.342 - - Medium 0.468 0.006 (0.018) 0.336 0.006 (0.018) High 0.461 0.005 (0.018) 0.323 0.005 (0.018)

By Decline in Mobility Low 0.457 - - 0.336 - - Medium 0.444 −0.011 (0.017) 0.344 −0.011 (0.017) High 0.488 0.042** (0.018) 0.321 0.042** (0.018)

Table 4: Pricing Response to the Panedemic

Note. The dependent variable in columns 1-3 (4-6) takes a value of one if the business increased (decreased) the average prices on its goods and services in the last 30 days. Results are from the April 2020 wave of the survey. The table presents both univariate means as well as coefficients and standard errors from multivariate regressions. All variables are defined in Table 1.

Price Increases Price Decreases Mean Reg Coeff SE Mean Reg Coeff SE All 0.039 0.242 By Age < 2 years 0.044 - - 0.284 - - 2-5 years 0.035 −0.000 (0.011) 0.277 −0.027 (0.025)

5 years 0.036 −0.005 (0.010) 0.219 −0.049** (0.022) By Industry Agriculture or Mining 0.068 0.038 (0.031) 0.205 0.030 (0.050) Construction 0.061 0.024 (0.018) 0.221 0.035 (0.034) Hotel/Café/Restaurant 0.067 0.035* (0.019) 0.271 0.044 (0.037) Information/Communications 0.040 0.007 (0.013) 0.257 0.059* (0.031) Manufacturing 0.026 −0.001 (0.020) 0.202 −0.013 (0.042) Retail and Wholesale Trade 0.042 0.006 (0.010) 0.259 0.042* (0.023) Services 0.031 0.002 (0.008) 0.258 0.050** (0.021) Transportation and Logistics 0.046 0.018 (0.023) 0.324 0.147*** (0.049) Other 0.027 - - 0.194 - - By Sales < $50k 0.047 - - 0.265 - - $50k-$1m 0.024 −0.032*** (0.008) 0.254 −0.015 (0.019) $1m 0.055 −0.005 (0.012) 0.198 −0.031 (0.024) By In-Person Interaction More than half 0.046 - - 0.230 - - Half or less 0.033 −0.012 (0.007) 0.250 0.036** (0.016) By Gender of Owner Female 0.030 - - 0.243 - - Male 0.047 0.017** (0.007) 0.242 0.006 (0.015) By Type of Shock Supply 0.061 - - 0.218 - - Demand 0.027 −0.033*** (0.009) 0.281 0.006 (0.018) Other/None 0.039 −0.016 (0.013) 0.138 −0.057*** (0.021) By COVID Cashflows Outflow <= Inflow 0.046 - - 0.199 - - Outflow > Inflow 0.029 −0.020*** (0.007) 0.296 0.034** (0.016) By Cashflow Concerns Low 0.043 - - 0.183 - - High 0.032 −0.004 (0.008) 0.336 0.086*** (0.018) By Access to Financing None 0.032 - - 0.251 - - Formal 0.027 −0.002 (0.011) 0.204 −0.024 (0.028) Informal 0.043 0.007 (0.007) 0.246 −0.019 (0.018) Both 0.054 0.016 (0.011) 0.256 0.019 (0.022) By Payment Struggles None 0.038 - - 0.153 - - Some 0.038 0.011 (0.008) 0.302 0.084*** (0.018)

Table 5:

Sources of Credit

Note.

The dependent variable takes a value of one if the business has access to a loan or line of credit from a financial institution (columns 1-3), informal

financing (columns 4-6), applied for a government grant during COVID-19 (columns 7-9), or applied for a new bank loan during COVID-19 (columns10-12). Results are from the April 2020 wave of the survey. The table presents both univariate means as well as coefficients and standard errors frommultivariate regressions. All variables are defined in Table 1.

