Policy Research Working Paper                      10562




     The Resilience of SMEs and Large Firms
          in the COVID-19 Pandemic
                      A Decomposition Analysis

                                Mohammad Amin
                                  Filip Jolevski
                                 Asif M. Islam




Global Indicators Group &
Middle East and North Africa Region
September 2023
Policy Research Working Paper 10562


  Abstract
 This study analyzes the difference in the decline in sales                         large firms from a given level of financial constraints, input
 between small and medium-size enterprises and large firms                          supply disruptions, and country-industry-specific factors,
 (the “gap”) following the outbreak of COVID-19 in 19                               and benefitted less from a given level of initial labor pro-
 developing countries. The decline in sales as a percentage                         ductivity. These differences in the returns to factors also
 of the pre-pandemic level was bigger for small and medi-                           widened the gap. Second, the gap was much larger at the
 um-size enterprises by 12.2 percentage points. The paper                           relatively high quantiles of sales decline distribution, indi-
 uses the Kitagawa-Oaxaca-Blinder and quantile decompo-                             cating that relative to large firms, small and medium-size
 sition methods to estimate individual factors’ contributions                       enterprises were much less resilient to large shocks than
 to the gap at the mean and across the sales decline distri-                        small shocks. Third, individual factors’ contribution to the
 bution. Several important results emerge. First, relative to                       gap varied across the sales decline distribution. Thus, the
 large firms, small and medium-size enterprises faced greater                       optimal policy mix depends on the size of the shock. Fourth,
 incidence of input supply disruptions during the pandemic,                         there were some important differences between geograph-
 had lower initial labor productivity levels, and were concen-                      ical regions in what drove the gap. Thus, an eclectic policy
 trated in country-industry cells with a bigger sales declines.                     approach is needed that duly accounts for the prevailing
 These differences in the level of factors widened the gap.                         local conditions.
 Small and medium-size enterprises also suffered more than




 This paper is a product of the Global Indicators Group, Development Economics and the Office of the Chief Economist,
 Middle East and North Africa Region. It is part of a larger effort by the World Bank to provide open access to its research
 and make a contribution to development policy discussions around the world. Policy Research Working Papers are also
 posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at mamin@worldbank.org.




         The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
         issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
         names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
         of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
         its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.


                                                       Produced by the Research Support Team
       The Resilience of SMEs and Large Firms in the COVID-19
                  Pandemic: A Decomposition Analysis
                                               August 2023
                      By: Mohammad Amin*, Filip Jolevski**, and Asif M. Islam***




Keywords: SMEs, crisis, COVID-19, resilience, firm size, decomposition
JEL Codes: D21, D22, E32, L25

* Corresponding author. Senior Economist, Enterprise Analysis Unit, DECIG, World Bank, Washington,
DC. Email: mamin@worldbank.org.
** Economist, Enterprise Analysis Unit, DECIG, World Bank. E-mail: fjolevski@worldbank.org
*** Senior Economist, Middle East and North Africa region Chief Economist Office of the World Bank.
E-mail: aislam@worldbank.org

The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They
do not necessarily represent the views of the World Bank and its affiliated organizations, or those of the
Executive Directors of the World Bank or the governments they represent.
We thank the Enterprise Analysis Unit of the Development Economics Global Indicators Department of
the World Bank Group for making the data available. We would also like to thank Jorge Luis Rodriguez
Meza, Norman Loayza, and participants at a seminar organized by DECIG, World Bank for useful


comments. All remaining errors are our own.
1. Introduction

Small and medium-size enterprises (SMEs) provide jobs and incomes to millions of individuals in

the developing countries (see ILO 2019, ILO and GIZ 2013, Ayyagari et al. 2011, 2014).

According to the World Bank’s Enterprise Surveys data on formal firms in 125 developing

countries, SMEs (5-99 workers) account for about 47 percent of the total formal sector

employment and 52 percent of the total output on average across countries. The rest is accounted

for by large firms (>99 workers). Ayyagari et al. (2014) report that over 95 percent of the new

formal sector jobs are created by SMEs in the developing world and only 5 percent by large firms.

SMEs are also expected to promote inclusive growth (see OECD 2019). The contribution of SMEs

is even higher if we include the self-employed, micro firms (fewer than 5 workers), and the

unregistered or informal firms (see ILO 2019). Thus, it is not surprising that most countries have

implemented policies to promote SMEs.

        However, a lingering concern with SMEs is that they may be unduly affected by economic

downturns and crises, such as the COVID-19 pandemic. That is, SMEs are less resilient to crises

than large firms. This may happen for several reasons that are discussed in detail in Section 2. The

relative underperformance of SMEs in periods of crises implies that SMEs act as destabilizers,

magnifying the adverse impact of crises. Furthermore, while it may not be a surprise that SMEs

were considerably affected by the COVID-19 pandemic, what is interesting is uncovering the

potential contributors that favored large firms while debilitating SMEs. A rigorous comparison

and quantification of SMEs’ and large firms’ performance during a crisis in the context of

developing countries is lacking (see Section 2). This study attempts to fill this gap in the literature.

It analyzes the difference in the decline in sales due to the COVID-19 pandemic between SMEs

and large firms (henceforth, the “sales decline gap” or the “gap”) in 19 developing countries spread



                                                   2
across several regions of the world. At the mean and various quantiles of the sales decline

distribution, the size of the sales decline gap and the contribution of individual factors to the gap

are estimated. This is achieved using the traditional or mean Kitagawa-Oaxaca-Blinder (KOB)

decomposition methodology (Kitagawa 1955; Oaxaca 1973; Blinder 1973) and the quantile

decomposition methodology (Firpo et al. 2007, 2009). The decomposition exercise yielded several

important and policy-relevant insights.

        The COVID-19 pandemic resulted in a decline in sales of 34.8 percent from the pre-

pandemic level for SMEs and 22.6 percent for large firms. Thus, the decline in sales was bigger

for SMEs than for large firms by 12.2 percentage points. This is the unconditional total gap. The

decline in sales is larger for SMEs in all the countries except Chad and Mongolia (see figure 1).

One reason for the sales decline gap is that the level or endowment of factors that influence the

decline in sales is different for SMEs and large firms (the “endowment” effect). Another reason is

that the determinants of sales decline have a different impact on SMEs and large firms (the

“structural” effect). Both the “endowment” and “structural” effects contribute significantly to our

sample’s sales decline gap at the mean. Of the total gap of 12.2 percentage points at the mean in

the full sample, the aggregate (of all the individual factors’) “endowment” effect, accounts for

about 7.96 percentage points, or about 65.1 percent of the total gap. The remaining gap is due to

the “structural” effects.

        Regarding the statistically significant individual “endowment” effects at the mean, SMEs

have higher levels of input supply disruptions and lower initial labor productivity than large firms,

which widen the sales decline gap. This is not surprising as large firms have stronger connections

to supply chains (see Bak et al. 2023 for a literature review) and higher levels of productivity that

allow them to buffer against shocks (confirmed below). Another factor widening the gap is that,



                                                 3
relative to large firms, SMEs are disproportionately located in country-industry pairs that

experienced a larger decline in sales. SMEs have less foreign ownership and are less constrained

by skills shortages, which help to narrow the gap in some of the specifications. For the “structural”

effects, SMEs are more adversely affected than large firms by input supply disruptions, financial

constraints, high share of women workers, higher share of exports in total sales, and country-

industry specific factors. These differences widen the gap. However, SMEs are less adversely

affected by high tax rates than large firms, which narrows the gap.

        The quantile decomposition methodology developed by Firpo et al. (2007; 2009) is used

to examine the sales decline gap at various quantiles of the sales decline distribution. One

motivation for the exercise is that the decline in sales varies substantially across firms. About 74.6

percent of the firms in our sample experienced a decline in sales, 17.3 percent no change, and 8.1

percent an increase. An increase in sales can result from very different factors than a decrease in

sales. The same applies to a small vs. large decline in sales. Thus, the factors that influence the

performance of SMEs and large firms may also vary at low vs. high levels of sales decline. Another

motivation is that the sales decline gap in our sample varies significantly across the sales decline

distribution. Figure 2 illustrates the point. The indication here is that the factors that drive the gap

are also different at low vs. high levels of the sales decline.

        The quantile decomposition exercise provides several interesting results. We preview some

of the findings based on the full sample. First, the sales decline gap is statistically significant at all

the quantiles, but it is much smaller at the lower quantiles (Figure 2). This indicates that SMEs

underperform large firms much more when the crisis (sales decline) is more severe. Second, the

aggregate “endowment” effect plays a much bigger role in driving the total gap at higher quantiles

than at lower quantiles of the sales decline distribution. The opposite holds for the aggregate



                                                    4
“structural” effects. Thus, for relatively large shocks, the weaker resilience of SMEs is mainly

because SMEs have fewer resources than large firms. However, for the relatively smaller shocks,

it is mainly because SMEs benefit less than large firms from a given level of resources. Third, the

individual determinants of the sales decline gap contribute differently across the sales decline

distribution. For instance, differences in the levels of labor productivity between SMEs and large

firms widen the gap significantly via the “endowment” effect at the 40th to 90th quantiles but not

at lower quantiles. Thus, efforts to reduce the sales decline gap by improving SMEs’ labor

productivity are more effective if targeted to SMEs at the relatively high quantiles of the sales

decline distribution. However, differences in the incidence of input supply disruption matter across

all quantiles, emphasizing their importance in closing the gap.

        We explore the role of regional factors. To this end, we perform the mean and quantile

decomposition analysis separately for countries in Sub-Saharan Africa (SSA), Eastern Europe and

Central Asia (ECA), Latin America and the Caribbean (LAC), and the Middle East and North

Africa (MENA). Some commonalities and differences emerge. Regarding commonalities, we find

the following statistically significant results hold in all the four regions: the decline in sales at the

mean is much bigger for SMEs than large firms; country-industry specific factors contribute to the

gap at several quantiles; supply disruptions widen the gap at the mean via the “endowment” effect;

the sales decline gap is positive and large at most quantiles; 10 of the 17 explanatory variables that

we consider contribute to the total gap at one or more quantiles. Further, we find that the aggregate

“endowment” effect at the mean widens the gap by a large 37 to 124 percent of the total gap across

the four regions, which is statistically significant in three regions. Also, 16 of the 17 variables that

we consider contribute significantly to the gap at one or more quantiles in at least three out of the

four regions. These commonalities suggest that these factors may be prominent and are likely to



                                                   5
matter for a global sample of economies beyond those included in our sample. Regarding some of

the differences, the sales decline gap is only 6 percentage points in MENA and about 11 percentage

points in the other regions; government support significantly narrows the gap in MENA at some

of the quantiles but widens it in all the other regions; a shortage of skilled workers widens the gap

in SSA and MENA at some of the quantiles and narrows it in ECA and LAC. Several other

differences are discussed in detail below. Thus, an eclectic approach is required that incorporates

the common findings in this study and the broader literature while appropriately accounting for

the prevailing regional or local conditions.

       We contribute to the literature in several ways. First, we use cross-country comparable

survey data on formal SMEs and large firms in 19 developing countries. Most previous studies on

the impact of COVID-19 or other crises on small vs. large firms are restricted to rich countries or

a handful of developing countries (see Section 2). Second, we identify several factors that

contribute to the sales decline gap between SMEs and large firms. An estimate of their

contributions is provided. Third, we use the mean and quantile decomposition methodology to

decompose the contribution of the individual factors to the sales decline gap into “endowment”

and “structural” effects. To the best of our knowledge, ours is the first such study. Fourth, we shed

light on how the size of the gap and the contributions of the individual factors to the gap via the

“endowment” and “structural” effects vary along the sales decline distribution. We are not aware

of any previous work that does so. Fifth, we explore heterogeneities between geographical regions

in the size and nature of the sales decline gap. This is important for assessing the relevance of

regional factors. Sixth, the breadth of the surveys we use enables specific policy recommendations

regarding the importance of salient firm factors, such as initial labor productivity, input supply




                                                 6
disruptions during the crisis, financial constraints, international trade and foreign ownership, skills

shortages, and the presence of women workers.

       We emphasize that the decomposition analysis is descriptive and not causal. The

descriptive approach is consistent with previous research (see, for example, Nix et al. 2016, Essers

et al. 2021, and Islam and Amin 2023).



2. Literature review

In this section, we review the literature on the impact of the COVID-19 pandemic and earlier

shocks on firms. The focus is on studies that compare the impact on small vs. large firms. There

are several other studies that deal with the impact of recessions and crises on small firms alone.

Some of these studies are discussed in Section 3, where we motivate our main explanatory

variables.



2.1 Small vs. large firm performance during crises

Recent studies have explored how firms of different sizes performed during the COVID-19

pandemic. Using firm-level survey data on 23 economies in ECA collected by the World Bank’s

Enterprise Surveys, Bruhn et al. (2023) found that small firms suffered larger declines in output

and employment due to the COVID-19 pandemic than large firms. Using data from the US, Bloom

et al. (2021) showed that the smallest offline firms experienced sales drops of over 40 percent due

to the pandemic, compared with a 10 percent drop for the largest offline firms. Ding et al. (2021)

used data on 61 mostly developed economies and found that the stock market response to the

COVID-19 shock was less negative for large corporations than for small firms. Dörr et al. (2022)

found that in Germany, small firms experienced stronger financial distress than large firms due to



                                                  7
the pandemic and were more likely to have gone bankrupt without policy assistance. There are

other studies, such as Liu et al. (2022) for China, that found a large adverse effect of the pandemic

on small firms. However, these studies do not provide a comparison with large firms. We discuss

some of these studies in Section 3, where we motivate our main explanatory variables.

        There is also evidence based on other types of shocks. For instance, Chen and Lee (2023)

reported that in 18 developed economies in Europe, the global financial crisis of 2008–2009

resulted in a much larger decline in the TFP growth rate for SMEs (including micro firms) than

for large firms. Peric and Vitezic (2016) found that during the recession of 2008–2013 in Croatia,

larger firms performed much better in terms of turnover growth. Gertler and Gilchrist (1994) found

that during the 1981–82 recession in the US, large firms performed better than small firms in terms

of total production. The study by Ozar et al. (2008) for Türkiye found that during the 2001 Turkish

financial crisis, large firms performed better in terms of employment growth rate. While most

studies confirm less resilience to crises among relatively small firms, there are some that show the

opposite. For example, Varum and Rocha (2013) found that periods of economic downturn or

crisis between 1988 and 2007 in Portugal led to a larger decline in firm growth for large firms than

SMEs.

        The studies discussed above are, at best, a starting point for understanding the resilience of

small vs. large firms during a crisis. These studies do not explore the mechanisms or contributions

of the underlying factors to the differential impact by firm size. Another limitation is that many of

these studies focus on developed countries. A few studies, such as Ozar et al. (2008) and Liu et al.

(2022), are based on developing countries. However, these studies are restricted to a single

country, with no indication if the results apply more broadly to the developing world.




                                                  8
2.2 Reasons for differential effects by firm size

There are several reasons for the differential impact of a crisis on small and large firms. First,

access to finance and cash reserves plays a key role in allowing firms to tide over a crisis (see

Amin and Viganola 2023 and Ding et al. 2021). Smaller firms have fewer cash reserves and poorer

access to formal finance. Thus, the liquidity crunch that followed the COVID-19 pandemic and

the earlier financial crises is likely to adversely impact small firms more than large firms (see Iyer

et al. 2014). Second, small firms’ relative shortcomings in terms of technological, managerial, and

human capabilities may reduce their capacity to deal with a crisis (see United Nations 2022, Varum

and Rocha 2013). Third, small firms’ greater dependence on a small set of customers, suppliers

(Herskovic et al. 2022, Marconatto et al. 2022), and markets (see United Nations 2022) may lead

to increased difficulties in maintaining their activity in the face of the crisis. Fourth, one argument

in favor of smaller firms is that smaller firms may suffer less from inertia, rigidity, and sunk costs

(see Varum and Rocha 2013, Tan and See 2004, and Dean et al. 1998). This allows smaller firms

to be more flexible when adapting to economic downturns. However, Belghitar et al. (2022) argue

that despite their lean organizations, smaller firms are quite often characterized by a cost structure

that is more rigid than that of larger organizations. As a result, even a small drop in sales can have

an important impact on how efficiently the assets are used. Further, smaller firms tend to have

fewer resources than large firms, which makes it more difficult for smaller firms to detect and

respond to ever-changing environments (Mueller and Jungworth 2022, Chan et al. 2019). Fifth,

another factor favoring smaller firms is that such firms tend to concentrate on activities

characterized by economies of agglomeration rather than scale. Thus, smaller firms are less

affected by a reduction in demand and the subsequent production cuts (Varum and Rocha 2013).

Sixth, there are several factors, such as management quality, ownership structure, the presence of



                                                    9
women in the workforce, and innovation capacity, that may affect firms’ ability to deal with the

COVID-19 pandemic and other such crises. Although there is no formal study, the possibility that

these factors have a differential impact on small and large firms cannot be ruled out. Section 3

discusses several of these factors in detail.



3. Data and Main Variables

3.1 Data description

Our data sources are the World Bank’s Enterprise Surveys (ES) and the World Bank’s Enterprise

Surveys COVID-19 follow-up (COV-ES). The ES are firm-level surveys representative of a

country’s registered or formal private sector employing 5 or more employees in most

manufacturing and select service industries. Informal firms and businesses operating in agriculture,

the extractive sector, and service sectors such as banking and finance, health care, and education

are excluded from the sampling universe. A standard sampling procedure based on location, sector,

and firm size stratifications ensures representativeness at the country, sector, and firm size level,

while the use of a common questionnaire guarantees comparability across countries. The ES

included in this paper were conducted between 2016 and 2020, covering the pre-pandemic period

defined as before March 2020. After the pandemic was declared, the same firms interviewed in the

ES were re-contacted with the aim of measuring the effects of the pandemic (COV-ES). These

interviews were administered over the phone. Our baseline sample includes all the firms in the

developing countries (low- and middle-income countries) that were operational at the time of

COV-ES, were successfully interviewed in COV-ES between May and December 2020, and for

which data on the main variables are available. There are 5,197 firms (SMEs and large) spread

across 19 developing countries in this sample. In Appendix A, Table A1 contains the list of



                                                 10
countries, the number of firms, and the period the ES and COV-ES cover; Table A2 contains a

formal definition of all the variables; and Table A3 contains the summary statistics of the variables.



3.2 Main variables

The outcome or dependent variable is the decline in sales in the last month (after the outbreak of

COVID-19) compared to sales in the same month one year ago (pre-pandemic), expressed as a

percentage of the latter (Sales decline). The mean decline in sales equals 32.2 percent, the median

decline is 30, and the standard deviation is 32.7. The decline ranges between 100 and -300 percent.

A positive value of Sales decline implies that sales were lower after the outbreak of the pandemic

than before, and a negative value implies an increase in sales following the pandemic. About 74.6

percent of the firms experienced a decline in sales, 17.3 percent no change, and 8.1 percent an

increase. By country, the mean decline was smallest in Belarus (16 percent) and highest in Niger

(53.8 percent).



3.2.1 Main explanatory variables

The main explanatory variable is a dummy variable equal to 1 if the firm is an SME and 0 otherwise

(SME). Following the definition used by the ES for survey stratification, SMEs are all firms that

had 5-99 full-time workers employed in the pre-pandemic period. The rest (>99 full-time workers)

are large firms. There are 4,097 SMEs in our baseline sample, which is about 78.8 percent of all

firms. We consider several determinants of the firm’s performance during the pandemic. Many of

these determinants are for the pre-pandemic period (taken from ES), and some are for the period

after the outbreak of the pandemic (taken from COV-ES). We focus more on the pre-pandemic

variables because information is available on a much larger set of variables before the outbreak of



                                                 11
the pandemic than after. This provides a sense of whether pre-existing conditions of businesses

play a part in their resilience. Also, it helps avoid the simultaneity problem.