Credit from FI

Access to Informal Financing

COVID Gov Loan App

COVID Bank Loan App

Mean

Reg Coeff

SE

Mean

Reg Coeff

SE

Mean

Reg Coeff

SE

Mean

Reg Coeff

SE

All

0.265

0.631

0.421

0.186

By Age < 2 years

0.140

-^ -^

0.625

-^ -^

0.286

-^ -^

0.123

-^

2-5 years

0.198

−

0.013

(0.021)

0.630

0.008

(0.028)

0.408

0.022

(0.026)

0.178

−

0.010

(0.023)

5 years

0.325

0.034*

(0.018)

0.635

0.036

(0.024)

0.468

0.007

(0.022)

0.212

−

0.029

(0.019)

By Industry Agriculture or Mining

0.319

0.122**

(0.050)

0.716

0.032

(0.055)

0.238

−

0.158***

(0.053)

0.149

−

0.029

(0.046)

Construction

0.371

0.075**

(0.037)

0.576

−

0.079*

(0.041)

0.490

0.005

(0.040)

0.254

0.022

(0.040)

Hotel/Café/Restaurant

0.343

0.050

(0.038)

0.597

−

0.062

(0.042)

0.640

0.122***

(0.039)

0.331

0.077*

(0.045)

Information/Communications

0.265

0.050*

(0.030)

0.694

0.030

(0.034)

0.384

−

0.026

(0.034)

0.181

−

0.004

(0.030)

Manufacturing

0.435

0.131***

(0.048)

0.486

−

0.163***

(0.056)

0.412

−

0.060

(0.057)

0.233

0.002

(0.058)

Retail and Wholesale Trade

0.219

0.020

(0.022)

0.628

−

0.035

(0.027)

0.341

−

0.039

(0.025)

0.152

−

0.009

(0.022)

Services

0.251

0.024

(0.020)

0.649

−

0.011

(0.024)

0.451

0.025

(0.023)

0.176

−

0.004

(0.021)

Transportation and Logistics

0.364

0.064

(0.050)

0.500

−

0.151***

(0.054)

0.457

−

0.006

(0.051)

0.270

0.012

(0.050)

Other

0.244

-^ -^

0.654

-^ -^

0.409

-^ -^

0.179

-^

By Gender of Owner Female

0.229

-^ -^

0.622

-^ -^

0.404

-^ -^

0.157

-^

Male

0.308

0.018

(0.015)

0.640

0.035**

(0.017)

0.445

−

0.020

(0.017)

0.224

0.013

(0.016)

By In-Person Interaction More than half

0.328

-^ -^

0.639

-^ -^

0.504

-^ -^

0.238

-^

Half or less

0.227

−

0.051***

(0.016)

0.628

−

0.020

(0.018)

0.371

−

0.068***

(0.017)

0.155

−

0.034**

(0.017)

By Sales < $50k

0.105

-^ -^

0.659

-^ -^

0.223

-^ -^

0.080

-^

$50k-$1m

0.310

0.181***

(0.017)

0.650

−

0.025

(0.020)

0.562

0.280***

(0.020)

0.270

0.139***

(0.018)

$1m

0.499

0.355***

(0.024)

0.567

−

0.103***

(0.027)

0.529

0.233***

(0.026)

0.312

0.149***

(0.025)

By Type of Shock Supply

0.394

-^ -^

0.170

-^

Demand

0.481

0.095***

(0.019)

0.230

0.060***

(0.018)

Other/None

0.237

−

0.132***

(0.025)

0.067

−

0.079***

(0.020)

By Access to Financing None

0.314

0.103

Formal

0.589

0.168***

(0.032)

0.382

0.204***

(0.037)

Informal

0.442

0.053***

(0.019)

0.199

0.015

(0.016)

Both

0.638

0.181***

(0.024)

0.459

0.283***

(0.028)

Table 6: Business-Household Interactions (Owners and Managers)

Note. The dependent variable in Columns 1-3 (4-6) takes a value of one if the business reported that their household (business) responsibilities affected their ability to focus on their business (household) during the COVID-19 pandemic “a lot” or “a great deal”. Panel A uses the April 2020 wave of the survey, Panel B the December 2020 wave. Not all covariates are available in the December wave. The table presents both univariate means as well as coefficients and standard errors from multivariate regressions. All variables are defined in Table 1.