       We begin with a set of dummy variables indicating the country times industry to which the

firm belongs (Country-Industry fixed effects). Industry is defined at the 2-digit level (ISIC

Revision 3.1). These dummy variables account for all industry-wide factors, specific to each

country, that influence the decline in sales. By construction, they also account for all country-wide

factors (country fixed effects) and industry-wide factors (industry fixed effects). Several studies

have documented the importance of country-specific factors in the incidence, spread, and intensity

of COVID-19 infections. Bargain and Aminjonov (2021) show that COVID-19 infections per

capita were higher in countries with a higher poverty rate. Pitterle and Niermann (2021) show that

a country’s GDP growth performance during the pandemic was affected by its stringency of

containment measures, good governance, provision of fiscal support to private agents, and the

strength of the macroeconomic fundamentals. Chang et al. (2022) identify 21 country-level factors,

some of which include population density, level of urbanization, education, democracy,

corruption, quality of institutions, and health care infrastructure. Although limited, direct evidence

on the impact on firms is also available. For instance, Hu and Zhang (2021) show that better health

care, financial development, and the quality of institutions at the country level significantly

improve firms’ ability to deal with the pandemic.

       Industry-specific differences may arise because some “essential” industries were not

affected or were less affected by the mandatory lockdown. Also, the scope of remote work and

online sales varies across industries depending on the need for face-to-face interaction with

customers and physical proximity to other workers (see Bloom et al. 2023, Fairlie and Fossen

2022, del Rio-Chanona et al. 2020, Stemmler 2022).



                                                 12
       Disruptions in the supply of inputs due to the pandemic are well documented, although

their impact on firms remains to be thoroughly analyzed (see Butt 2021, Pitschner 2022, Meier

and Pinto 2020). We capture supply disruptions with a dummy variable equal to 1 if the firm’s

supply of inputs, raw materials, or finished goods and materials purchased to resell decreased in

the last month (after the outbreak of the pandemic) compared to the same month a year ago (pre-

pandemic). The data source is COV-ES.

       Voluntary dropouts from the labor market due to health or family reasons may have

resulted in firms facing a labor shortage (see Pitschner 2022, Forsythe et al. 2022). For instance,

Forsythe et al. (2022) find that for the US economy, the labor market remained surprisingly tight

throughout the entire pandemic, despite the dramatic job losses. A shortage of hard-to-replace

skilled workers may be especially worrisome. The COV-ES did not ask firms about skilled

shortages. We use a question in the ES that asks firms how severe (on a 0–4 scale) inadequately

educated workers is as an obstacle to their current operations (Skills obstacle).

       Next, we account for the labor productivity of the firm prior to the pandemic using ES. It

equals the log of total sales of the firm (in 2009 USD) in the last fiscal year divided by the number

of full-time workers at the end of the fiscal year covered by the ES. Several studies have found a

positive link between a firm’s pre-pandemic productivity and performance during the pandemic in

terms of sales, employment, and survival (see Bloom et al. 2023, Bruhn et al. 2023, Muzi et al.

2022, Fernández-Cerezo et al. 2022, Kozeniauskas et al. 2020). Unlike earlier recessions, COVID-

19 resulted in a larger drop in women’s relative to men’s employment (Bundervoet et al. 2022,

Adams-Prassl et al. 2020, Russell and Sun 2020, Bluedorn et al. 2023). One reason for this could

be the mandatory closure of schools and the collapse of childcare facilities due to the pandemic.

As a result, firms that are more dependent on women workers are likely to be more adversely



                                                 13
affected. We capture this effect using a dummy variable equal to 1 if the share of women workers

in the firm prior to the pandemic was more than 25 percent and 0 otherwise (High share of women

workers). The data source is ES.

       We also consider the financial condition of the firm. Studies find that a stronger financial

condition (more cash, less debt, and better access to finance) prior to the pandemic helped firms

perform better during the pandemic (Amin and Viganola 2023, Ding et al. 2021). Since many

SMEs may obtain finance from non-bank financial institutions, our main measure of a firm’s

financial constraints is a broad one. This equals the pre-pandemic severity level (on a 0–4 scale)

reported by the firms of access to finance as an obstacle to their operations (Finance obstacle). The

second measure focuses on bank finance. It is equal to 1 if the firm had an overdraft facility prior

to the pandemic and 0 otherwise (Overdraft). The data source for the two finance variables is ES.

The age of the firm and location (urban vs. rural) have been found to impact the firm’s performance

during the pandemic. Using data on over 4,000 firms in Spain, Fernández-Cerezo et al. (2022)

report a much larger decline in sales due to the pandemic among younger firms and firms located

in urban vs. rural areas. Age may matter because younger firms are less experienced and have

fewer resources to deal with the pandemic. Younger firms are also disadvantaged because they

have a less developed supplier network and a less dedicated customer base. Urban areas may be

more susceptible to the virus because of greater contact between people due to higher population

density, more crowded housing, and more interconnected public transport. Thus, we account for a

dummy variable equal to 1 if the firm is 10 years old or younger (in 2020) and 0 otherwise (Young

firm), and a dummy variable equal to 1 if the firm is in the country’s capital city and 0 otherwise

(Capital city). The data source for both variables is ES.




                                                 14
       We account for the proportion of the firm’s sales made directly abroad (Exports) and the

proportion of the firm owned by foreign entities (Foreign ownership). Both the variables are pre-

pandemic and taken from the ES. Studies have found that exporters (see Demir and Javorcik 2020)

and firms with foreign ownership (see Liu et al. 2021, Gu et al. 2020) may be more at risk from

the pandemic. This could be because exporters and foreign-affiliated firms tend to rely more on

global supply chains, which have undergone tremendous shocks during the pandemic. Lack of

proper international law means that an unexpected adverse economic shock, such as COVID-19,

may increase the risks of non-payment or non-delivery of prepaid goods across borders.

       Another important issue is the support provided by the government to help private firms

deal with the pandemic. Cirera et al. (2021) find that policies such as credit and cash transfers

appear to be helping firms address liquidity constraints, while receiving wage subsidies seems to

be associated with a lower probability of firing workers. Guerrero-Amezaga et al. (2022) find that

across eight Latin American countries, government support led to improved outcomes at the firm

level. Interestingly, both of these studies find that the smallest firms benefit less from government

programs. Thus, we account for a dummy variable equal to 1 if the firm received any national or

local government support in response to the pandemic and 0 otherwise (Government support). The

data source for the variable is COV-ES.

       A high tax burden squeezes profits and drains the firm of much-needed cash reserves. As

a result, high taxes may adversely affect firm performance during the pandemic. Our country-

industry fixed effects capture the tax burden that is common to all firms within a country or

country-industry cell, but not the difference between SMEs and large firms. We account for such

effects using a dummy variable equal to 1 if the firm in the pre-pandemic period reported the tax

rates as a major or very severe obstacle to its current operations and 0 if it reported them as a no



                                                 15
obstacle, minor obstacle, or moderate obstacle (Tax rates are a major obstacle). Note that even

through SMEs and large firms may face similar tax rates, the burden they impose to SMEs vs large

firms may be different. The data source is ES.

       Apart from taxes, there may be differences in enforcement or in the ability of the firms to

deal with corruption, power outages, and other such factors. We account for the overall quality of

the business environment by using the average severity level (on a 0–4 scale) of the following as

obstacles for the firms’ operations in the pre-pandemic period: electricity, tax administration,

obtaining licenses and permits, corruption, and labor laws (Obstacles severity). The data source is

ES.

       The capacity of the firms to adapt existing products, delivery methods, and operating

methods (such as work from home) or introduce new ones may impact their performance during

the pandemic (see Clampit et al. 2022). We capture such effects through firms’ innovation and

R&D activity prior to the pandemic. That is, we use a dummy variable equal to 1 if the firm spent

on R&D activity during the three years prior to the year the ES was administered and 0 otherwise,

and a dummy variable equal to 1 if the firm introduced a new product during the three years prior

to the year the ES was administered and 0 otherwise. Both variables are for the pre-pandemic

period and taken from the ES. We show that our results do not change if we use adjustments made

by the firm after the outbreak of the pandemic in product, process, and delivery.

       We consider the firm’s legal status as a possible factor influencing its performance. A few

studies have shown that business legal forms may differ with respect to risk-sharing and risk-

limiting possibilities, which in turn impact firms’ ability to cope with external shocks (see Doś et

al. 2022). To this end, we use two dummy variables for the pre-pandemic period: one for sole




                                                 16
proprietorship firms vs. the rest, and another for partnerships or shareholding companies with non-

traded shares vs. the rest. The data source is ES.



3.2.2 Additional explanatory variables

For robustness, we experimented with a few additional variables. These variables did not have any

significant effects on the sales decline gap or on the results for the main explanatory variables

listed above. The additional variables for the pre-pandemic period and taken from the ES include:

a dummy variable equal to 1 if the firm has a woman owner and 0 otherwise; a dummy variable

equal to 1 if the firm suffered losses due to crime during the year and 0 otherwise; and the growth

rate of employment over the last three years. Additional variables for the period after the outbreak

of the pandemic taken from COV-ES include the percentage of sales made online and the

percentage of the workforce working remotely.



4. Empirical results

4.1 Mean differences between SMEs and large firms

Table 1 provides descriptive statistics and means tests of the main variables between SMEs and

large firms. The decline in sales for SMEs is higher than for large firms by 12.2 percentage points

(34.8 vs. 22.6 percentage points). The difference is significant at the 1 percent level. This is the

(unconditional) total gap in sales decline at the mean between SMEs and large firms (henceforth

“gap” or “total gap”). Figure 3 provides the kernel density plot of the sales decline for SMEs and

large firms.

       The differences in the level of the explanatory variables between SMEs and large firms

help explain the “endowment” effects in the decomposition analysis. Unless stated otherwise, all



                                                 17
the differences between SMEs and large firms discussed in this paragraph are statistically

significant at the 1 percent level. The labor productivity of SMEs is significantly lower by 0.51

log points, or about 40 percent, than that of large firms. About 67 percent of the SMEs, compared

to a significantly lower 52 percent of large firms, experienced input supply disruptions. As

expected, large firms have significantly better access to finance than small firms. That is, 48

percent of large firms have an overdraft facility, compared to 29 percent of SMEs. The severity of

the finance obstacle equals 1.45 for SMEs and a significantly lower 1.24 for large firms. In

contrast, large firms report a significantly higher level of the skills obstacle than SMEs (1.54 vs.

1.28). A significantly higher proportion of SMEs are young than large firms (21 percent vs. 11.3

percent). Compared to SMEs, large firms export significantly more of their output (19.3 percent

vs. 5.2 percent) and have significantly higher foreign ownership (18.8 percent vs. 6.2 percent).

Similarly, a significantly higher share of large firms spent on R&D than SMEs (20.5 vs. 9.3

percent) and introduced a new product or service (36.6 vs. 30.4 percent, a difference significant at

the 5 percent level).

       The levels of the remaining baseline variables are not significantly different between SMEs

and large firms. These include proportion of firms that receive government support, proportion of

firms that have a high share of women workers, proportion of firms located in the capital city,

proportion of firms that report tax rates are a major obstacle, overall obstacle severity level, and

the proportion of sole proprietorship and shareholding or partnership firms.



4.2 Differences in returns to drivers of sales decline

In addition to the level of factors, the gap also depends on the differences in the returns to the

factors between SMEs and large firms (“structural” effect). A difference in returns could be



                                                 18
because of differences in the ability to benefit from a given level of endowments or because of

unobserved or unaccounted-for determinants of sales decline, such as management quality.

       To uncover the differences in the returns to factors, we regress sales decline on the baseline

explanatory variables discussed above. Regression results are provided in Table 2 for the full

sample (column 1) and the subsamples of SMEs (column 2) and large firms (column 3). The main

findings are summarized below.

       An increase in labor productivity is associated with a smaller decline in sales, significant

at the 1 percent level, in all three samples. Large firms benefit more from higher productivity than

SMEs. That is, a 1 log point increase in labor productivity is associated with a 2.8 percentage point

smaller decline in sales for large firms and 2.1 percentage points for SMEs. Disruptions in the

supply of inputs are associated with a much bigger decline in sales in all three samples, significant

at the 1 percent level. SMEs tend to suffer more. That is, input supply disruptions lead to a 27.9

percentage point bigger decline in sales for large firms and 33.1 percentage points for SMEs.

       The relationship between the severity level of the finance obstacle and sales decline is

positive and significant at the 5 percent level for SMEs. It is positive for the full sample too, but

much smaller and statistically insignificant. For the large firm sample, the relationship is negative

and statistically insignificant. Thus, SMEs are adversely affected by a more severe finance obstacle

but not the large firms. However, having an overdraft facility reveals a different picture. Overdraft

facility helps arrest some of the decline in sales due to the pandemic, but the effect is small and

statistically insignificant in all three samples. Nevertheless, overdraft facility benefits large firms

more than SMEs. It lowers sales decline by 2.9 percentage points for large firms compared to 0.72

percentage points for SMEs. One reason for this difference could be that banks typically favor

large firms and may grant them bigger overdraft limits.



                                                  19
         An increase in the severity level of the skills obstacle is associated with a larger decline in

sales for large firms, which is significant at the 5 percent level. The impact on SME and in the full

sample is also positive, but much smaller and statistically insignificant. The decline in sales due to

a unit increase in the skills obstacle is higher by a mere 0.26 percentage points for SMEs compared

to 1.65 percentage points for large firms.

         Higher exports are associated with a significantly larger decline in sales in the full sample

and among SMEs. For large firms, the opposite result holds, although it is not significant. In short,

exporting hurts SMEs in dealing with the pandemic but not large firms. The opposite result is

obtained for foreign ownership. That is, higher foreign ownership has no significant impact on the

decline in sales in the full sample or the SME sample but has a significant positive impact on large

firms.

         Like foreign ownership, large firms are adversely affected when tax rates are a major

obstacle, and this effect is significant at the 5 percent level. However, the impact on SMEs and the

full sample is close to zero and statistically insignificant. Firms that introduce new products or

services experience a smaller decline in sales during the pandemic. However, this effect is

statistically significant only in the full sample. The impact is also quantitatively small, at around

1.5 to 2 percentage points across the three samples.

         The remaining variables do not have any significant or quantitatively large impact on the

performance of firms in any of the three samples. These variables include a high share of women

workers, young vs. old firms, government support, overall obstacles severity level, legal

organization of the firm, location in a capital city, and R&D spending. However, there are some

noticeable differences between the impact on SMEs and large firms. That is, a higher share of

women workers leads to a higher decline in sales for SMEs by 2.1 percentage points but reduces



                                                   20
it by 1.6 percentage points for large firms. Being a young firm is associated with a larger decline

in sales for SMEs by 1.9 percentage points, and a smaller decline in sales for large firms by 1.72

percentage points. Location in the capital city is associated with a smaller decline in sales for SMEs

by 0.46 percentage points and by 2.65 percentage points for large firms. Last, spending on R&D

is associated with a smaller decline in sales for SMEs by 1.46 percentage points and a larger decline

in sales by 0.21 percentage points for large firms.



5. Mean Decomposition Results

This section analyzes whether the differences in the levels and returns to the determinants of sales

decline discussed above contribute to the sales decline gap at the mean. The traditional KOB

decomposition methodology is used (see Appendix B for a summary). The main or baseline results

in this section are based on the twofold mean decomposition with the large firms as the reference

group (see equation 4 in Appendix B). Twofold decomposition results with SMEs and the pooled

sample as the reference groups and threefold decomposition results are also provided as robustness

checks in Sections 5.2 and 5.3, respectively. 1 One issue is that the estimated impact of country-

industry dummies depends on which country-industry pair is “omitted”. To overcome this

indeterminacy, we follow the “normalization” strategy of Yun (2005), which involves using

deviations of the country-industry dummies from their grand mean. This is equivalent to averaging

all possible estimates of the impact of country-industry dummies with permuting reference or

omitted group.



5.1 Baseline results


1
 Without much loss of generality, throughout the remaining sections, only the combined effect of R&D and new
product or process development is reported.

                                                    21
The baseline mean KOB decomposition results are provided in Table 3. The “endowment” effects

are provided in column 2 and the “structural” effects in column 3. Note that a positive contribution

of a variable to the gap in sales decline means that the variable widens the gap, and a negative

contribution narrows the gap.

       As stated above, the decline in sales among SMEs is on average 12.2 percentage points

higher than that of large firms. This is the total gap. The aggregate “endowment” effect, defined

as the sum of all the individual “endowment” effects, is positive and significant at the 1 percent

level. Thus, “endowment” effects widen the gap by 7.96 percentage points, or about 65.1 percent

(7.96*100/12.22) of the total gap. Likewise, the aggregate “structural” effect is positive and

significant at the 1 percent level. Thus, structural effects widen the total gap by 4.26 percentage

points, or about 34.9 percent of the total gap. We now analyze the contributions of the individual

factors to the total gap. Figure 4 shows the main results that are discussed below.



5.1.1 Endowment effects

The biggest individual contribution via the “endowment” effect comes from the input supply

disruptions. We previously documented that input supply disruptions are more prevalent among

SMEs than large firms and that such disruptions lead to a significantly larger decline in sales. As

a result, the difference in the incidence of input supply disruptions between SMEs and large firms

widens the gap by 4.2 percentage points, or about 34.7 percent of the total gap. This effect is

significant at the 1 percent level. In other words, about one-third of the sales decline gap between

SMEs and large firms can be eliminated by improving the supply of inputs for the SMEs.

       We find that the differences in the distribution of SMEs and large firms across country-

industry pairs widen the gap by about 2.7 percentage points, or about 22.4 percent of the total gap.



                                                22
This effect is significant at the 5 percent level. Thus, relative to large firms, SMEs tend to be

disproportionately located in country-industry pairs that suffered a larger decline in sales due to

the pandemic.

        Recall that large firms have higher labor productivity than SMEs, and a higher labor

productivity leads to a smaller decline in sales. It follows that the difference in the level of labor

productivity between SMEs and large firms widens the gap by 1.4 percentage points, or about 11.7

percent of the total gap, which is significant at the 1 percent level.

        We also found that large firms have more foreign ownership than SMEs. Further, the

impact of foreign ownership on sales decline for large firms, which is used for evaluating the

“endowment” effect, is positive. Likewise, large firms report a higher severity of the skills

obstacle, and a higher skills obstacle is associated with a significant positive impact on sales

decline for large firms. Thus, the difference in the level of foreign ownership between SMEs and

large firms narrows the gap by 0.55 percentage points, or about 4.5 percent of the total gap. This

effect is significant at the 5 percent level. Similarly, the difference in the level of skills obstacle

between SMEs and large firms narrows the gap by about 0.44 percentage points, or about 3.6

percent of the total gap. This effect is significant at the 5 percent level.

        The remaining variables do not contribute significantly to the total gap via the

“endowment” effect.



5.1.2 Structural effects

The biggest significant individual contribution via the “structural” effect comes from the country-

industry dummies. Differences in the returns to country-industry dummies widen the gap by about

4.9 percentage points, or about 39.7 percent of the total gap. This effect is significant at the 1



                                                   23
percent level. Thus, SMEs benefit less or suffer more from the country-industry specific factors

than large firms.

        Recall that large firms benefit more from an increase in labor productivity than SMEs. As

a result, we find that “structurally,” labor productivity widens the gap by about 7.4 percentage

points, or about 60.2 percent of the total gap. While this is a quantitatively large effect, it is not

statistically significant. In the next sections, we show that there are several cases (regions and

quantiles) where the impact of labor productivity is large and significant.