Panel A — April 2020

Impact of HH on business Impact of business on HH Mean Reg Coeff SE Mean Reg Coeff SE All 0.319 0.296 By Age < 2 years 0.385 - - 0.302 - - 2-5 years 0.380 0.010 (0.024) 0.325 0.009 (0.025)

5 years 0.284 −0.014 (0.022) 0.287 −0.014 (0.022) By Industry Agriculture or Mining 0.242 −0.040 (0.040) 0.185 −0.086** (0.040) Construction 0.256 −0.017 (0.032) 0.287 0.016 (0.036) Hotel/Café/Restaurant 0.375 0.044 (0.034) 0.478 0.127*** (0.038) Information/Communications 0.321 0.037 (0.030) 0.251 −0.007 (0.030) Manufacturing 0.282 0.019 (0.040) 0.264 −0.004 (0.043) Retail and Wholesale Trade 0.325 0.000 (0.022) 0.298 0.007 (0.024) Services 0.344 0.018 (0.020) 0.299 0.004 (0.021) Transportation and Logistics 0.355 −0.012 (0.043) 0.364 0.020 (0.050) Other 0.290 - - 0.267 - - By Gender of Owner Female 0.347 - - 0.305 - - Male 0.284 −0.002 (0.015) 0.286 −0.011 (0.016) By Sales < $50k 0.363 - - 0.272 - - $50k-$1m 0.335 −0.006 (0.018) 0.324 0.060*** (0.019) $1m 0.257 −0.025 (0.022) 0.314 0.098*** (0.023) By In-Person Interaction More than half 0.317 - - 0.335 - - Half or less 0.320 0.017 (0.015) 0.268 −0.023 (0.016) By HH Responsibilities None - - - - Child care 0.515 0.229*** (0.017) 0.381 0.096*** (0.017) Adult/ HH member care 0.476 0.099*** (0.018) 0.388 0.068*** (0.018)

Other 0.401 0.017 (0.015) 0.321 −0.009 (0.016) By HH Expenses Easy 0.177 - - 0.163 - - Neutral 0.267 0.058*** (0.017) 0.245 0.078*** (0.017) Difficult 0.473 0.204*** (0.019) 0.441 0.242*** (0.020) By Time on HH Activities Low 0.176 - - 0.227 - - Medium 0.424 0.151*** (0.017) 0.346 0.048*** (0.017) High 0.667 0.326*** (0.026) 0.457 0.125*** (0.027) By Covid Case Intensity Low 0.287 - - 0.273 - - Medium 0.325 0.022 (0.017) 0.297 0.016 (0.018) High 0.348 0.027 (0.018) 0.318 0.019 (0.019) By Decline in Mobility Low 0.312 - - 0.304 - - Medium 0.310 −0.016 (0.017) 0.288 −0.010 (0.018) High 0.334 −0.002 (0.017) 0.293 −0.011 (0.018)

Business-Household Interactions (Owners and Managers) — Continued

Panel B — December 2020

Impact of HH on business Impact of business on HH Mean Reg Coeff SE Mean Reg Coeff SE All 0.287 0.247

By Age < 2 years 0.335 - - 0.210 - - 2-5 years 0.421 0.103* (0.057) 0.344 0.147** (0.059)

5 years 0.276 0.009 (0.042) 0.281 0.142*** (0.042)

By Industry Agriculture or Mining 0.250 −0.164* (0.093) 0.200 −0.055 (0.107) Construction 0.259 −0.081 (0.084) 0.259 0.012 (0.096) Hotel/Café/Restaurant 0.250 −0.109 (0.070) 0.364 0.069 (0.080) Information/Communications 0.276 −0.072 (0.060) 0.288 0.019 (0.068) Manufacturing 0.273 0.003 (0.108) 0.227 0.039 (0.095) Retail and Wholesale Trade 0.364 0.022 (0.058) 0.247 −0.063 (0.056) Services 0.299 −0.096** (0.048) 0.271 −0.020 (0.053) Transportation and Logistics 0.200 −0.181* (0.093) 0.100 −0.159 (0.098) Other 0.286 - - 0.230 - -

By Gender of Owner Female 0.359 - - 0.316 - - Male 0.257 −0.016 (0.037) 0.201 −0.098** (0.038)

By HH Responsibilities None - - - - Child care 0.506 0.191*** (0.042) 0.323 0.067* (0.040) Adult/ HH member care 0.508 0.162*** (0.046) 0.316 0.018 (0.043) Other 0.405 0.025 (0.038) 0.262 −0.070* (0.038)