        The next biggest contribution via the “structural” effect comes from input supply

disruptions. Recall that the impact of input supply disruptions on sales decline is bigger for SMEs

than for large firms. This difference widens the gap by about 3.5 percentage points, or about 28.7

percent of the total gap, which is significant at the 5 percent level.

        Recall from our results that a more severe finance obstacle leads to a significantly bigger

decline in sales for SMEs but not for large firms. This difference in returns to the finance obstacle

widens the gap by 2.6 percentage points, or about 20.9 percent of the total gap. This effect is

significant at the 5 percent level.

        As previously noted, SMEs with a high share of women workers suffered a larger decline

in sales during the pandemic. The opposite result was found for large firms. While neither of these

effects was statistically significant, their net impact on the total gap is large and significant. That

is, the difference in the returns to high share of women workers between SMEs and large firms

widens the gap by 2.2 percentage points, or about 18.3 percent of the total gap. This effect is

significant at the 10 percent level.

        As previously documented, a more severe skills obstacle led to a much larger decline in

sales for large firms than SMEs. This difference in the return to skills obstacle narrows the gap by



                                                  24
about 1.8 percentage points, or about 14.5 percent of the total gap. However, this effect is not

statistically significant.

        We also found that among large firms, the decline in sales was significantly larger for firms

that reported tax rates as a major or higher obstacle. No effect of the obstacle was found on SMEs.

Thus, the difference in returns to tax rates as a major constraint between SMEs and large firms

narrows the total gap by 1.5 percentage points, or about 12.2 percent of the total gap. This effect

is significant at the 5 percent level.

        The last statistically significant “structural” effect is due to exports. We documented that a

higher level of exports leads to a larger decline in sales for SMEs but not for large firms. Thus,

“structurally,” exports widen the gap by 0.39 percentage points, or about 3.2 percent of the total

gap, which is significant at the 10 percent level.

        The remaining variables do not contribute significantly “structurally” to the gap. However,

like labor productivity, a few of them do make a quantitatively noticeable contribution. That is, as

a percentage of the total gap, being a shareholding company or partnership narrows the gap by

10.5 percent; being a young firm widens the gap by 6.3 percent; having a location in the capital

city widens the gap by 5.7 percent; and having an overdraft facility widens the gap by 5.1 percent.



5.2 SME and pooled samples as the reference group

For robustness, we repeat the mean decomposition results above using SMEs as the reference

group (see equation 5 in Appendix B) and the full or pooled sample (see equation 6 in Appendix

B). These decomposition results are provided in Table A4 in Appendix A. Columns 1 to 3 contain

the baseline decomposition results discussed above for an easy comparison, columns 4-6 contain

the decomposition results with SMEs as the reference group, and columns 7-9 contain the results



                                                 25
with the pooled sample as the reference group. As is evident from Table A4, most of the baseline

results discussed remain intact qualitatively when we use SME and pooled samples as the reference

groups. A few noticeable changes are as follows.

       First, the contribution of input supply disruptions to the total gap via the “endowment”

effect is larger when the reference group is SMEs (5.0 percentage points) and the pooled sample

(4.8 percentage points) compared to the baseline results above (4.2 percentage points). In contrast,

its contribution via the “structural” effect is lower but statistically significant with SME and pooled

samples as the reference group (2.7 and 2.9 percentage points, respectively) than the baseline

results above (3.5 percentage points). Second, exporting makes a much larger and statistically

significant contribution to the total gap via the “endowment” and “structural” effects with SMEs

and pooled samples as the reference groups than in the baseline model. Third, foreign ownership

and skills obstacle contributed significantly to the total gap via the “endowment” effect in the

baseline results. Both effects are much smaller and statistically insignificant with the SME and the

pooled samples as the reference groups.



5.3 Threefold mean decomposition results

The threefold decomposition (see equation 7 in Appendix B) results are provided in Table A5 in

Appendix A. Columns 1 to 3 contain the baseline decomposition results discussed above for an

easy comparison. Columns 4–8 contain the threefold decomposition results. In the threefold

decomposition, the aggregate “interaction” effect is small and statistically insignificant. Results

for the individual factors are roughly the same as for the baseline model above qualitatively.



5.4 Additional controls



                                                  26
The mean decomposition results with the additional controls are provided in Table A6 in Appendix

A. None of the additional controls contribute significantly to the total gap. The baseline results

discussed above are roughly unchanged except that the contribution of country-industry fixed

effects “structurally” is much smaller and statistically insignificant. However, these fixed effects

continue to contribute via the “endowment” effect significantly.



6. Quantile decomposition results

In this section, we provide the quantile decomposition results based on the approach of Firpo et al.

(2007, 2009). A summary of the methodology is provided in Appendix B.



6.1 Aggregate quantile decomposition results

The results for the total gap, aggregate “endowment” effects, and aggregate “structural” effects at

the 10th to 90th quantiles are provided in Table 4. As can be seen in Panel A of the table, the

(unconditional) total gap is positive and significant (at the 1 percent level) at all the quantiles. That

is, SMEs suffered a significantly larger decline in sales during the pandemic than large firms at all

the quantiles. However, the gap is much larger at higher quantiles (above 30th quantile) than lower

quantiles. The mean gap at the 10th, 20th, and 30th quantiles is 7 percentage points, and 16.6

percentage points between the 40th and 90th quantiles. In short, SMEs underperform large firms

much more when the decline in sales is relatively high or the crisis is more severe (also see figure

2).

        Panel B in Table 4 provides the aggregate “endowment” effect and the “structural” effect.

Panel C expresses these aggregate effects as a percentage of the total gap at the quantile. The main

findings are as follows. First, as with the mean decomposition (Table 3), the aggregate



                                                   27
“endowment” effect is positive and statistically significant at all the quantiles. The aggregate

“structural” effect is positive and significant at all the quantiles except the 50th and 80th (positive

but insignificant) and the 90th quantile (negative but insignificant). Thus, overall, the difference in

both the endowments and returns to factors widens the gap at most quantiles.

       Second, as for the total gap, both the aggregate “endowment” and “structural” effects

increase in value as we move from lower to higher quantiles, although not monotonically.

However, the aggregate “endowment” effect rises proportionately more than the aggregate

“structural” effect (see Panel C in Table 4). Thus, the aggregate “endowment” effect accounts for

a smaller share of the total gap at lower quantiles (below the 50th quantile) than at higher quantiles.

The opposite holds for the aggregate “structural” effect. For instance, from Panel C in Table 4, we

see that between the 10th and 40th quantiles, the aggregate “structural” effect accounts for on

average about 62 percent of the total gap, and only 41 percent between the 50th and 90th quantiles.

The remaining part of the gap is accounted for by the aggregate “endowment” effect. In other

words, for small shocks or sales declines, SMEs underperformed large firms, mainly due to the

latter’s better use of the existing endowments. However, for a large shock, SMEs underperformed

large firms, mostly because the latter had more endowments or resources.



6.2 Decomposition results for individual factors

Quantile decomposition results for the individual factors are provided in Table 5. Table 6 expresses

the estimates in Table 5 as a percentage of the total gap at each quantile. In both these tables, Part

A contains the “endowment” effects and Part B the “structural” effects.



6.2.1 Individual endowment effects



                                                  28
From Part A in tables 5 and 6, the main individual “endowment” effects are as follows. First, input

supply disruptions widen the gap significantly at all the quantiles. However, the impact, both

absolutely and relative to the total gap, is much larger at the intermediate quantiles. That is, on

average, input supply disruptions widen the gap by 6.2 percentage points, or about 41.3 percent of

the total gap, at the 40th, 50th, and 60th quantiles. They widen the gap by 1.6 percentage points, or

22.6 percent of the total gap, below the 40th quantile and by 4.2 percentage points, or 23.5 percent

of the total gap, above the 60th quantile.

       Second, in the baseline mean decomposition results, we found that differences in labor

productivity between SMEs and large firms widened the gap by about 1.4 percentage points, or

11.7 percent of the mean total gap. This effect rises, absolutely and relative to the total gap, almost

monotonically with the quantiles. It ranges between 0.127 percentage points, or 1.7 percent of the

total gap (30th quantile), and 3.8 percentage points, or 20.9 percent of the total gap (90th quantile).

These contributions to the gap are statistically insignificant at the 30th quantile and below and

significant above it.

       Third, the impact of the country-industry fixed effects on the total gap via the “endowment”

effect is mostly confined to the higher quantiles. From the 10th to the 40th quantiles, country-

industry fixed effects widen the gap by 3.5 to 9.3 percent of the total gap. However, these effects

are statistically insignificant. For the remaining quantiles (50th to 90th), country-industry fixed

effects widen the gap significantly by 23.5 to 73.6 percent of the total gap.

       Fourth, the baseline mean decomposition results above showed no significant contribution

of the overdraft facility via the “endowment” effect. The quantile decomposition results show that

having an overdraft facility widens the gap via the “endowment” effect at the 10th, 20th, and 30th

quantiles significantly. It does so by about 7.9 percent of the total gap on average. Above the 30th



                                                  29
quantile, overdraft facility has a much smaller and statistically insignificant impact on the total gap

(averaging 2.8 percent of the total gap).

        In the baseline mean decomposition results, the skills obstacle narrowed the gap via the

“endowment” effect by 3.6 percent of the total gap. Across quantiles, we find that the skills

obstacle narrows the gap, and significantly so, at low and high quantiles (20th, 30th, 70th, 80th, and

90th). The impact at intermediate quantiles (40th, 50th, and 60th) is much smaller and statistically

insignificant.



6.2.2 Individual structural effects

Regarding the “structural” effect, the country-industry fixed effects were the biggest contributors

to the total gap in the baseline mean decomposition results. The quantile decomposition results

reveal that country-industry fixed effects make a significant contribution “structurally” to the total

gap at all the quantiles. Further, as a percentage of the total gap, these fixed effects narrow the total

gap at the 10th and 20th quantiles by 172 percent and 231 percent, respectively; widen the gap at

the 30th to 90th quantiles by 210 to 1071 percent. Thus, there is a large variation in how the country-

fixed effects impact the total gap across the different quantiles.

        Differences in the returns to labor productivity have varied effects at different quantiles.

They narrow the gap at lower quantiles (10th to 30th quantiles) and widen it thereafter. Except at

the 40th quantile, the “structural” contribution of the variable as a percentage of the total gap at the

various quantiles is large and above 48 percent. However, these contributions are statistically

significant only at the 30th and 80th quantiles.

        As a percentage of the total gap, the difference in returns to input supply disruptions

significantly widens the gap by 40 and 57 percent, respectively, at the 10th and 20th quantiles and



                                                   30
by 414 percent at the 30th quantile. It narrows the gap significantly by 23.6 percent and 37 percent

at the 40th and 50th quantiles, respectively. Above the 50th quantile, the impact is statistically

insignificant.

        Having an overdraft facility “structurally” widens the gap significantly by 15.2 and 14.2

percent of the total gap at the 10th and 20th quantiles, respectively. However, above the 20th

quantile, the contribution of the overdraft facility “structurally” is statistically insignificant and

ranges between close to 0 percent and 7 percent of the total gap. The finance obstacle “structurally”

widens the gap by 14 to 44 percent of the total gap at the 70th to 90th quantiles. These effects are

statistically significant at 10 percent or 5 percent levels. There is no significant impact of the

variable “structurally” at the remaining quantiles. Thus, “structurally,” the finance obstacle widens

the gap between SMEs and large firms when the sales decline shock is relatively severe.

        The difference in the returns to being young widens the gap significantly by 7.9 to 15.2

percent of the total gap at the 60th to 80th quantiles. The contribution of the variable at the remaining

quantiles is statistically insignificant and typically much smaller. The tax obstacle “structurally”

narrows the gap significantly at all the quantiles except the 40th, 80th, and 90th quantiles. This effect,

however, is much stronger at lower quantiles than at higher quantiles. It ranges from 22.6 to 27.4

percent of the total gap between the 10th and 30th quantiles and from about 0 to 15.7 percent at the

remaining quantiles. Like the tax obstacle, the skills obstacle “structurally” significantly narrows

the gap by 21.6 to 25.5 percent of the total gap at the 10th to 30th quantiles. It narrows the gap by

17.7 percent of the total gap at the 70th quantile and 20 percent at the 80th quantile, but only the

former is statistically significant.

        Some other variables have a significant impact “structurally” at select quantiles. That is,

government support widens the gap significantly by 14.3 percent of the total gap at the 30th



                                                   31
quantile, and R&D and innovation activity (combined) narrow the gap by 9.8 percent of the total

gap at the 40th quantile.

        To summarize, the quantile decomposition results reveal substantial variation in the

contribution of the variables to the sales decline gap via the “endowment” and “structural” effects.

Most variables contribute significantly across the sales decline quantiles. The few variables that

do not contribute significantly either “structurally” or via the “endowment” effect at any quantile

are having a higher share of women workers, being in the capital city, overall obstacles severity

level, and the legal form of the firm (sole proprietorship and shareholding/partnership firms). In

the next section, we will show that the effect of these variables is restricted to specific regions.




7. Regional results

This section analyzes the sales decline gap across groups of countries. Unless stated otherwise,

the discussion below relates to the mean decomposition. All “percentage” contributions are as a

percentage of the total gap at the mean or the relevant quantile.



7.1 Total sales decline gap

A significantly poorer resilience of SMEs relative to large firms is observed in all the four regions

(Table 7). The mean gap is the smallest in MENA, equal to 6.2 percentage points. It is roughly the

same in all the other regions, equaling 11.4 percentage points in LAC and SSA, and 11.2

percentage points in ECA. 2 All these gaps are significant at the 1 percent level. Likewise, the gap




2
 However, do note that the formal private sector in the MENA region has a limited role in the presence of an
overbearing public sector (see Islam et al. 2022).


                                                    32
is positive, large, and significant at most of the quantiles in each region (see Table A7 in Appendix

A).

       In the full sample, we found that the gap was larger at higher quantiles. This result holds

for ECA and LAC. In MENA and SSA, there is no noticeable and robust increase or decrease in

the total gap as we move from lower to higher quantiles. Thus, relative to large firms, SMEs are

less resilient to large downturns than small ones in ECA and LAC, but not in MENA and SSA.



7.2 Aggregate endowment vs. structural effects

Figure 5 shows the contributions of the aggregate “endowment” and “structural” effects to the total

gap. Interesting differences emerge between the regions. Both effects are significant in SSA,

widening the gap by roughly equal amounts. In LAC and MENA, the aggregate “endowment”

effect is dominant, widening the gap significantly. While noticeable, the aggregate “structural”

effect is statistically insignificant. In contrast, the opposite holds for the ECA region. See Table 7

for more details. Within each region, there are substantial differences across the sales decline

distribution (see Table A7 in Appendix A). For instance, in SSA, the aggregate “endowment”

effect is significant at 4 of the 9 quantiles, and it ranges between -6.7 percent of the total gap and

159.4 percent. The differences within and across regions highlight the need for a proper targeting

of policies aimed at improving SMEs’ resilience to economic shocks.



7.3 Individuals factors

Next, we consider the individual factors in the different regions. These are provided in Table 8

(mean decomposition) and Tables A8 to A11 in Appendix A (quantile decomposition). Due to

space limitations, the discussion is restricted to a few important results.



                                                  33
       Input supply disruptions significantly widen the gap via the “endowment” effect at the

mean and at most quantiles in each region. However, as a percentage of the total gap, the impact

at the mean and on average across quantiles is much larger in LAC and MENA (93 percent of the

total gap on average when significant) than in SSA (31 percent) and ECA (25.5 percent).

“Structurally,” input supply disruptions significantly affect the gap at one or more quantiles in all

the regions. In SSA and LAC, the “structural” effect of the variable is to significantly widen the

gap at some of the relatively low quantiles and narrow it at some of the higher quantiles. A similar

result holds for MENA, except that the narrowing of the gap at the higher quantiles is not

significant. In ECA, “structurally,” input supply disruptions widen the gap significantly at the

lowest as well as the highest quantiles and narrow it at intermediate quantiles. Several important

policy implications follow from these results. For example, closing the gap can be more effectively

achieved in ECA by improving the “returns” to supply chains for firms at the low and high end of

the sales decline distribution. In contrast, in SSA, it is important to focus on firms at the low end

of the distribution rather than at the high end.

       Differences in the level of the finance obstacle between SMEs and large firms

(“endowment” effect) contribute significantly to the gap only in LAC. They do so by widening the

gap at the 20th, 60th, and 70th quantiles by 17.3 to 21 percent of the total gap. “Structurally,” the

finance obstacle contributes significantly to the total gap in all the regions, but only at select

quantiles. It narrows the gap in LAC (by 43.4 percent of the total gap at the 20th quantile) and

widens it in all the other regions. MENA stands out with the very large “structural” effects of the

finance obstacle, which widen the gap at the mean by 146 percent and at the 60th to 90th quantiles

by 209 to 793 percent.




                                                   34
        Differences in the level of labor productivity between SMEs and large firms widen the gap

significantly at the mean and/or some of the quantiles in SSA (by 14.5 to 49.4 percent of the total

gap) and LAC (by 20 to 236.5 percent), but not in ECA and MENA. “Structurally”, labor

productivity significantly widens the gap in all the regions at one or more quantiles. While the

impact varies substantially across the quantiles within a region, when significant, it is quite large

as a proportion of the total gap. For instance, in SSA, the significant “structural” effect occurs at

the 40th to 70th quantiles, which widen the gap by 261 to 355 percent, and in MENA by 635 to

2,826 percent between the 60th and 90th quantiles.

        Recall that in the full sample, the skills obstacle significantly narrowed the gap

“structurally” by at most 26 percent of the total gap at some of the quantiles. This overall impact

is somewhat muted likely because of contrasting effects at the regional level. Focusing on the

statistically significant “structural” effects at the various quantiles, the skills obstacle narrows the

gap in SSA by 33 to 34 percent of the total gap, in LAC by 41 percent or more, and in ECA by 20

to 136 percent. In contrast, in MENA, it significantly widens the gap by 685 percent at the 90th

quantile.

        In the full sample, we found that government support did not reduce the gap between SMEs

and large firms. In fact, government support widened the gap significantly at some quantiles (by

14.3 percent “structurally” at the 30th quantile). A qualitatively similar result obtains in ECA, LAC,

and SSA. However, in MENA, government support significantly narrows the gap “structurally”

and via the “endowment” effect. For example, “structurally”, the gap is smaller at the mean by

21.3 percent of the total gap and at several quantiles by 40.5 to 67.1 percent. Thus, MENA stands

out as the only region where government support helps SMEs more than large firms. This is an




                                                  35
important result from a policy-targeting point of view if we assume that SMEs are more in need

of government support.

       There are several other interesting results. We provide a few examples focusing on the

statistically significant results. First, in MENA, R&D and innovation activity significantly widen

the mean gap by 14.9 percent via the “endowment” effect and 15.3 percent “structurally.” There

is no significant impact of the variable on the mean gap in any of the other regions. By quantiles,

R&D and innovation significantly widen the gap in MENA “structurally” at 5 out of the 9

quantiles. In all the other regions, the variable significantly widens (in SSA) or narrows (in LAC

and ECA) the gap at one or two quantiles. In short, R&D activity plays a much bigger role in

MENA in widening the gap between SMEs and large firms than in the other regions. Second, in

the full sample, we did not find any significant effects of the legal form of the firm on the total

gap. This changes at the regional level. For instance, being a sole proprietorship has a significant

impact on the total gap at the mean or one or more quantiles in all the regions. To provide an

example, in SSA, sole proprietorship narrows the mean gap by 40 percent via the “endowment”

effect and widens it by 41 percent “structurally.” Both effects are significant at the 5 percent level.

Third, as in the full sample, country-industry fixed effects contribute significantly to the total gap

at the mean and/or at one or more quantiles in each region. Fourth, looking at all the explanatory

variables one-by-one, we find that most of them contribute significantly to the total gap either at

the mean or at one or more quantiles in all the regions. The exceptions are as follows: high share

of women workers in LAC, foreign ownership in ECA, overdraft facility in LAC, capital city in

ECA, tax obstacle in SSA, obstacles severity in ECA, and shareholding company or partnerships

in LAC and ECA. Last, MENA stands out with a much larger and more widespread (across

quantiles) impact of foreign ownership on the total gap than the other regions. Foreign ownership



                                                  36
significantly narrows the gap in MENA at the 20th to 50th quantiles by 24.8 to 36.1 percent of the

total gap via the “endowment” effect and 10.7 to 16.5 percent “structurally”. In contrast, foreign

ownership significantly narrows the gap in SSA and LAC only at the 50th quantile by about 8.8

and 11.1 percent, respectively. There is no significant impact of the variable in ECA.



8. Conclusion

The key questions that this study investigates are whether SMEs suffered a larger decline in sales

due to the COVID-19 pandemic than large firms, and if so, by how much and what are the

contributors that exacerbate or alleviate the gap between the two. We ascertain a significantly

larger decline in sales for SMEs than large firms. This gap is widespread, as it exists across the

entire sample, within geographical regions, and in most individual countries. Factors that

contribute to the gap are identified. This is done at the mean and various quantiles of the sales

decline distribution. We identified 10 explanatory variables that contribute significantly to the gap

at either the mean or one or more quantiles in all the regions, and another six variables that

contribute significantly in three out of four regions.

       Our results are important to policy makers in several ways. First, they suggest that SMEs

are more vulnerable to crisis situations. As a result, additional policy measures may be required to

help SMEs overcome economic downturns, natural disasters, armed conflicts, infectious disease

outbreaks, and climate change. Second, because of their potential for creating jobs and inclusive

growth, several countries have implemented national policies to promote SMEs. By showing that

SMEs may suffer more during a crisis, our findings indicate that the development of SMEs comes

at a cost in terms of stability. Thus, support for SMEs must note the potential tradeoffs.




                                                  37
       Third, we identified and estimated the contribution of several factors to the decline in sales

for SMEs relative to large firms during the pandemic. These factors highlight the specific

challenges faced by SMEs during the pandemic and possibly in future shocks. Thus, they are prime

candidates for policy intervention following a crisis. For instance, we found that lower labor

productivity and a high incidence of input supply disruption result in poorer performance by SMEs

relative to large firms. Thus, policies that improve SMEs’ labor productivity and input supply

networks can help them close the gap with large firms currently and in future crises. Fourth, our

results indicate that a substantial part of the gap between SMEs and large firms can be closed by

providing more resources to SMEs. This is encouraging, as it is typically easier to provide more

resources than improve the returns to factors for a particular group. Of course, where possible,

policies can also help SMEs by enabling them to use their resources better.

       Fifth, our results show that policies that are effective in one region may be ineffective or

less effective in other regions. As a result, an eclectic approach is needed that combines the general

findings in the literature with the prevailing regional or local factors. This poses a significant

challenge to policy makers and researchers to identify the relevant local conditions and their

implications for optimal policies. Sixth, there are important differences in the size of the gap and

its drivers across the quantiles of the sales decline distribution. Thus, policy effectiveness can be

improved through proper targeting of the relevant quantiles. Last, governments around the globe

provided support to the private sector during the pandemic. Our results show that government

support helped SMEs significantly more than large firms only in MENA. The opposite holds for

ECA, LAC, and SSA. This is a worrying result if we assume that SMEs suffered more and are

more in need of government support. Thus, a review of existing policies, their targeting, and their

delivery to SMEs may be helpful.



                                                 38
       There are several avenues for future research. First, the analysis can be extended to the

impact of the pandemic on other performance measures, such as employment, product and process

innovation, use of online platforms, and exports. Second, extending our results to countries not in

our sample will be a fruitful exercise. Third, identifying the mechanisms through which the

identified factors affect the performance of SMEs and large firms could further help in the design

and targeting of policies for SMEs. Fourth, heterogeneous effects by various groups, such as young

vs. old firms and countries with weak vs. strong rule of law, are another avenue for future research.

Finally, future researchers can explore the causal effects of the factors we identified as important.




                                                 39
References
Adams-Prassl, A., T. Boneva, M. Golin, and C. Rauh (2020). Inequality in the impact of the
coronavirus shock: Evidence from real time surveys. Journal of Public Economics, 189: 104245.

Amin, M. and D. Viganola (2023). Does better access to finance help firms deal with the COVID-
19 pandemic? Evidence from firm-level survey data. Journal of International
Development, 1– 31. https://doi.org/10.1002/jid.3787

Ayyagari, M., A. Demirguc-Kunt, and V. Maksimovic (2014). Who creates jobs in developing
countries? Small Business Economics 43 (1): 75-99. https://doi.org/10.1007/s11187-014-9549-5

Ayyagari, M., A. Demirguc-Kunt, and V. Maksimovic (2011). Small vs. young firms across the
world: Contribution to employment, job creation, and growth. Policy Research Working Paper No.
5631, World Bank, Washington, DC. https://doi.org/10.1596/1813-9450-5631

Bak, O., S. Shaw, C. Colicchia and V. Kumar, "A Systematic Literature Review of Supply Chain
Resilience in Small–Medium Enterprises (SMEs): A Call for Further Research," in IEEE
Transactions on Engineering Management, vol. 70, no. 1, pp. 328-341, Jan. 2023, doi:
10.1109/TEM.2020.3016988

Bargain, O. and U. Aminjonov (2021). Poverty and COVID-19 in Africa and Latin
America. World Development, 142 (June): 105422.

Belghitar, Y., A. Moro, A. and N. Radić (2022). When the rainy day is the worst hurricane ever:
the effects of governmental policies on SMEs during COVID-19. Small Business Economics 58,
943–961. https://doi.org/10.1007/s11187-021-00510-8

Blinder, A. S. (1973). Wage discrimination: Reduced form and structural estimates. Journal of
Human Resources, 8(4): 436–455.

Bloom, N., P. Bunn, P. Mizen, P. Smietanka, and G. Thwaites (2023). The impact of COVID-19
on productivity. The Review of Economics and Statistics. https://doi.org/10.1162/rest_a_01298

Bloom, N., R. S. Fletcher, and E. Yeh (2021). The impact of COVID-19 on US firms. National
Bureau of Economic Research Working Paper No. 28314.

Bluedorn, J., F. Caselli, N.-J. Hansen, I. Shibata, and M. M. Tavares (2023). Gender and
employment in the COVID-19 recession: Cross-Country evidence on “She-Cessions. Labour
Economics, 81: 102308.

Bundervoet, T., M. E. Dávalos, and N. Garcia (2022). The short-term impacts of COVID-19 on
households in developing countries: An overview based on a harmonized dataset of high-
frequency surveys. World Development, 153(May): 105844.

Bruhn, M., A. Demirguc-Kunt, and D. Singer (2023). Competition and firm recovery post-
COVID-19. Small Business Economics. https://doi.org/10.1007/s11187-023-00750-w


                                              40
Butt, A. S. (2021). Understanding the implications of pandemic outbreaks on supply chains: An
exploratory study of the effects caused by the COVID-19 across four South Asian countries and
steps taken by firms to address the disruptions. International Journal of Physical Distribution &
Logistics Management, 52(4): 370–392. https://doi.org/10.1108/IJPDLM-08-2020-0281

Chan, C. M. L., S. Y. Teoh, A. Yeow, and G. Pan (2019). Agility in responding to disruptive digital
innovation: Case study of an SME. Information Systems Journal, 29(2): 436–455.
https://doi.org/10.1111/isj.12215.

Chang, D., X. Chang, H. Yu, and K. J. K. Tan (2022). The determinants of COVID-19 morbidity
and mortality across countries. Scientific Reports, 12: 5888 https://doi.org/10.1038/s41598-022-
09783-9

Chen, S. and D. Lee (2023). Small and vulnerable: SME productivity in the great productivity
slowdown. Journal of Financial Economics, 147(1): 49-74.

Cirera, X., M. Cruz, E, Davies, A. Grover, L. Iacovone, J. E. L. Cordova, D. Medvedev, F. O.
Maduko, G. Nayyar, S.R. Ortega, and J. Torres (2021). Policies to support businesses through the
COVID-19 shock: A firm level perspective. The World Bank Research Observer, 36(1): 41-66.

Clampit, J. A., M. P. Lorenz, J. E. Gamble, and J. Lee (2022). Performance stability among small
and medium-sized enterprises during COVID-19: A test of the efficacy of dynamic
capabilities. International Small Business Journal, 40(3): 403-419.

Dean, T. J., R. L. Brown, and C. E. Bamford (1998). Differences in large and small firm responses
to environmental context: Strategic implications from a comparative analysis of business
formations. Strategic Management Journal, 19(8): 709–728. https://doi.org/10.1002/(SICI)1097-
0266(199808)19:8%3C709::AID-SMJ966%3E3.0.CO;2-9

del Rio-Chanona, R. M., P. Mealy, A. Pichler, F. Lafond, and J. D. Farmer (2020). Supply and
demand shocks in the COVID-19 pandemic: An industry and occupation perspective. Oxford
Review of Economic Policy, 36: S94-S137.

Demir, B. and B. Javorcik (2020). Trade finance matters: Evidence from the COVID-19 crisis.
Oxford Review of Economic Policy, 36(S1): S397-S408.

Ding, W., R. Levine, C. Lin, and W. Xie (2021). Corporate immunity to the COVID-19
pandemic. Journal of Financial Economics, 141(2): 802-830.

Dörr, J., G. Licht, and S. Murmann (2022). Small firms and the COVID-19 insolvency gap. Small
Business Economics, 58: 887–917. https://doi.org/10.1007/s11187-021-00514-4

Doś, A., M. Wieczorek-Kosmala, and J. Błach (2022). The effect of business legal form on the
perception of COVID-19-related disruptions by households running a business. Risks, 10(4): 82.
https://doi.org/10.3390/risks10040082


                                                41
Essers, D., K. Megersa, and M. Sanfilippo (2021). The productivity gaps of female-owned firms:
Evidence from Ethiopian census data. Economic Development and Cultural Change, 69(2): 645-
683.
Fairlie, R., and F. M. Fossen (2022). The early impacts of the Covid-19 pandemic on business
sales. Small Business Economics 58: 1853–1864.

Fernández-Cerezo, A., B. Gonzalez, M. I. Peinado, and E. Moral-Benito (2022). Firm-level
heterogeneity in the impact of the COVID-19 pandemic. Applied Economics
DOI: 10.1080/00036846.2022.2133894

Firpo, S., N. Fortin, and T. Lemieux (2007). Decomposing wage distributions using recentered
influence function regressions. Working paper, University of British Columbia.
——— (2009). Unconditional quantile regressions. Econometrica, 77(3): 953–73.


Forsythe, E., L. B. Kahn, F. Lange, and D. Wiczer (2022). Where have all the workers gone?
Recalls, retirements, and reallocation in the COVID recovery. Labour Economics, 78(October):
102251.

Gertler, M., and S. Gilchrist (1994). Monetary policy, business cycles, and the behavior of small
manufacturing firms. The Quarterly Journal of Economics, 109(2): 309–340.

Gu, X., S. Ying, W. Zhang, and Y. Tao (2020). How do firms respond to COVID-19? First
evidence from Suzhou, China. Emerging Markets Finance and Trade, 56(10): 2181-2197.

Guerrero-Amezaga, M. E., J. E. Humphries, C. A. Neilson, N. Shimberg, and G. Ulyssea (2022).
Small firms and the pandemic: Evidence from Latin America. Journal of Development Economics,
155(March): 102775.

Herskovic, B., B. Kelly, H. Lustig, and S. Van Nieuwerburgh (2020). Firm volatility in granular
networks. Journal of Political Economy, 128(11): 4097-4162.

Hu, S. and Y. Zhang (2021). COVID-19 pandemic and firm performance: Cross-country
evidence. International Review of Economics & Finance, 74 (July): 365-372.
https://doi.org/10.1016/j.iref.2021.03.016

ILO (2019). Small matters: Global evidence on the contribution to employment by the self-
employed, micro-enterprises and SMEs. International Labor Organization (ILO).
https://www.ilo.org/wcmsp5/groups/public/---dgreports/---dcomm/---
publ/documents/publication/wcms_723282.pdf

ILO and GIZ (2013). Is small still beautiful? Literature review of recent empirical evidence on the
contribution of SMEs to employment creation (Geneva and Bonn). International Labor
Organization (ILO) and Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ).

                                                42
https://www.ilo.org/employment/Whatwedo/Publications/employment-
reports/WCMS_216909/lang--en/index.htm

Islam, A. M. and M. Amin (2023). The gender labor productivity gap across informal firms, World
Development, 167: 106229.


Islam, A. M, D. Moosa, and F. Saliola (2022). Jobs undone: Reshaping the role of governments
toward markets and workers in the Middle East and North Africa. Washington DC: World Bank.
https://www.worldbank.org/en/region/mena/publication/jobs-undone-reshaping-the-role-of-
governments-toward-markets-and-workers-in-the-middle-east-and-north-africa

Iyer, R. J.-L. Peydro, S. da-Rocha-Lopes, and A. Schoar (2014). Interbank liquidity crunch and
the firm credit crunch: Evidence from the 2007–2009 Crisis. The Review of Financial Studies,
27(1): 347–372. https://doi.org/10.1093/rfs/hht056

Kitagawa, E. M. (1955). Components of a difference between two rates. Journal of the American
Statistical Association, 50(272): 1168–1194.

Kozeniauskas, N., P. Moreira, and C. Santos (2020). COVID-19 and firms: Productivity and
government policies. CEPR Discussion Paper 15156, Centre for Economic Policy Research.

Liu, Y., Y. Zhang, H. Fang, and X. Chen (2022). SMEs’ line of credit under the COVID-19:
Evidence from China. Small Business Economics, 58: 807–828. https://doi.org/10.1007/s11187-
021-00474-9

Liu, Y., M. W. Peng, Z. Wei, J. Xu, and L. C. Xu (2021). Organizational resources, country
institutions, and national culture behind firm survival and growth during COVID-19. Policy
Research Working Paper Series No 9633, The World Bank, Washington, DC.

Marconatto, D. A. B., E. G. Teixeira, G. A. Peixoto, and L, Faccin (2022). Weathering the storm:
what successful SMEs are doing to beat the pandemic. Management Decisions, 60(5): 1369-1386.
https://doi.org/10.1108/MD-11-2020-1507

Meier, M. and E. Pinto (2020). COVID-19 supply chain disruptions, CRC TR 224 Discussion
Paper Series, University of Bonn and University of Mannheim, Germany.

Mueller, E. F. and C. Jungwirth (2022). Are cooperative firms more agile? A contingency
perspective on small and medium-sized enterprises in agglomerations and peripheral areas. Small
Business Economics 58: 281–302 (2022). https://doi.org/10.1007/s11187-020-00410-3

Muzi, S., F. Jolevski, K. Ueda, D. Viganola (2023). Productivity and firm exit during the COVID-
19 crisis: Cross-country evidence. Small Business Economics, 60: 1719–1760.
https://doi.org/10.1007/s11187-022-00675-w




                                              43
Nix, E., E. Gamberoni, and R. Heath (2016). Bridging the gender gap: Identifying what is holding
self-employed women back in Ghana, Rwanda, Tanzania, and the Republic of Congo. World Bank
Economic Review, 30(3): 501–21.

Oaxaca, R. (1973). Male–female wage differentials in urban labor markets. International
Economic Review, 14(3): 693–709.

OECD (2019). Strengthening SMEs and entrepreneurship for productivity and inclusive growth:
OECD 2018 ministerial conference on SMEs. OECD Studies on SMEs and Entrepreneurship,
OECD Publishing, Paris. https://doi.org/10.1787/c19b6f97-en.

Ozar, S., G. Ozertan, and Z. Irfanoglu (2008). Micro and small enterprise growth in Turkey: Under
the shadow of financial crisis. The Developing Economies, 46(4): 331–362.

Peric, M., and V. Vitezic (2016). Impact of global economic crisis on firm growth. Small Business
Economics, 46: 1–12. https://doi.org/10.1007/s11187-015-9671-z

Pitschner, S. (2022). Supply chain disruptions and               labor   shortages:   COVID     in
perspective. Economics Letters, 221(December): 110895.

Pitterle, I. A. and L. Niermann (2021). The COVID-19 crisis: What explains cross-country
differences in the pandemic’s short-term economic impact? DESA Working Paper No. 174, United
Nations. https://www.un.org/en/desa/covid-19-crisis-what-explains-cross-country-differences-
pandemic%E2%80%99s-short-term-economic-impact

Russell, L. and C. Sun (2020). The effect of mandatory child care center closures on women's labor
market outcomes during the COVID-19 pandemic. Covid Economics: Vetted and Real-Time
Papers, 62: 124 -154.

Stemmler, H. (2022). The effects of COVID-19 on businesses key versus non-key firms. ILO
Working Papers, International Labour Organization.
https://www.ilo.org/legacy/english/intserv/working-papers/wp077/index.html#ID0ECC

Tan, H. and H. See (2004). Strategic reorientation and responses to the Asian financial crisis: The
case of the manufacturing industry in Singapore. Asia Pacific Journal of Management, 21: 189–
211.

United Nations (2022). The Covid-19 pandemic impact on micro, small and medium sized
enterprises: Market access challenges and competition policy. United Nations Conference on
Trade and Development, UNCTAD, UN.
https://unctad.org/system/files/official-document/ditcclp2021d3_en.pdf

Varum, C. A. and V. C. Rocha (2013). Employment and SMEs during crises. Small Business
Economics, 40(1): 9–25.




                                                44
Yun, Myeong-Su (2005). Normalized equation and decomposition analysis: Computation and
inference. No 1822, IZA Discussion Papers, Institute of Labor Economics (IZA),
https://EconPapers.repec.org/RePEc:iza:izadps:dp1822.




                                         45
                                                                                               Figure 1: Sales decline gap (mean values)
                                                                    20.0

                                                                                                                                                         17.0
                                                                                                             16.3 16.4
  Percentage decline in sales of SMEs minus percentage decline in




                                                                                                                               14.7
                                                                    15.0
                                                                                                                                                                                   13.0
                                                                                                                                                                                                 12.2
                                                                                  11.3                                                                                11.4                11.4
              sales of large firms (percentage points)




                                                                                         9.7          10.0
                                                                    10.0
                                                                                                                         7.9
                                                                                                                                             6.2                6.5
                                                                                                                                                   5.8
                                                                                                                                                                             5.3
                                                                     5.0    3.9




                                                                     0.0


                                                                                                                                      -2.0

                                                                     -5.0                      -4.1




                                                                    -10.0

Note: Sales decline gap is the decline in sales in the last month (post COVID-19 outbreak)
compared to the same month last year (pre-COVID-19 outbreak) and expressed as a percentage of
the latter.
Source: Authors’ own calculations based on data from World Bank Enterprise Surveys.




                                                                                                                               46
                                                      Figure 2: Sales decline gap at the various quantiles

                                          30


                                          25
  Sales decline gap (percentage points)




                                          20


                                          15


                                          10


                                           5


                                           0
                                               10th   20th     30th        40th         50th          60th        70th   80th   90th

                                                               Sales decline gap               95% confidence interval



Source: Authors’ own calculations based on data from World Bank’s Enterprise Surveys.
Note: Sales decline gap is the decline in sales due to the pandemic as a percentage of the pre-
pandemic sales level of SMEs minus the same for large firms. A positive value of sales decline
gap implies that the percentage decline in sales is bigger for SMEs than large firms.