By HH Expenses Easy 0.191 - - 0.184 - - Neutral 0.241 0.012 (0.045) 0.200 −0.007 (0.046) Difficult 0.440 0.128*** (0.047) 0.373 0.180*** (0.048)

By Time on HH Activities Low 0.156 - - 0.234 - - Medium 0.481 0.184*** (0.045) 0.308 0.011 (0.042) High 0.688 0.361*** (0.070) 0.359 0.085 (0.073)

By Covid Case Intensity Low 0.307 - - 0.310 - - Medium 0.316 0.008 (0.042) 0.238 −0.059 (0.046) High 0.340 −0.000 (0.043) 0.271 −0.035 (0.046)

By Decline in Mobility Low 0.366 - - 0.317 - - Medium 0.315 −0.030 (0.045) 0.257 −0.047 (0.048) High 0.290 −0.034 (0.044) 0.249 −0.075 (0.048)

Table 7: Business-Household Interactions (Employee Responses)

Note. The dependent variable in columns 1-3 (4-6) takes a value of one if the employee reported that their household (business) responsibilities affected their ability to focus on this business (household) during the coronavirus (COVID-19) pandemic “a lot” or “a great deal”. Results are from the April 2020 wave of the survey. The table presents both univariate means as well as coefficients and standard errors from multivariate regressions. All variables are defined in Table 1.

Panel A — Employed Respondents

Impact of HH on business Impact of business on HH Mean Reg Coeff SE Mean Reg Coeff SE All 0.240 0.184 By Age < 25 years 0.228 - - 0.232 - - 25-45 years 0.287 0.008 (0.017) 0.213 −0.066*** (0.018) 45+ years 0.186 −0.062*** (0.018) 0.139 −0.118*** (0.018) By Industry Agriculture or Mining 0.173 −0.050 (0.037) 0.135 −0.062* (0.034) Construction 0.201 −0.038 (0.025) 0.115 −0.058*** (0.022) Hotel/Café/Restaurant 0.279 0.026 (0.028) 0.246 0.018 (0.029) Information/Communications 0.204 −0.010 (0.018) 0.139 −0.024 (0.017) Manufacturing 0.202 −0.017 (0.019) 0.161 −0.031* (0.019) Retail and Wholesale Trade 0.273 0.044** (0.021) 0.235 0.034 (0.022) Services 0.248 0.008 (0.016) 0.176 −0.023 (0.015) Transportation and Logistics 0.212 −0.012 (0.023) 0.175 −0.029 (0.022) Other 0.253 - - 0.196 - - By Firm Size < 50 0.251 - - 0.184 - - 50 - 250 0.234 0.002 (0.014) 0.182 0.024* (0.014)

250 0.228 −0.002 (0.013) 0.183 0.036*** (0.012) By Remote work No 0.227 - - 0.205 - - At least some time 0.250 0.020* (0.012) 0.166 −0.051*** (0.012) By Gender Female 0.263 - - 0.206 - - Male 0.189 −0.047*** (0.011) 0.140 −0.052*** (0.011) By HH Responsibilities None 0.130 - - 0.121 - - Child care 0.150 −0.148*** (0.012) 0.130 −0.084*** (0.011) Adult/ HH member care 0.212 −0.177*** (0.021) 0.167 −0.113*** (0.020) By Education High school or less 0.218 - - 0.168 - - Non-college degree 0.361 −0.021 (0.037) 0.227 −0.019 (0.036) College degree 0.255 −0.058 (0.038) 0.199 −0.029 (0.037) By Time on HH Activities Low 0.132 - - 0.120 - - Medium 0.332 0.142*** (0.013) 0.238 0.091*** (0.012) High 0.569 0.334*** (0.021) 0.377 0.196*** (0.021) By Covid Case Intensity Low 0.231 - - 0.193 - - Medium 0.235 0.009 (0.013) 0.183 0.006 (0.013) High 0.252 0.028** (0.013) 0.175 −0.003 (0.013) By Decline in Mobility Low 0.115 - - 0.115 - - Medium 0.104 −0.009 (0.013) 0.104 −0.004 (0.012) High 0.102 −0.009 (0.013) 0.102 −0.031** (0.012) By Government assistance Not applied 0.230 - - 0.176 - - Applied 0.387 0.101*** (0.023) 0.296 0.067*** (0.024)