                                                                                   47
                        Figure 3: Kernel density plot
   .02
   .015
Density
 .01
   .005
   0




          -300   -200           -100             0      100
                          Sales decline (%)

                                 SMEs
                                 Large firms




                                      48
                                   Figure 4: Baseline mean decomposition results

                                              Endowment effects (% of the total gap)

              Skills obstacle (0-4)
                       -3.6

 Foreign ownership (proportion)
                 -4.5
       Labor productivity (logs)                                  11.7

  Country-Industry fixed effects                                                                22.4

     Supply disruption Y:1 N:0                                                                                             34.6

        -10            -5             0          5          10           15           20           25        30           35           40


                                                Structural effects (% of total gap)

         Exports (proportion of sales)           3.2

Tax rates are a major obstacle Y:1 N:0
                         -12.2
High share of women workers Y:1 N:0                              18.3

                 Finance obstacle (0-4)                               20.9

           Supply disruption Y:1 N:0                                           28.7

              Labor productivity (logs)                                                                            60.2

        Country-Industry fixed effects                                                                                          70.6

                 -20         -10          0            10        20           30           40           50    60           70          80


Source: Authors’ own calculations based on the results in Table 3. A positive value implies that
the variable widens the sales decline gap between SMEs and large firms. A negative implies that
the variable narrows the gap.




                                                                 49
       Figure 5: Aggregate endowment and structural effects at the mean by regions

                                        130

                                        110

              % of the total mean gap    90
                                                                                     37.7
                                         70    55.7**           63.3***
                                                                                                       124.1***
                                         50

                                         30                                         62.3*
                                               44.3**             36.7
                                         10

                                        -10     SSA               ECA                LAC          -24.1 MENA

                                        -30

                                        -50

                                              Aggregate endowment effect      Aggregate structural effect


Source: Authors’ own calculations based on the results in Table 7.
*** significant at the 1% level, ** significant at the 5% level, * significant at the 10% level. See
Table 7 for more details.




                                                                         50
                                Table 1: Mean differences
                                                  (1)          (2)           (3)

                                               Mean value Mean value Difference
                                               for SMEs     for Large  (SME minus
                                                sample    firms sample Large firms)
Sales decline (%)                                34.77        22.55       12.22***
Labor Productivity (logs)                         10.4        10.91       -0.511***
Supply disruption Y:1 N:0                         0.67        0.52        0.152***
Overdraft                                         0.29        0.48        -0.194***
Finance obstacle (0-4)                            1.45        1.24        0.210***
Skills obstacle (0-4)                             1.28        1.54        -0.266***
Young firm (<=10 years old): Y:1 N:0              0.21        0.11        0.098***
Exports (proportion of sales)                     0.05        0.19        -0.141***
Foreign ownership (proportion)                    0.06        0.19        -0.126***
Introduced new product/service Y:1 N:0           0.304        0.366       -0.062**
Spent on R&D Y:1 N:0                             0.093        0.204       -0.111***
Government support Y:1 N:0                        0.18        0.17          0.013
High share of women workers: Y:1 N:0              0.61        0.61          -0.002
Capital city Y:1 N:0                              0.32        0.32          -0.006
Taxes are a major obstacle Y:1 N:0               29.88        26.36         3.512
Obstacles severity                                1.24        1.22          0.019
Sole proprietorship Y:1 N:0                       0.29        0.09          0.209
Shareholding company (non-traded) or              0.57        0.63          -0.054
Partnership Y:1 N:0

Significance level is denoted by ***(1%), **(5%). Standard errors used are Huber-White
robust and clustered on the country.




                                          51
                                         Table 2: OLS

Dependent variable: Sales decline (%)             (1)          (2)          (3)

                                              Full sample    SMEs       Large firms
SME                                            4.362***
                                                (0.914)
Labor productivity (logs)                     -2.177***     -2.090***   -2.797***
                                                (0.388)       (0.473)     (0.721)
Supply disruption Y:1 N:0                     31.954***     33.145***   27.914***
                                                (1.293)       (1.467)     (2.139)
Overdraft Y:1 N:0                               -0.819        -0.720      -2.875
                                                (0.992)       (1.101)     (1.776)
Finance obstacle (0-4)                           0.428       0.880**      -0.880
                                                (0.430)       (0.433)     (0.877)
Skills obstacle (0-4)                            0.509         0.258     1.647**
                                                (0.327)       (0.348)     (0.809)
Young firm (<=10 years) Y:1 N:0                  1.682         1.949      -1.724
                                                (1.376)       (1.635)     (2.060)
High share of women workers Y:1 N:0              1.636         2.065      -1.614
                                                (1.128)       (1.256)     (1.780)
Exports (proportion of sales)                   3.395*       5.449**      -2.166
                                                (1.759)       (2.201)     (3.164)
Foreign ownership (proportion)                   0.633        -0.853      4.407*
                                                (1.489)       (2.451)     (2.256)
Tax rates are a major obstacle Y:1 N:0           0.003        -0.006      0.044*
                                                (0.010)       (0.009)     (0.023)
Government support Y:1 N:0                       1.340         1.597       0.234
                                                (1.880)       (2.180)     (3.330)
Capital city Y:1 N:0                            -1.035        -0.460      -2.650
                                                (1.371)       (1.663)     (1.981)
Obstacles severity                              -0.425        -0.596      -0.621
                                                (0.667)       (0.742)     (1.458)
Sole proprietorship Y:1 N:0                      0.850         0.540      -0.681
                                                (1.886)       (2.073)     (3.488)
Shareholding company (non-traded) or            -0.260        -1.351       0.891
Partnership Y:1 N:0                             (1.447)       (1.806)     (2.005)
Introduced new product/service Y:1 N:0         -1.783**       -1.592      -1.498
                                                (0.860)       (1.040)     (1.718)
Spent on R&D Y:1 N:0                            -0.686        -1.456       0.213
                                                (1.262)       (1.633)     (2.130)
Country-Industry fixed effects                    Yes           Yes         Yes
Constant                                      29.644***     32.973***   39.691***


                                             52
                                                     (4.178)           (5.258)             (8.267)
Number of observations                                5,197             4,097               1,100
Huber-White robust standard errors in brackets clustered on country-industry pairs. *** (1%), ** (5%),
* (10%).




                                                  53
                        Table 3: Baseline mean decomposition
Dependent variable: Sales decline (%)            (1)         (2)          (3)
                                                          Endowment   Structural
                                                            effect      effect
SMEs                                          34.766***
                                               (1.849)
Large firms                                   22.548***
                                               (2.209)
Difference (SME-Large firms)                  12.218***
                                               (1.883)
Country-Industry                                            2.732**   4.852***
                                                            (1.163)    (1.200)
Labor productivity (logs)                                  1.430***      7.354
                                                            (0.512)    (8.726)
Supply disruption Y:1 N:0                                  4.233***    3.509**
                                                            (0.758)    (1.515)
Overdraft Y:1 N:0                                             0.557      0.618
                                                            (0.341)    (0.518)
Finance obstacle (0-4)                                       -0.185    2.554**
                                                            (0.184)    (1.275)
Skills obstacle (0-4)                                      -0.439**     -1.776
                                                            (0.223)    (1.100)
Young firm (<=10 years) Y:1 N:0                              -0.168      0.773
                                                            (0.193)    (0.576)
High share of women workers Y:1 N:0                           0.003     2.237*
                                                            (0.052)    (1.164)
Exports (proportion of sales)                                 0.306     0.393*
                                                            (0.430)    (0.205)
Foreign ownership (proportion)                             -0.554**     -0.329
                                                            (0.280)    (0.231)
Tax rates are a major obstacle Y:1 N:0                        0.155   -1.488**
                                                            (0.129)    (0.659)
Government support Y:1 N:0                                    0.003      0.251
                                                            (0.042)    (0.646)
Capital city Y:1 N:0                                          0.015      0.697
                                                            (0.077)    (0.782)
Obstacles severity                                           -0.012      0.031
                                                            (0.043)    (1.928)
Sole proprietorship Y:1 N:0                                  -0.142      0.359
                                                            (0.690)    (0.999)

                                         54
Shareholding company (non-traded) or                        -0.048          -1.288
Partnership Y:1 N:0                                        (0.108)         (1.390)
R&D and New product/process                                  0.069          -0.183
                                                           (0.243)         (0.628)
Constant                                                                   -14.303
                                                                           (9.536)
Total or Aggregate effect                                 7.956***        4.262***
                                                           (1.927)         (1.215)
Number of observations                        5,197          5,197           5,197
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), **
(5%), * (10%). Without much loss of generality, only the combined contribution of
R&D and new product/process development is shown.




                                        55
                                                    Table 4: Baseline quantile decomposition results
Dependent variable: Sales decline (%)      (1)           (2)          (3)          (4)         (5)        (6)        (7)           (8)          (9)
                                          10th          20th         30th         40th        50th       60th       70th          80th         90th
                                         quantile     quantile     quantile     quantile   quantile    quantile   quantile       quantile     quantile
                                                                               Panel A: Total gap (unconditional)
SMEs                                     2.392***    11.163*** 18.058*** 30.776*** 35.634*** 47.300*** 56.919***                66.328***     84.028***
                                          (0.707)      (0.707)      (2.529)      (2.283)    (2.669)     (2.677)    (2.542)       (2.995)       (2.431)
Large firms                             -4.690***     4.911*** 10.368*** 11.853*** 23.501*** 32.331*** 36.359***                51.088***     66.107***
                                          (0.959)      (0.891)      (0.891)      (2.407)    (2.769)     (2.836)    (3.162)       (4.380)       (5.354)
Difference (SME-Large firms)             7.082***     6.252***     7.690*** 18.923*** 12.134*** 14.968*** 20.561***             15.240***     17.921***
                                          (0.838)      (0.804)         (2.319)      (2.283)     (2.620)  (2.702)      (2.508)     (3.435)      (4.565)
                                                                         Panel B: Aggregate endowment and structural effects
Aggregate endowment effect                  2.443**      2.270**      2.270**      9.958*** 10.668*** 9.397***       8.713*** 12.042***       19.370***
                                            (1.062)      (0.905)       (0.905)      (2.370)     (2.594)  (2.686)      (2.828)     (3.763)      (4.805)
Aggregate structural effect                4.639***     3.983***      5.420**      8.966***      1.466  5.571**     11.847***      3.199        -1.449
                                            (1.092)      (0.894)       (2.227)      (1.883)     (2.116)  (2.202)      (1.991)     (3.087)      (4.860)
                                                       Panel C: Aggregate endowment and structural effects as a percentage of the total gap
Aggregate endowment effect                   34.5%        36.3%         29.5%       52.6%       87.9%    62.8%        42.4%       79.0%        108.1%
Aggregate structural effect                  65.5%        63.7%         70.5%       47.4%       12.1%    37.2%        57.6%       21.0%         -8.1%
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), * (10%).




                                                                          56
                                     Table 5: Baseline quantile decomposition results for the individual factors
Dependent variable: Sales decline     (1)            (2)           (3)         (4)          (5)          (6)         (7)        (8)         (9)
(%)
                                     10th          20th          30th              40th       50th      60th        70th       80th       90th
                                    quantile      quantile      quantile        quantile   quantile   quantile     quantile   quantile   quantile
                                                                                Panel A: Endowment effects
Country-Industry                       0.248         0.496         0.496           1.759    2.960**   3.523**      4.834***   8.122***   13.183***
                                     (0.601)       (0.520)       (0.520)         (1.074)    (1.324)    (1.508)      (1.782)    (2.271)     (3.492)
Labor productivity (logs)              0.163         0.127         0.127        1.083**    1.645*** 2.051***        1.868**   3.115***     3.751**
                                     (0.169)       (0.155)       (0.155)         (0.463)    (0.594)    (0.675)      (0.749)    (1.107)     (1.705)
Supply disruption Y:1 N:0           1.683***      1.517***      1.517***        7.137*** 6.285*** 5.148***         4.301***   3.757***    4.483***
                                     (0.367)       (0.338)       (0.338)         (1.240)    (1.082)    (0.895)      (0.823)    (0.844)     (1.184)
Overdraft Y:1 N:0                    0.632**       0.508**       0.508**           0.528      0.424     0.703         0.408      0.012       0.667
                                     (0.263)       (0.216)       (0.216)         (0.410)    (0.519)    (0.444)      (0.466)    (0.628)     (1.124)
Finance obstacle (0-4)                 0.003        -0.055        -0.055           0.016     -0.093     0.037        -0.088     -0.450      -0.759
                                     (0.092)       (0.081)       (0.081)         (0.198)    (0.220)    (0.236)      (0.233)    (0.390)     (0.614)
Skills obstacle (0-4)                 -0.257      -0.270**      -0.270**           0.045     -0.185    -0.304       -0.707*    -0.752*    -0.944**
                                     (0.157)       (0.136)       (0.136)         (0.241)    (0.256)    (0.337)      (0.383)    (0.435)     (0.476)
Young firm (<=10 years) Y:1 N:0        0.054         0.033         0.033          -0.316     -0.113   -0.698*       -0.516*    -0.774*      -0.108
                                     (0.142)       (0.108)       (0.108)         (0.361)    (0.391)    (0.389)      (0.279)    (0.406)     (0.728)
High share of women workers            0.002         0.002         0.002           0.003      0.007     0.004         0.004     -0.002      -0.003
Y:1 N:0                              (0.029)       (0.034)       (0.034)         (0.046)    (0.118)    (0.065)      (0.066)    (0.037)     (0.055)
Exports (proportion of sales)         -0.004        -0.119        -0.119          1.113*      0.376    -0.614        -0.338     -0.396       0.866
                                     (0.519)       (0.417)       (0.417)         (0.674)    (0.583)    (0.602)      (0.545)    (0.828)     (1.166)
Foreign ownership (proportion)         0.056         0.026         0.026         -0.728*     -0.260    -0.384      -0.942**     -0.826      -0.628
                                     (0.200)       (0.176)       (0.176)         (0.407)    (0.416)    (0.431)      (0.454)    (0.593)     (0.897)
Tax rates are a major obstacle         0.171         0.155         0.155           0.121      0.207     0.194         0.174      0.071       0.077
Y:1 N:0                              (0.127)       (0.113)       (0.113)         (0.128)    (0.172)    (0.148)      (0.151)    (0.170)     (0.196)
Government support Y:1 N:0            -0.041        -0.033        -0.033           0.067      0.065     0.080         0.084      0.035      -0.093
                                     (0.078)       (0.062)       (0.062)         (0.129)    (0.126)    (0.154)      (0.168)    (0.115)     (0.184)

                                                                           57
Capital city Y:1 N:0                0.003        0.009        0.009      0.001        0.002        0.008     0.014     0.022       0.043
                                   (0.018)      (0.045)      (0.045)    (0.021)      (0.024)      (0.045)  (0.074)    (0.113)     (0.222)
Obstacles severity                 -0.017       -0.014       -0.014     -0.013       -0.013        0.000    -0.012     0.029       0.031
                                   (0.053)      (0.044)      (0.044)    (0.049)      (0.050)      (0.031)  (0.048)    (0.095)     (0.102)
Sole proprietorship Y:1 N:0        -0.089        0.018        0.018     -0.503       -0.911       -0.680    -0.531    -0.622      -1.997
                                   (0.478)      (0.420)      (0.420)    (0.834)      (0.931)      (0.829)  (0.836)    (1.868)     (2.197)
Shareholding company (non-         -0.077       -0.031       -0.031     -0.066       -0.077        0.060     0.107     0.193       0.073
traded) or Partnership Y:1 N:0     (0.101)      (0.076)      (0.076)    (0.152)      (0.169)      (0.155)  (0.184)    (0.240)     (0.282)
R&D and New                        -0.085       -0.099       -0.099     -0.289        0.348        0.270     0.052     0.507       0.730
product/process                    (0.254)      (0.214)      (0.214)    (0.338)      (0.309)      (0.316)  (0.314)    (0.378)     (0.562)
                                                                         Panel B: Structural effects
                                      10th         20th        30th       40th        50th         60th      70th        80th        90th
                                   quantile     quantile     quantile  quantile     quantile     quantile quantile    quantile    quantile
Country-Industry                  -12.167***   -14.472***   42.149*** 39.686*** 44.021*** 49.094*** 51.892***        95.260***   192.326***
                                    (3.630)      (3.577)      (6.239)   (7.758)      (7.148)      (6.547)  (6.961)     (9.183)    (14.151)
Labor productivity (logs)            -3.992       -4.720    -17.272**    0.599        6.592       14.808     9.908    36.437**      42.681
                                    (3.777)      (3.666)      (7.432)   (8.916)      (9.838)     (10.499) (12.113)    (18.207)    (30.460)
Supply disruption Y:1 N:0           2.838**     3.570***    31.807*** -4.471**      -4.491**       0.248    -0.246       1.765      -4.461
                                    (1.286)      (1.284)      (2.058)   (2.021)      (1.937)      (1.668)  (1.791)     (2.313)     (3.976)
Overdraft Y:1 N:0                  1.073***      0.889**       0.468     0.774        0.385        1.040     0.961      -0.004       0.210
                                    (0.414)      (0.356)      (0.562)   (0.699)      (0.807)      (0.680)  (0.724)     (1.061)     (2.023)
Finance obstacle (0-4)               -0.285        0.114       0.754     0.887        1.631        0.934   2.963*      6.450**     7.955**
                                    (0.796)      (0.751)      (1.060)   (1.686)      (1.688)      (1.704)  (1.749)     (2.758)     (3.865)
Skills obstacle (0-4)              -1.529**     -1.595**     -1.694*     0.280       -0.644       -1.169  -3.642**      -3.048      -2.262
                                    (0.759)      (0.679)      (0.932)   (1.309)      (1.474)      (1.837)  (1.810)     (2.037)     (2.156)
Young firm (<=10 years) Y:1 N:0      -0.129       -0.084       0.215     1.400        0.722      2.048**   1.624**     2.316**       0.580
                                    (0.363)      (0.296)      (0.412)   (0.910)      (0.954)      (0.933)  (0.698)     (1.016)     (1.868)
High share of women workers           0.692        0.801       0.721     1.450        2.394        1.952     2.796       2.684       3.769
Y:1 N:0                             (0.845)      (0.752)      (1.133)   (1.863)      (1.656)      (1.620)  (1.741)     (2.461)     (3.623)
Exports (proportion of sales)         0.038       -0.004       0.127    0.520*        0.343        0.065     0.204       0.615     1.247**
                                    (0.216)      (0.186)      (0.248)   (0.266)      (0.259)      (0.242)  (0.265)     (0.383)     (0.593)


                                                                       58
Foreign ownership (proportion)          0.031          0.017          -0.244      -0.506*      -0.306     -0.449    -0.672**      -0.468       -0.374
                                      (0.137)         (0.130)        (0.253)      (0.265)      (0.273)    (0.304)    (0.311)     (0.432)       (0.540)
Tax rates are a major obstacle       -1.850***      -1.715***      -1.740***       -1.274     -1.910**  -1.706**     -1.713*       0.034       -1.111
Y:1 N:0                               (0.530)         (0.468)        (0.479)      (0.898)      (0.970)    (0.768)    (0.949)     (1.412)       (1.823)
Government support Y:1 N:0              0.573          0.452         1.102**       -0.394      -0.538     -0.809      -1.244      -0.316        1.370
                                      (0.398)         (0.377)        (0.502)      (0.681)      (0.667)    (0.672)    (0.964)     (1.395)       (1.063)
Capital city Y:1 N:0                   -0.131          0.198           0.498        0.053       0.434      0.809       0.893       1.331        2.367
                                      (0.584)         (0.456)        (0.614)      (1.288)      (1.287)    (1.070)    (1.289)     (1.490)       (2.112)
Obstacles severity                      1.543          1.349           0.402       -0.173       0.645     -0.343      -0.477      -4.456       -2.099
                                      (1.667)         (1.420)        (1.553)      (2.344)      (2.580)    (2.371)    (2.408)     (3.650)       (3.803)
Sole proprietorship Y:1 N:0             0.311          0.160          -0.261        1.244       1.415      1.135       0.353       0.709        4.323
                                      (0.769)         (0.728)        (0.852)      (1.116)      (1.204)    (1.129)    (1.188)     (2.736)       (3.446)
Shareholding company (non-             -0.654         -0.168          -0.801       -0.538      -1.290      0.308      -0.250       0.981        1.246
traded) or Partnership Y:1 N:0        (1.174)         (1.107)        (1.351)      (1.818)      (1.947)    (1.734)    (1.989)     (2.478)       (3.303)
R&D and New product/process             0.028         -0.193          -0.688     -1.858**      -0.699     -0.835      -1.161       0.004        2.080
                                      (0.630)         (0.570)        (0.706)      (0.894)      (0.825)    (0.757)    (0.827)     (1.069)       (1.332)
Constant                             18.25***        19.38***      -50.12*** -28.71*** -47.24*** -61.56*** -50.34*** -137.1*** -251.30***
                                      (6.778)         (6.691)       (10.360)     (15.075)     (16.037)   (15.960)   (14.276)    (19.296)      (38.197)
Observations                            5,197          5,197           5,197        5,197       5,197      5,197       5,197       5,197        5,197
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), * (10%). Without much loss of generality, only the combined
contribution of R&D and New product/process development is shown.




                                                                         59
                                      Table 6: Baseline quantile decomposition results as a % of the total gap
Dependent variable: Sales decline (%)     (1)           (2)         (3)        (4)          (5)         (6)        (7)        (8)        (9)
                                          10th        20th        30th          40th        50th    60th          70th       80th       90th
                                         quantile    quantile    quantile    quantile quantile quantile          quantile   quantile   quantile
                                                                                Panel A: Endowment effects
Country-Industry                          3.50%       7.93%        6.45%       9.30%      24.40%   23.54%        23.51%     53.29%      73.56%
Labor productivity (logs)                 2.31%       2.04%        1.66%       5.72%      13.56%   13.70%        9.09%      20.44%      20.93%
Supply disruption Y:1 N:0                23.77%      24.27%       19.73%     37.72%       51.80%   34.39%        20.92%     24.65%      25.01%
Overdraft Y:1 N:0                         8.92%       8.12%        6.60%       2.79%       3.50%   4.70%         1.99%       0.08%       3.72%
Finance obstacle (0-4)                    0.04%      -0.88%       -0.72%       0.08%      -0.77%   0.25%         -0.43%     -2.95%      -4.23%
Skills obstacle (0-4)                    -3.62%      -4.32%       -3.51%       0.24%      -1.52%   -2.03%        -3.44%     -4.93%      -5.27%
Young firm (<=10 years) Y:1 N:0           0.76%       0.53%        0.43%      -1.67%      -0.93%   -4.66%        -2.51%     -5.08%      -0.60%
High share of women workers Y:1 N:0       0.02%       0.03%        0.03%       0.01%       0.06%   0.03%         0.02%      -0.01%      -0.02%
Exports (proportion of sales)            -0.06%      -1.90%       -1.54%       5.88%       3.10%   -4.10%        -1.65%     -2.60%       4.83%
Foreign ownership (proportion)            0.78%       0.42%        0.34%      -3.85%      -2.14%   -2.57%        -4.58%     -5.42%      -3.50%
Tax rates are a major obstacle Y:1 N:0    2.41%       2.48%        2.02%       0.64%       1.70%   1.29%         0.84%       0.47%       0.43%
Government support Y:1 N:0               -0.58%      -0.52%       -0.43%       0.35%       0.54%   0.53%         0.41%       0.23%      -0.52%
Capital city Y:1 N:0                      0.04%       0.14%        0.11%       0.00%       0.02%   0.05%         0.07%       0.14%       0.24%
Obstacles severity                       -0.24%      -0.23%       -0.19%      -0.07%      -0.11%   0.00%         -0.06%      0.19%       0.17%
Sole proprietorship Y:1 N:0              -1.26%       0.28%        0.23%      -2.66%      -7.51%   -4.54%        -2.58%     -4.08%     -11.14%
Shareholding company (non-traded) or     -1.09%      -0.50%       -0.41%      -0.35%      -0.64%   0.40%         0.52%       1.27%       0.41%
Partnership Y:1 N:0
R&D and New product/process               -1.20%     -1.59%       -1.29%      -1.53%        2.87%     1.80%       0.25%      3.33%      4.07%
                                                                                 Panel B: Structural effects
Country-Industry                         -171.80%   -231.47%      548.12%    209.72% 362.81% 327.98%             252.39%    625.05%    1073.18%
Labor productivity (logs)                 -56.37%    -75.50%     -224.62%      3.16%       54.32%     98.93%      48.19%    239.08%     238.16%
Supply disruption Y:1 N:0                 40.07%      57.11%      413.63%    -23.63%      -37.02%     1.66%       -1.20%     11.58%     -24.89%
Overdraft Y:1 N:0                         15.15%      14.22%       6.09%       4.09%        3.18%     6.95%       4.67%      -0.03%       1.17%
Finance obstacle (0-4)                     -4.03%      1.82%       9.81%       4.69%       13.44%     6.24%       14.41%     42.32%      44.39%
Skills obstacle (0-4)                     -21.59%    -25.51%      -22.03%      1.48%       -5.31%     -7.81%     -17.71%    -20.00%     -12.62%
Young firm (<=10 years) Y:1 N:0            -1.82%     -1.35%       2.80%       7.40%        5.95%     13.68%      7.90%      15.20%       3.24%

                                                                        60
High share of women workers Y:1 N:0        9.77%    12.81%     9.38%     7.66%    19.73%     13.04%    13.60%     17.61%    21.03%
Exports (proportion of sales)              0.54%    -0.06%     1.65%     2.75%     2.83%     0.43%     0.99%       4.03%     6.96%
Foreign ownership (proportion)             0.44%     0.26%    -3.17%    -2.67%    -2.52%     -3.00%    -3.27%     -3.07%    -2.08%
Tax rates are a major obstacle Y:1 N:0   -26.13%   -27.44%   -22.63%    -6.73%   -15.74%    -11.40%    -8.33%      0.22%    -6.20%
Government support Y:1 N:0                 8.09%     7.23%    14.33%    -2.08%    -4.43%     -5.40%    -6.05%     -2.08%     7.64%
Capital city Y:1 N:0                      -1.84%     3.17%     6.48%     0.28%     3.58%     5.40%     4.35%       8.74%    13.21%
Obstacles severity                       21.79%     21.58%     5.22%    -0.91%     5.32%     -2.29%    -2.32%    -29.24%   -11.71%
Sole proprietorship Y:1 N:0                4.39%     2.56%    -3.40%     6.57%    11.66%     7.58%     1.72%       4.65%    24.12%
Shareholding company (non-traded) or      -9.24%    -2.69%   -10.41%    -2.84%   -10.63%     2.06%     -1.22%      6.44%     6.95%
Partnership Y:1 N:0
R&D and New product/process              0.40%     -3.09%    -8.94%     -9.82%    -5.76%    -5.58%     -5.65%     0.02%    11.61%
    Without much loss of generality, only the combined contribution of R&D and New product/process development is shown.




                                                                   61
                            Table 7: Mean decomposition by regions
Dependent variable: Sales decline (%)     (1)          (2)                (3)            (4)
                                         SSA          ECA                LAC            MENA
                                   Panel A: Aggregate effects

SMEs                                     49.499***     28.216***        27.483***      45.674***
                                           (1.153)       (1.693)         (2.003)         (2.655)
Large firms                              38.097***     16.994***        16.052***      39.509***
                                           (1.709)       (2.650)         (2.479)         (1.569)
Difference (SME-Large firms)             11.402***     11.222***        11.432***        6.165**
                                           (1.290)       (2.676)         (2.832)         (2.420)
Total or Aggregate                        5.056**         4.120           7.120*        7.651***
endowment effect                           (2.445)       (2.703)         (4.010)         (1.965)
Total or Aggregate structural             6.346**       7.102***           4.311          -1.486
effect                                     (2.497)       (1.487)         (3.731)         (2.947)
Number of observations                      1,087         2,695             726            444
                Panel B: Aggregate effects as a % of the total unconditional gap
Total or Aggregate endowment effect       44.34%         36.71%          62.29%         124.11%
Total or Aggregate structural effect      55.66%         63.29%          37.71%         -24.11%
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), * (10%).




                                                62
                                                        Table 8: Mean decomposition by regions
                                      (1)             (2)             (3)              (4)       (5)             (6)      (7)            (8)
                                             SSA                            ECA                         LAC                    MENA
Total unconditional gap                   11.402***                       11.222***                  11.432***                 6.165**
                                           (1.290)                         (2.676)                     (2.832)                 (2.420)
                                                             Panel A; Mean decomposition estimates (Percentage points)
                                 Endowment Structural Endowment                  Structural Endowment Structural Endowment Structural
                                    effect          effect          effect           effect     effect          effect  effect          effect
Country-Industry                    -2.169          -1.866           1.045         3.731***     1.415          -0.730    0.531         -3.066
                                   (1.637)         (3.428)         (1.956)          (1.358)    (1.511)         (2.584) (0.725)         (3.164)
Labor productivity (logs)          2.615**          14.327           0.502           -8.923     2.090          -4.003    0.441       28.271***
                                   (1.326)        (14.629)         (0.411)         (12.581)    (1.344)        (23.750) (1.223)         (8.768)
Supply disruption Y:1 N:0         3.586***           2.001        3.094***           2.703*  7.448***           2.420  3.101**       18.866***
                                   (1.314)         (5.251)         (0.896)          (1.569)    (1.731)         (2.879) (1.284)         (5.500)
Overdraft Y:1 N:0                    1.893           2.003           0.249            0.487    -0.138          -0.641    1.667         -0.742
                                   (1.337)         (1.282)         (0.317)          (0.629)    (0.895)         (2.168) (3.227)         (2.560)
Finance obstacle (0-4)               0.007           2.066          -0.012           2.359*     0.744          -0.842   -1.108         8.993*
                                   (0.148)         (3.109)         (0.156)          (1.233)    (0.865)         (2.768) (1.148)         (5.043)
Skills obstacle (0-4)               -0.092           0.146         -0.499*           -1.690    -0.128        -6.141**    1.292          6.680
                                   (0.233)         (1.973)         (0.290)          (1.318)    (0.412)         (2.883) (1.202)         (4.126)
Young firm (<=10 years) Y:1          0.080         -1.391*           0.093            0.973    -0.034           0.227   -0.032          0.976
N:0                                (0.288)         (0.815)         (0.313)          (0.944)    (0.144)         (0.331) (1.186)         (2.216)
High share of women                 -0.030           2.376          -0.066           2.643*    -0.018           0.790    0.645          4.107
workers Y:1 N:0                    (0.233)         (1.603)         (0.137)          (1.488)    (0.122)         (3.067) (0.731)         (3.353)
Exports (proportion of sales)        1.048           0.721          -0.058            0.428     0.484         0.494**    1.307          0.107
                                   (0.907)         (0.491)         (0.493)          (0.304)    (1.002)         (0.217) (1.099)         (0.354)
Foreign ownership                   -0.144          -0.640          -0.185           -0.049    -2.227          -0.463   -1.217         -0.040
(proportion)                       (1.048)         (0.829)         (0.214)          (0.187)    (1.475)         (0.577) (1.176)         (0.942)
Tax rates are a major obstacle      -0.055          -0.182           0.261          -1.519*    -0.055           0.768   -0.041         -0.513
Y:1 N:0                            (0.142)         (1.445)         (0.244)          (0.879)    (0.380)         (2.021) (0.220)         (1.868)
Government support Y:1 N:0           0.149          -0.159          -0.102            1.533    -0.675           0.508  -1.312*       -3.127***


                                                                       63
                                  (0.184)   (0.336)        (0.172)        (1.109)       (0.789)      (1.616)       (0.713)    (0.920)
Capital city Y:1 N:0                1.414    3.393         -0.030          0.478        -0.159        1.558        -0.008      0.612
                                  (0.956)   (2.658)        (0.062)        (0.713)       (0.420)      (2.071)       (0.101)    (0.885)
Obstacles severity                 -0.030    0.512         -0.001         -0.415        -0.140        0.319         0.588    -22.195*
                                  (0.383)   (4.645)        (0.028)        (1.308)       (0.427)      (6.137)       (1.889)   (13.117)
Sole proprietorship Y:1 N:0      -4.510**   4.712**        -0.404          1.145         0.263        0.152         1.921     -2.113
                                  (1.943)   (2.339)        (0.459)        (0.801)       (1.568)      (3.070)       (3.714)    (6.283)
Shareholding company (non-        1.209**    1.040          0.166         -1.026         0.180       -1.212        -1.042     -0.951
traded) or Partnership Y:1        (0.473)   (1.310)        (0.259)        (2.730)       (0.928)      (2.520)       (1.295)    (2.265)
N:0
R&D and New                       0.084     -0.135         0.069          -0.033        -1.929        -4.445      0.918***   0.941**
product/process
                                 (0.427)     (1.033)      (0.206)       (0.635)       (1.540)       (3.806)        (0.338)     (0.434)
Constant                                    -22.576                      4.275                      15.552                   -38.293**
                                            (19.971)                   (12.989)                    (31.439)                   (10.931)
Number of observations             1,087        1,087        2,695         2,695          726           726            444        444
                                             Panel B: Mean decomposition estimates above as a percentage of the total gap
Country-Industry                 -19.0%      -16.4%         9.3%         33.3%        12.4%          -6.4%           8.6%     -19.7%
Labor productivity (logs)         22.9%     125.7%          4.5%        -79.5%        18.3%         -35.0%           7.2%     458.6%
Supply disruption Y:1 N:0         31.4%       17.6%       27.6%          24.1%        65.1%          21.2%         50.3%      306.0%
Overdraft Y:1 N:0                 16.6%       17.6%         2.2%          4.3%         -1.2%         -5.6%         27.0%      -12.0%
Finance obstacle (0-4)             0.1%       18.1%        -0.1%         21.0%          6.5%         -7.4%         -18.0%     145.9%
Skills obstacle (0-4)             -0.8%        1.3%        -4.4%        -15.1%         -1.1%        -53.7%         21.0%      108.4%
Young firm (<=10 years) Y:1        0.7%      -12.2%         0.8%          8.7%         -0.3%          2.0%          -0.5%      15.8%
N:0
High share of women               -0.3%     20.8%          -0.6%          23.6%         -0.2%         6.9%         10.5%      66.6%
workers Y:1 N:0
Exports (proportion of sales)      9.2%      6.3%          -0.5%           3.8%          4.2%         4.3%         21.2%       1.7%
Foreign ownership                 -1.3%      -5.6%         -1.7%          -0.4%         -19.5%        -4.1%        -19.7%     -0.6%
(proportion)
Tax rates are a major obstacle    -0.5%      -1.6%         2.3%           -13.5%        -0.5%         6.7%         -0.7%      -8.3%
Y:1 N:0
Government support Y:1 N:0        1.3%       -1.4%         -0.9%          13.7%         -5.9%         4.4%         -21.3%     -50.7%

                                                                     64
Capital city Y:1 N:0              12.4%           29.8%         -0.3%           4.3%          -1.4%        13.6%         -0.1%         9.9%
Obstacles severity                -0.3%            4.5%          0.0%          -3.7%          -1.2%         2.8%          9.5%       -360.0%
Sole proprietorship Y:1 N:0      -39.6%           41.3%         -3.6%          10.2%           2.3%         1.3%        31.2%         -34.3%
Shareholding company (non-        10.6%            9.1%          1.5%          -9.1%           1.6%       -10.6%        -16.9%        -15.4%
traded) or Partnership Y:1
N:0
R&D and New                        0.7%           -1.2%          0.6%          -0.3%         -16.9%       -38.9%        14.9%         15.3%
product/process
Huber-White robust standard errors clustered on   country-industry pairs. *** (1%), ** (5%), * (10%). Without much loss of generality, only the
combined contribution of R&D and New product/process development is shown.




                                                                        65
                                                Appendix A
                                                  May 2023


                                     Table A1: Sample description.
                                                        Fiscal year               Last completed month
Country                     Number of firms
                                                      covered by ES                covered by COV-ES
Albania                              282                   2018                         May, 2020
Belarus                              462                   2017                         July, 2020
Bulgaria                             370                   2018                  October-November, 2020
Chad                                  75                   2017                         May, 2020
El Salvador                          313                   2015                  October-December, 2020
Georgia                              348                   2018                  September-October, 2020
Guatemala                            143                   2016                 November-December, 2020
Guinea                                40                   2015                         May, 2020
Honduras                             104                   2015                  October-December, 2020
Moldova                              220                   2018                  September-October, 2020
Mongolia                             245                   2018                         July, 2020
Morocco                              444                   2018                      June-July, 2020
Nicaragua                            166                   2016                 November-December, 2020
Niger                                 39                   2016                         May, 2020
North Macedonia                      179                   2018                      September, 2020
Russian Federation                   834                   2018                         May, 2020
Togo                                  39                   2015                         May, 2020
Zambia                               439              2018/2019/2020                 May-June, 2020
Zimbabwe                             455                   2015                      May-June, 2020
All firms                           5,197
Note: The data is extracted from the World Bank Enterprise Surveys and the Covid-19-ES Follow up Survey.




                                                        66
                             Table A2: Description of variables
Variable                                Description
Sales decline                           The decline in sales in the last month covered by the
                                        survey (after the outbreak of the pandemic) compared
                                        to sales in the same month in 2019 (pre-pandemic)
                                        and expressed as a percentage of the latter.
                                        A positive value of the variable that sales were lower
                                        after the outbreak of the pandemic than before. A
                                        negative value of the variable implies higher sales
                                        after the outbreak of the pandemic than before.
                                        Source: COVID follow-up Enterprise Surveys
                                        (COV-ES), World Bank
SME                                     A dummy variable equal to 1 if the firm has less than
                                        100 full-time workers at the end of the last fiscal year
                                        and 0 otherwise. Full-time workers include
                                        permanent and temporary workers with the latter
                                        adjusted for the number of months worked in the last
                                        fiscal year.
                                        The variable is for the pre-pandemic period.
                                        Source: Enterprise Surveys (ES), World Bank
Large (firm)                            A dummy variable equal to 1 if the firm has 100 or
                                        more full-time workers at the end of the last fiscal
                                        year and 0 otherwise. Full-time workers include
                                        permanent and temporary workers with the latter
                                        adjusted for the number of months worked in the last
                                        fiscal year.
                                        The variable is for the pre-pandemic period.
                                        Source: Enterprise Surveys (ES), World Bank
Country-industry fixed effects          A set of dummy variables indicating the country
                                        times industry to which the firm belongs. Industry is
                                        defined a the 2-digit ISIC Rev. 3.1 level. There are
                                        seven industries in the baseline sample. The variable
                                        is for the pre-pandemic period.
                                        Source: Enterprise Surveys (ES), World Bank
Labor Productivity (logs)               Log of annual sales of the firm in the last fiscal year
                                        (expressed in 2009 USD) divided by the total number
                                        of full-time workers at the firm at the end of the last
                                        fiscal year. Full-time workers include permanent and
                                        temporary workers with the latter adjusted for the
                                        number of months worked in the last fiscal year.
                                        The variable is for the pre-pandemic period.
                                        Source: Enterprise Surveys (ES), World Bank
Supply disruption Y:1 N:0               A dummy variable equal to 1 if the firm’s supply of
                                        inputs, raw materials, or finished goods and materials
                                        purchased to resell decreased in the last month (after
                                              67
                                       the outbreak of the pandemic) compared to the same
                                       month in 2019 and 0 otherwise.
                                       This information is based on firms’ self-reports of the
                                       change in inputs supplies. The variable is for the
                                       period after the outbreak of the COVID-19 pandemic.
                                       Source: COVID follow-up Enterprise Surveys
                                       (COV-ES), World Bank
Overdraft                              A dummy variable equal to 1 if the firm has overdraft
                                       facility and 0 otherwise. The variable is for the pre-
                                       pandemic period.
                                       Source: Enterprise Surveys (ES), World Bank
Finance obstacle (0-4)                 Firm’s self-report on how severe access to finance is
                                       as an obstacle to the firm’s current operations. The
                                       severity level ranges between 0 and 4 as follows: No
                                       obstacle (0), minor obstacle (1), moderate obstacle
                                       (2), major obstacle (3), and a very severe obstacle (4).
                                       The variable is for the pre-pandemic period.
                                       Source: Enterprise Surveys (ES), World Bank
Skills obstacle (0-4)                  Firm’s self-report on how severe inadequately
                                       educated workers is as an obstacle to the firm’s
                                       current operations. The severity level ranges between
                                       0 and 4 as follows: No obstacle (0), minor obstacle
                                       (1), moderate obstacle (2), major obstacle (3), and a
                                       very severe obstacle (4).
                                       The variable is for the pre-pandemic period.
                                       Source: Enterprise Surveys (ES), World Bank
Young firm (<=10 years old): Y:1 N:0   A dummy variable equal to 1 if the firm is 10 years
                                       or younger in 2020 and 0 otherwise.
                                       Source: Enterprise Surveys (ES), World Bank
Exports (proportion of sales)          Proportion of firm’s sales last fiscal year that were
                                       made abroad directly by the firm.
                                       The variable is for the pre-pandemic period.
                                       Source: Enterprise Surveys (ES), World Bank
Foreign ownership (proportion)         Proportion of firm owned by foreign individuals,
                                       companies, or organizations.
                                       The variable is for the pre-pandemic period.
                                       Source: Enterprise Surveys (ES), World Bank
Introduced new product/service Y:1     A dummy variable equal to 1 if the firm introduced a
N:0                                    new product or service in the last three years and 0
                                       otherwise.
                                       The variable is for the pre-pandemic period.
                                       Source: Enterprise Surveys (ES), World Bank
Spent on R&D Y:1 N:0                   A dummy variable equal to 1 if the firm spent on
                                       R&D activity, either in-house or contracted with
                                       other companies, during the last fiscal year and 0
                                       otherwise.

                                           68
                                     The variable is for the pre-pandemic period.
                                     Source: Enterprise Surveys (ES), World Bank
Government support Y:1 N:0           A dummy variable equal to 1 if the firm received nay
                                     national or local government support since May 2020
                                     in response to the COVID-19 crisis and 0 otherwise.
                                     Source: COVID follow-up Enterprise Surveys
                                     (COV-ES), World Bank.
High share of women workers: Y:1 N:0 A dummy variable equal to 1 if the share of women
                                     in total full-time permanent workers at the firm at the
                                     end of the last fiscal year is more than 25 percent and
                                     0 otherwise.
                                     The variable is for the pre-pandemic period.
                                     Source: Enterprise Surveys (ES), World Bank
Capital city Y:1 N:0                 A dummy variable equal to 1 if the firm is in the
                                     country’s capital city and 0 otherwise.
                                     The variable is for the pre-pandemic period.
                                     Source: Enterprise Surveys (ES), World Bank
Taxes are a major obstacle Y:1 N:0   A dummy variable equal to 1 if the firm reported tax
                                     rates as a major or very severe obstacle for its current
                                     operations and 0 if it reported them as no obstacle,
                                     minor obstacle, or a moderate obstacle.
                                     The variable is for the pre-pandemic period.
                                     Source: Enterprise Surveys (ES), World Bank
Obstacles severity                   In separate questions, firms were asked if electricity,
                                     tax administration, obtaining licenses and permits,
                                     corruption, and labor laws are an obstacle for their
                                     current operations. Responses were recorded as no
                                     obstacle (0), minor obstacle (1), moderate obstacle
                                     (2), major obstacle (3), and a very severe obstacle (4).
                                     For each firm, we take the average over the reported
                                     severity levels of all the obstacles.
                                     The variable is for the pre-pandemic period.
                                     Source: Enterprise Surveys (ES), World Bank
Sole proprietorship Y:1 N:0          A dummy variable equal to 1 if the firm is a sole
                                     proprietorship firm and 0 otherwise.
                                     The variable is for the pre-pandemic period.
                                     Source: Enterprise Surveys (ES), World Bank
Shareholding company (non-traded) or A dummy variable equal to 1 if the firm is a
Partnership Y:1 N:0                  shareholding company with non-traded shares or a
                                     Partnership and 0 otherwise.
                                     The variable is for the pre-pandemic period.
                                     Source: Enterprise Surveys (ES), World Bank




                                             69
                                 Table A3: Summary statistics
                                                     Mean         Std.   Min    Max   Obs.
                                                                 Dev.
Sales decline (%)                                       32.18   32.673   -300   100   5,197
SME                                                     0.788    0.409     0     1    5,197
Labor productivity (logs)                              10.507    1.546   1.250 16.978 5,197
High share of women workers Y:1 N:0                     0.608    0.488     0     1    5,197
Overdraft Y:1 N:0                                       0.328    0.470     0     1    5,197
Finance obstacle (0-4)                                  1.406    1.280     0     4    5,197
Young firm (<=10 years) Y:1 N:0                         0.190    0.392     0     1    5,197
Exports (proportion of sales)                           0.082    0.225     0     1    5,197
Foreign ownership (proportion)                          0.089    0.262     0     1    5,197
Tax rates are a major obstacle Y:1 N:0                  0.291    0.454     0     1    5,197
Skills obstacle (0-4)                                   1.335    1.241     0     4    5,197
Supply disruption Y:1 N:0                               0.639    0.480     0     1    5,197
Government support Y:1 N:0                              0.181    0.385     0     1    5,197
Obstacles severity                                      1.238    0.885     0     4    5,197
Sole proprietorship Y:1 N:0                             0.250    0.433     0     1    5,197
Shareholding company (non-traded) or Partnership Y:1    0.586    0.493     0     1    5,197
N:0
Capital city Y:1 N:0                                   0.319    0.466     0      1    5,197
Introduced new product/service Y:1 N:0                 0.317    0.465     0      1    5,197
Spent on R&D Y:1 N:0                                   0.116    0.321     0      1    5,197




                                               70
                                                Table A4: Robustness of mean decomposition
                   Baseline results (Refence groups:         Reference groups: Large firms         Reference group: Pooled (SME and
                                SMEs)                                                                     large firms) sample
Dependent          (1)         (2)            (3)          (4)             (5)         (6)          (7)         (8)          (9)
variable:
Sales decline
(%)
                            Endowment     Structural                Endowment       Structural               Endowment    Structural
                              effect        effect                    effect          effect                   effect       effect
SMEs            34.766***                               34.766***                                34.766***
                 (1.849)                                 (1.849)                                  (1.849)
Large firms     22.548***                               22.548***                                22.548***
                 (2.209)                                 (2.209)                                  (2.209)
Difference      12.218***                               12.218***                                12.218***
(SME-Large
firms)           (1.883)                                 (1.883)                                  (1.883)
Country-                     2.732**       4.852***                      1.848**    5.737***                  1.776**     5.808***
Industry
                              (1.163)       (1.200)                       (0.781)    (1.304)                   (0.865)     (1.171)
Labor                        1.430***        7.354                       1.068***     7.716                   1.113***      7.671
productivity
(logs)
                              (0.512)      (8.726)                        (0.367)    (9.156)                   (0.351)     (9.076)
Supply                       4.233***      3.509**                       5.026***    2.716**                  4.846***     2.896**
disruption
Y:1
N:0                          (0.758)        (1.515)                      (0.852)     (1.180)                  (0.817)      (1.262)
Overdraft Y:1                 0.557          0.618                        0.140       1.036                    0.159        1.017
N:0
                             (0.341)        (0.518)                      (0.211)     (0.865)                  (0.192)      (0.784)

                                                                    71
Finance           -0.185    2.554**         0.185*     2.185**    0.090     2.280**
obstacle (0-4)
                  (0.184)   (1.275)         (0.107)    (1.092)    (0.094)   (1.128)
Skills           -0.439**    -1.776          -0.069     -2.145     -0.135    -2.079
obstacle (0-4)
                 (0.223)    (1.100)         (0.092)    (1.329)    (0.090)   (1.284)
Young firm        -0.168     0.773           0.190      0.414      0.164     0.440
(<=10 years)
Y:1 N:0          (0.193)    (0.576)         (0.160)    (0.309)    (0.136)   (0.344)
High share of     0.003     2.237*           -0.004    2.244*      -0.003   2.243*
women
workers Y:1      (0.052)    (1.164)         (0.067)    (1.175)    (0.053)   (1.173)
N:0
Exports           0.306     0.393*          -0.770**   1.469*     -0.480*   1.179**
(proportion of
sales)            (0.430)   (0.205)         (0.361)    (0.793)    (0.273)   (0.586)
Foreign          -0.554**    -0.329          0.107      -0.990     -0.080    -0.803
ownership
(proportion)     (0.280)     (0.231)        (0.303)     (0.690)   (0.185)    (0.518)
Tax rates are     0.155     -1.488**         -0.020    -1.313**    0.012    -1.345**
a major
obstacle Y:1     (0.129)    (0.659)         (0.035)    (0.585)    (0.034)   (0.597)
N:0
Government        0.003      0.251           0.021      0.233     0.018      0.236
support Y:1
N:0              (0.042)    (0.646)         (0.048)    (0.600)    (0.040)   (0.609)
Capital city      0.015      0.697           0.003      0.709      0.006     0.706
Y:1 N:0
                 (0.077)    (0.782)         (0.016)    (0.799)    (0.031)   (0.795)
Obstacles         -0.012     0.031           -0.011     0.031      -0.008    0.027
severity
                 (0.043)    (1.928)         (0.035)    (1.898)    (0.026)   (1.903)


                                       72
Sole                          -0.142          0.359                       0.113         0.104                       0.178          0.040
proprietorship
Y:1
N:0                           (0.690)        (0.999)                      (0.426)       (0.291)                    (0.389)        (0.402)
Shareholding                   -0.048         -1.288                       0.073         -1.409                     0.014          -1.350
company
(non-traded)                  (0.108)        (1.390)                      (0.108)       (1.519)                    (0.077)        (1.477)
or Partnership
Y:1 N:0
R&D and New                    0.069         -0.183                       0.262         -0.376                      0.188         -0.302
product/process
                              (0.243)        (0.628)                      (0.184)       (0.855)                    (0.147)        (0.792)
Constant                                     -14.303                                    -14.303                                   -14.303
                                             (9.536)                                    (9.536)                                   (9.536)
Total or                     7.956***       4.262***                     8.161***      4.057***                  7.856***        4.362***
Aggregate
effect                        (1.927)         (1.215)                     (1.663)       (1.100)                   (1.698)         (0.842)
Number of         5,197        5,197           5,197         5,197         5,197         5,197        5,197        5,197           5,197
observations
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), * (10%). Without much loss of generality, only the
combined contribution of R&D and New product/process development is shown.




                                                                     73
                                   Table A5: Threefold mean decomposition
                             Baseline mode                           Threefold decomposition
Dependent           (1)         (2)           (3)          (4)          (5)         (6)              (7)
variable:
Sales decline
(%)
                             Endowment    Structural                    Endowment   Structural   Interaction
                               effect       effect                        effect      effect        effect
SMEs             34.766***                                  34.766***
                  (1.849)                                    (1.849)
Large firms      22.548***                                  22.548***
                  (2.209)                                    (2.209)
Difference       12.218***                                  12.218***
(SME-
Large firms)      (1.883)                                    (1.883)
Country-                      2.732**      4.852***                      2.732**    5.737***       -0.885
Industry
                               (1.163)      (1.200)                       (1.163)    (1.304)       (0.863)
Labor                         1.430***       7.354                       1.430***     7.716         -0.362
productivity
(logs)                         (0.512)      (8.726)                       (0.512)    (9.156)      (0.439)
Supply                        4.233***      3.509**                      4.233***    2.716**      0.793**
disruption
Y:1 N:0                       (0.758)       (1.515)                      (0.758)     (1.180)       (0.365)
Overdraft Y:1                  0.557         0.618                        0.557       1.036         -0.418
N:0
                              (0.341)       (0.518)                      (0.341)     (0.865)       (0.355)
Finance                        -0.185       2.554**                       -0.185     2.185**       0.369*
obstacle (0-
4)                             (0.184)      (1.275)                       (0.184)    (1.092)       (0.219)
Skills                        -0.439**       -1.776                      -0.439**     -2.145        0.370
obstacle (0-4)
                              (0.223)       (1.100)                      (0.223)     (1.329)       (0.241)
Young firm                     -0.168        0.773                        -0.168      0.414         0.359
(<=10
years) Y:1                    (0.193)       (0.576)                      (0.193)     (0.309)       (0.273)
N:0
High share of                  0.003        2.237*                        0.003      2.244*         -0.007
women                         (0.052)       (1.164)                      (0.052)     (1.175)       (0.119)
workers Y:1
N:0
Exports                        0.306        0.393*                        0.306      1.469*       -1.076*
(proportion
                                                       74
of sales)                      (0.430)         (0.205)                      (0.430)         (0.793)        (0.600)
Foreign                       -0.554**          -0.329                     -0.554**          -0.990         0.661
ownership
(proportion)                   (0.280)         (0.231)                      (0.280)         (0.690)        (0.464)
Tax rates are                   0.155         -1.488**                       0.155         -1.313**         -0.175
a
major                          (0.129)         (0.659)                      (0.129)         (0.585)        (0.141)
obstacle Y:1
N:0
Government                      0.003           0.251                        0.003          0.233           0.018
support
Y:1 N:0                        (0.042)         (0.646)                      (0.042)         (0.600)        (0.056)
Capital city                    0.015           0.697                        0.015           0.709          -0.012
Y:1 N:0
                               (0.077)         (0.782)                      (0.077)         (0.799)        (0.064)
Obstacles                       -0.012          0.031                        -0.012          0.031          0.000
severity
                               (0.043)         (1.928)                      (0.043)         (1.898)        (0.030)
Sole                            -0.142          0.359                        -0.142          0.104          0.255
proprietorship
Y:1 N:0                        (0.690)         (0.999)                      (0.690)         (0.291)        (0.709)
Shareholding                    -0.048          -1.288                       -0.048          -1.409         0.121
company                        (0.108)         (1.390)                      (0.108)         (1.519)        (0.155)
(non-traded)
or Partnership
Y:1 N:0
R&D and New                     0.069           -0.183                       0.069          -0.376          0.193
product/process
                               (0.243)         (0.628)                      (0.243)         (0.855)        (0.306)
Constant                                       -14.303                                      -14.303
                                               (9.536)                                      (9.536)
Total or                       7.956***       4.262***                       7.956***      4.057***          0.205
Aggregate
effect                          (1.927)          (1.215)                      (1.927)         (1.100)       (1.348)
Number of           5,197        5,197            5,197        5,197           5,197           5,197         5,197
observations
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), * (10%). Without much loss
of generality, only the combined contribution of R&D and New product/process development is shown.




                                                         75
       Table A6: Robustness of mean decomposition for additional controls
Dependent variable: Sales decline (%)    (1)           (2)              (3)
                                                  Endowment         Structural
                                                     effect            effect
SMEs                                  34.794***
                                          (1.868)
Large firms                              22.468***
                                          (2.228)
Difference (SME-Large firms)             12.326***
                                          (1.880)
Country-Industry                                     2.923***          1.776
                                                      (1.025)         (1.361)
Labor productivity (logs)                            1.491***         12.265
                                                      (0.502)         (8.543)
Supply disruption Y:1 N:0                            4.171***        3.189**
                                                      (0.843)         (1.551)
Overdraft Y:1 N:0                                       0.505          0.529
                                                      (0.336)         (0.557)
Finance obstacle (0-4)                                 -0.177        2.389**
                                                      (0.170)         (1.200)
Skills obstacle (0-4)                                 -0.488*         -2.017
                                                      (0.262)         (1.304)
Young firm (<=10 years) Y:1 N:0                        -0.077          0.685
                                                      (0.148)         (0.478)
High share of women workers Y:1 N:0                    -0.002        2.464**
                                                      (0.074)         (1.218)
Exports (proportion of sales)                           0.110          0.345
                                                      (0.439)         (0.223)
Foreign ownership (proportion)                         -0.432         -0.234
                                                      (0.280)         (0.214)
Tax rates are a major obstacle Y:1 N:0                  0.196       -1.642***
                                                      (0.150)         (0.622)
Government support Y:1 N:0                              0.007          0.281
                                                      (0.037)         (0.700)
Capital city Y:1 N:0                                    0.034          0.872
                                                      (0.096)         (0.802)
Obstacles severity                                     -0.018          0.824
                                                      (0.056)         (2.045)
Sole proprietorship Y:1 N:0                            -0.023          0.196
                                                      (0.731)         (1.080)
Shareholding company (non-traded) or                   -0.037         -1.288

                                          76
Partnership Y:1 N:0                                          (0.107)            (1.568)
R&D and New product/process                                   0.019              0.042
                                                             (0.244)            (0.646)
Firm has female owners Y:1 N:0                                0.010              0.353
                                                             (0.031)            (0.787)
Losses from crime Y:1 N:0                                     0.156              0.160
                                                             (0.099)            (0.385)
Employment growth rate (annual)                               0.062             -0.101
                                                             (0.066)            (0.270)
Share of online sales (%)                                     0.011              0.200
                                                             (0.032)            (0.406)
Share of workforce working remotely                          -0.107             -0.224
(%)                                                          (0.124)            (0.419)
Constant                                                                       -17.077*
                                                                               (10.077)
Total or Aggregate effect                                   8.336***           3.990***
                                                             (1.975)            (1.383)
Number of observations                       4.623             4.623              4.623
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), *
(10%). Without much loss of generality, only the combined contribution of R&D and New
product/process development is shown.




                                            77
                                           Table A7: Aggregate quantile decomposition results by region
Dependent             (1)         (2)          (3)            (4)            (5)               (6)           (7)         (8)         (9)
variable: Sales
decline (%)
                     10th       20th           30th          40th              50th            60th         70th       80th        90th
                    quantile   quantile      quantile       quantile         quantile         quantile    quantile    quantile    quantile
                                                                       Panel A: SSA (N=1,087)
SMEs                5.970***   30.297***    38.349***      46.107***          55.565***      61.868***    68.108***   78.045***   90.591***
                     (1.525)    (1.659)      (1.535)        (1.346)            (1.199)        (1.344)       (1.679)    (1.773)     (1.995)
Large firms         9.719***   17.211***    25.266***      33.380***          37.500***      51.300***    59.642***   66.864***   77.387***
                     (2.951)    (2.652)      (2.791)        (3.011)            (2.936)        (2.562)       (2.562)    (3.629)     (3.407)
Difference           -3.750    13.086***    13.083***      12.727***          18.065***      10.568***     8.466***   11.182***   13.204***
(SME-Large
firms)              (3.535)     (2.203)      (2.441)        (2.872)            (2.348)          (2.407)    (2.723)     (3.505)     (3.532)
Total or            -0.250      11.119*       4.767        12.396**           12.342**           5.653      5.653     17.822**      7.450
Aggregate
endowment           (2.463)     (5.800)      (4.925)        (5.639)            (5.721)          (5.914)    (5.914)     (7.210)     (6.569)
effect
Total or             -3.499      1.967        8.316          0.331              5.723           4.915       2.813      -6.640       5.754
Aggregate
structural
 effect             (3.512)     (5.849)      (5.387)        (6.192)            (6.071)          (6.281)    (6.122)     (7.889)     (7.320)
Total or            -6.67%      84.97%       36.44%         97.40%             68.32%           53.49%     66.77%     159.38%      56.42%
Aggregate
endowment
effect (% of
total gap)
Total or            -93.33%     15.03%       63.56%          2.60%             31.68%           46.51%     33.23%     -59.38%      43.58%
Aggregate
structural effect
(% of total gap)
                                                                       Panel B: ECA (N=2,695)
SMEs                 1.528*    8.780***     10.733***      21.786***          30.820***      38.420***    45.799***   58.165***   77.768***


                                                                         78
                      (0.801)    (0.801)    (2.172)     (2.232)          (1.956)      (1.956)       (2.658)     (3.715)     (6.284)
Large firms         -6.946***   4.146***   9.082***   14.019***         16.165***    25.378***     33.802***   36.866***   54.506***
                      (1.296)    (1.103)    (1.103)     (1.103)          (2.909)      (3.224)       (3.331)     (3.416)     (6.831)
Difference           8.474***   4.634***     1.651     7.767***         14.655***    13.042***     11.997***   21.299***   23.262***
(SME-Large
firms)
                     (1.127)    (1.016)    (1.739)     (1.953)           (2.944)         (3.171)    (3.823)     (4.112)     (7.411)
Total or              1.589      1.416      1.416       1.416            7.706**         7.721**    6.769*       5.256       5.458
Aggregate
endowment
effect               (1.332)     (1.006)   (1.006)     (1.006)           (3.221)         (3.310)    (3.674)     (3.671)     (6.892)
Total or            6.885***    3.218***    0.235     6.351***          6.949***         5.320**    5.228*     16.043***   17.804***
Aggregate
structural effect
                     (1.614)    (1.177)    (2.052)     (2.250)           (2.367)         (2.347)    (3.137)     (3.237)     (4.488)
Total or             18.75%     30.56%     85.79%      18.23%            52.58%          59.20%     56.42%      24.68%      23.46%
Aggregate
endowment
effect (% of
total gap)
Total or             81.25%     69.44%     14.21%      81.77%            47.42%          40.80%     43.58%      75.32%      76.54%
Aggregate
structural effect
(% of total gap)
                                                                  Panel C: LAC (N=726)

SMEs                -8.852***   4.449***    1.051     21.742***         31.801***    42.164***     51.970***   61.586***   74.596***
                      (1.544)    (1.414)   (2.969)     (3.451)           (3.376)      (3.412)       (3.100)     (3.100)     (2.501)
Large firms         -9.751***    -4.584*    4.135     10.239***         11.606***    22.249***     30.349***   45.254***   60.806***
                      (1.416)    (2.677)   (2.722)     (2.722)           (3.463)      (4.221)       (3.619)     (3.763)     (7.148)
Difference             0.900    9.033***   -3.084     11.503***         20.195***    19.916***     21.622***   16.332***    13.790*
(SME-Large
firms)
                     (2.104)    (2.763)    (3.251)     (3.592)           (4.653)         (5.139)    (4.618)     (4.840)     (8.002)


                                                                   79
Total or             2.028      7.185*      8.137**      8.137**          13.718***   18.206***       10.198*     15.239*      9.717
Aggregate
endowment
effect              (3.847)     (4.028)      (3.505)     (3.505)           (4.798)         (4.713)    (6.135)     (8.254)     (8.480)
Total or            -1.497       1.848     -11.221***     2.366             6.476           1.710    11.423**      1.092       4.073
Aggregate
structural
 effect             (4.639)     (4.540)      (4.015)     (4.176)           (4.791)         (4.594)    (5.382)     (7.339)    (12.081)
Total or            225.5%      79.54%      263.82%      70.74%            67.93%          91.41%     47.17%      93.31%     70.46%
Aggregate
endowment
effect (% of
total gap)
Total or            -125.6%     20.46%     -363.82%      29.26%            32.07%          8.59%      52.83%      6.69%       29.54%
Aggregate
structural effect
(% of total gap)
                                                                   Panel D: MENA (N=444)
SMEs                8.614***   31.776***   42.508***    49.987***         55.537***   60.678***      67.336***   73.345***   85.873***
                     (1.345)    (3.986)     (2.671)      (2.671)           (2.257)     (2.139)        (2.139)      (2.550)    (2.144)
Large firms         8.203***   12.500***   33.918***    40.560***         46.718***   53.513***      59.954***   65.372***   83.180***
                     (1.385)    (3.564)     (3.017)      (1.660)           (1.660)     (2.344)        (2.344)      (2.269)    (9.747)
Difference            0.411    19.276***     8.590*      9.428**           8.820**      7.165*        7.382**     7.973***     2.693
(SME-Large
firms)
                    (2.318)     (5.949)     (4.386)      (3.956)           (3.487)         (3.724)    (3.724)     (2.642)     (8.787)
Total or            4.873**    18.300**    7.874***      4.816**           4.816**          1.186      1.186       1.625     10.100***
Aggregate
endowment
effect              (2.083)     (7.816)     (2.358)      (1.882)           (1.882)     (2.486)        (2.486)     (3.062)     (3.595)
Total or            -4.461       0.975       0.715        4.611             4.003      5.979**        6.196**     6.349*      -7.407
Aggregate
structural effect
                    (2.770)     (8.516)     (5.114)      (3.322)           (2.779)         (2.995)    (2.995)     (3.780)     (7.568)

                                                                     80
Total or           1184.87%     94.94%          91.67%         51.09%          54.61%           16.55%   16.07%   20.38%   375.07%
Aggregate
endowment
effect (% of
total gap)
Total or          -1084.87%      5.06%           8.33%         48.91%          45.39%           83.45%   83.93%   79.62%   -275.07%
Aggregate
structural effect
(% of total gap)
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), * (10%).




                                                                         81
                             Table A8: Quantile decomposition results for SSA (Percentage points)
Dependent           (1)            (2)          (3)         (4)           (5)         (6)             (7)        (8)         (9)
variable: Sales
decline (%)
                    10th          20th         30th         40th           50th        60th       70th           80th        90th
                  quantile      quantile     quantile     quantile       quantile    quantile   quantile       quantile    quantile
Total              -3.750      13.086***    13.083***    12.727***      18.065***   10.568***   8.466***      11.182***   13.204***
unconditional      (3.535)       (2.203)      (2.441)      (2.872)        (2.348)     (2.407)    (2.723)        (3.505)     (3.532)
gap
                                                          Panel A: Endowment effects
Country-          -0.815        5.638**      3.637*        3.692      4.902*       0.770             0.770      4.798      -2.553
Industry          (2.050)       (2.704)      (1.959)      (2.546)     (2.914)     (2.931)           (2.931)    (4.771)     (2.479)
Labor             -0.422        1.894*       2.853**     4.713***    5.552***     4.186**           4.186**     3.272       0.436
productivity      (0.426)       (1.014)      (1.455)      (1.539)     (2.133)     (1.662)           (1.662)    (3.137)     (1.961)
(logs)
Supply             2.123        4.257**      3.272**     3.823***       4.507***    4.396***    4.396***      2.692***     2.400**
disruption Y:1    (1.897)       (1.704)      (1.429)      (1.438)        (1.572)     (1.420)     (1.420)       (0.953)     (1.000)
N:0
Overdraft         -1.221          0.525       -0.923      -0.511          2.679       2.034          2.034     5.365**     5.863**
Y:1 N:0           (0.970)        (1.806)     (1.684)      (2.105)        (2.401)     (2.177)        (2.177)    (2.124)     (2.684)
Finance           -0.154          0.092       -0.073       0.075         -0.240      -0.079         -0.079      -0.186       0.067
obstacle (0-4)    (0.168)        (0.209)     (0.238)      (0.300)        (0.302)     (0.314)        (0.314)    (0.307)     (0.189)
Skills            -0.131          0.361        0.313       0.446         -0.078      -0.475         -0.475      -0.345      -0.533
obstacle (0-4)    (0.191)        (0.406)     (0.370)      (0.561)        (0.450)     (0.553)        (0.553)    (0.438)     (0.700)
Young              0.102         -1.035      -1.104*      -0.895         -1.148      -0.430         -0.430     1.631**       1.847
firm (<=10        (0.214)        (0.733)     (0.611)      (0.624)        (0.852)     (0.716)        (0.716)    (0.716)     (1.357)
years) Y:1 N:0
High share of      -0.186        -1.029       -0.703       0.582         -0.123       0.152          0.152     -0.009      -0.203
women workers
Y:1 N:0
                  (0.142)        (0.714)     (0.504)      (0.432)        (0.443)     (0.643)        (0.643)    (0.606)     (0.645)




                                                                   82
Exports            0.014     2.497*    0.899     0.311          -0.215      0.161       0.161      1.183    1.339
(proportion of
sales)
                   (0.307)   (1.362)   (1.097)   (0.798)       (1.172)     (0.961)     (0.961)    (1.261)   (1.257)
Foreign            -0.297     0.368     0.028     1.041        -1.287       0.049       0.049      1.803    -0.955
ownership
(proportion)
                   (0.431)   (1.838)   (1.173)   (0.818)       (1.127)     (1.354)     (1.354)    (1.642)   (2.444)
Tax rates are a    -0.119     0.096     0.001     0.009         0.002      -0.125      -0.125     -0.149     0.090
major obstacle
Y:1 N:0
                   (0.229)   (0.223)   (0.143)   (0.091)       (0.147)     (0.271)     (0.271)    (0.296)   (0.185)
Government         -0.032     0.019     0.079     0.144         0.110       0.116       0.116      0.127     0.407
support Y:1 N:0
                   (0.064)   (0.182)   (0.204)   (0.227)       (0.205)     (0.180)     (0.180)    (0.233)   (0.403)
Capital city Y:1    0.896     0.761     1.320     2.045         1.833       1.424       1.424      0.408     0.998
N:0
                   (0.852)   (0.832)   (1.029)   (1.288)       (1.305)     (1.051)     (1.051)    (0.877)   (1.248)
Obstacles           0.175     0.218    -0.279    -0.682        -0.514       0.055       0.055     -0.283     1.387
severity
                   (0.243)   (0.603)   (0.677)   (0.785)        (0.747)     (0.893)     (0.893)   (0.721)   (1.061)
Sole               -1.213    -2.859    -4.919    -3.168        -5.294**   -9.681***   -9.681***   -5.181    -2.187
proprietorship
Y:1 N:0
                   (1.472)   (3.927)   (4.221)   (4.256)       (2.485)     (3.425)     (3.425)    (3.210)   (2.519)
Shareholding        0.961    -0.119    -0.196     0.860         1.385      2.460*      2.460*     2.398**   -1.060
company (non-
traded) or
Partnership Y:1    (0.698)   (0.834)   (0.814)   (0.850)       (0.912)     (1.260)     (1.260)    (1.215)   (0.759)
N:0
R&D and New        0.068     -0.565    0.562     -0.087         0.272       0.639       0.639      0.298    0.105
product/process
                   (0.280)   (0.899)   (0.676)   (0.581)       (0.816)     (0.951)     (0.951)    (0.587)   (1.215)

                                                 Panel B: Structural effects

                                                          83
                      10th       20th       30th        40th        50th      60th        70th          80th       90th
                    quantile   quantile   quantile    quantile   quantile   quantile    quantile      quantile   quantile
Country-Industry   -33.888**    6.527     -20.438     -11.873     -22.217   47.655**    41.635**     80.737***   21.601*
                    (16.073)   (19.168)   (18.955)    (19.325)   (20.508)   (19.794)    (18.970)      (21.653)   (12.662)
Labor                 0.107      8.886     18.380    40.518***   47.384**    27.567*     30.009*       14.984    -13.821
productivity
(logs)
                    (18.239)   (12.412)   (19.393)   (14.960)    (19.248)    (15.439)     (15.509)   (34.337)    (24.423)
Supply             52.042***    10.572     9.900      -1.666      -6.156    -10.216**   -14.819***   -7.144*      -7.447
disruption Y:1
N:0
                   (13.540)    (8.930)    (7.387)     (6.199)    (5.999)     (4.297)     (4.222)      (3.902)    (5.013)
Overdraft Y:1      -2.682**     0.281     -1.489      -0.274      2.928       2.985       2.897       6.174**    6.886**
N:0
                    (1.161)    (2.077)    (1.852)     (2.042)    (2.275)     (2.140)     (2.131)      (2.490)    (2.803)
Finance obstacle     2.877     -2.325      2.587       0.566      6.773       5.751       6.138       8.596*      6.363
(0-4)
                    (3.246)    (4.803)    (5.157)     (5.597)    (4.271)     (6.073)     (5.588)      (4.487)    (4.117)
Skills obstacle      1.644     4.267*     4.501**      3.981     -0.209      -3.215      -4.120       -2.676     -4.602
(0-4)
                    (2.794)    (2.553)    (2.082)     (3.096)    (4.203)     (2.504)     (3.032)       (2.512)    (4.079)
Young firm          -1.179      2.342      1.576       1.154      1.988       0.271      -0.483      -6.107***   -6.617**
(<=10 years) Y:1
N:0
                    (1.268)    (1.864)    (1.466)     (1.640)    (2.060)     (1.618)     (2.222)      (1.186)    (3.258)
High share of       2.749**     5.592     4.808*      -0.391      3.531       2.593       2.252        3.585      3.236
women workers
Y:1 N:0
                    (1.232)    (3.462)    (2.683)     (2.207)    (2.502)     (3.540)     (3.504)      (2.384)    (2.903)
Exports              0.561     1.436**     0.769       0.100     -0.123       0.038       0.244        0.842      1.061
(proportion of
sales)
                    (0.481)    (0.661)    (0.579)     (0.471)    (0.607)     (0.558)     (0.533)      (0.661)    (0.753)



                                                            84
 Foreign              -0.672        -0.860       -0.968        -0.454        -1.598*      -0.509       -0.925         0.957        -0.738
 ownership
 (proportion)
                      (1.421)      (1.093)       (0.779)      (0.514)        (0.822)     (0.844)       (1.115)       (1.011)       (1.375)
 Tax rates are a      -0.054       -3.258        -0.188       -0.754         -0.695       2.091         0.987         0.686        -1.940
 major obstacle
 Y:1 N:0
                      (1.370)      (2.476)       (2.434)      (1.651)        (2.120)     (2.126)       (2.030)       (2.359)      (1.883)
 Government           -0.709       -0.543        -0.158       -0.019         -0.045      -0.273        -0.082        -0.003       0.905**
 support Y:1 N:0
                      (0.504)      (0.539)       (0.535)      (0.480)        (0.449)     (0.385)       (0.371)       (0.510)       (0.408)
 Capital city Y:1      3.436        0.809         3.526       5.215*          3.911       2.385         3.321         0.287         0.614
 N:0
                      (2.359)      (3.200)       (3.283)      (2.758)        (4.271)     (2.988)       (3.118)       (4.276)      (3.796)
 Obstacles            -6.419        2.748        -3.737       -4.562         -2.424       0.816         4.177         1.027       14.848*
 severity
                      (4.598)      (8.058)       (7.421)      (7.620)        (7.567)     (9.329)      (9.079)        (5.979)       (8.527)
 Sole                 -1.531        2.969         5.778        3.151         6.125*     11.037***    11.548***        6.504         0.681
 proprietorship
 Y:1 N:0
                      (3.654)      (4.954)       (4.899)      (4.910)        (3.338)     (4.070)       (3.954)       (4.365)       (2.769)
 Shareholding         -1.136       -0.726        -0.375        0.372          1.841       3.811         3.739         3.535       -3.833**
 company (non-
 traded) or
 Partnership Y:1      (2.464)      (1.927)       (1.843)      (1.212)        (1.352)     (2.450)       (2.325)       (2.365)       (1.564)
 N:0
 R&D and New          2.045*        -1.901        0.232        -1.758        -0.017       -0.013        0.617         0.279        0.145
 product/process
                      (1.211)       (1.914)      (1.218)       (1.588)        (2.209)      (2.052)      (2.048)      (1.873)       (1.711)
 Constant            -20.689       -34.849      -16.390       -32.976        -35.275    -87.858***   -84.323***    -118.902**     -11.589
                     (37.580)      (33.643)     (39.979)      (34.687)       (33.802)     (25.831)     (25.463)     (47.202)      (33.827)
 Observations          1,087        1,087        1,087         1,087          1,087       1,087          1,087        1,087         1,087
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), * (10%).. Without much loss of generality, only
the combined contribution of R&D and New product/process development is shown.

                                                                        85
                                       Table A9: Quantile decomposition results for ECA (Percentage points)
Dependent variable: Sales decline     (1)         (2)          (3)          (4)         (5)         (6)          (7)         (8)         (9)
(%)
                                      10th       20th          30th        40th        50th      60th            70th        80th        90th
                                    quantile   quantile      quantile    quantile    quantile  quantile        quantile    quantile    quantile
Total unconditional gap             8.474***   4.634***        1.651     7.767*** 14.655*** 13.042***         11.997***   21.299***   23.262***
                                     (1.127)    (1.016)       (1.739)     (1.953)     (2.944)   (3.171)         (3.823)     (4.112)     (7.411)
                                                                           Panel A: Endowment effects
Country-Industry                     -0.267      0.179        0.179        0.179       1.733     1.526          2.962        3.538      6.453
                                     (0.722)    (0.543)      (0.543)      (0.543)     (1.863)   (1.990)        (2.443)     (2.732)     (5.520)
Labor productivity (logs)             0.099      0.060        0.060        0.060       0.697     0.549          1.117        0.663      1.339
                                     (0.207)    (0.166)      (0.166)      (0.166)     (0.529)   (0.491)        (0.758)     (0.601)     (1.060)
Supply disruption Y:1 N:0           1.176***   0.920***     0.920***     0.920*** 5.181***     4.722***       3.654***    2.776***    2.973***
                                     (0.376)    (0.305)      (0.305)      (0.305)     (1.451)   (1.309)        (1.027)     (0.835)     (0.962)
Overdraft Y:1 N:0                    0.619*     0.474*       0.474*       0.474*       0.244     0.436          0.712        0.338     -0.751
                                     (0.349)    (0.253)      (0.253)      (0.253)     (0.354)   (0.479)        (0.436)     (0.411)     (0.849)
Finance obstacle (0-4)               -0.005     -0.004       -0.004       -0.004      -0.014    -0.013         -0.009       -0.008     -0.030
                                     (0.058)    (0.051)      (0.051)      (0.051)     (0.175)   (0.165)        (0.120)     (0.106)     (0.380)
Skills obstacle (0-4)                -0.283     -0.224       -0.224       -0.224      -0.243    -0.249         -0.455      -0.932*     -0.879
                                     (0.254)    (0.174)      (0.174)      (0.174)     (0.350)   (0.335)        (0.462)     (0.513)     (0.747)
Young firm (<=10 years) Y:1 N:0       0.229      0.141        0.141        0.141      -0.247     0.136         -0.497       -0.197     -0.459
                                     (0.253)    (0.175)      (0.175)      (0.175)     (0.550)   (0.627)        (0.546)     (0.386)     (0.842)
High share of women workers          -0.054     -0.060       -0.060       -0.060      -0.086    -0.100         -0.117       -0.059     -0.048
Y:1 N:0                              (0.115)    (0.123)      (0.123)      (0.123)     (0.191)   (0.211)        (0.242)     (0.137)     (0.126)
Exports (proportion of sales)        -0.337     -0.378       -0.378       -0.378       0.256    -0.002         -0.802       -0.328     -0.316
                                     (0.810)    (0.563)      (0.563)      (0.563)     (0.808)   (0.770)        (0.863)     (0.744)     (1.154)
Foreign ownership (proportion)        0.156      0.074        0.074        0.074      -0.453    -0.004         -0.106       -0.340     -0.529
                                     (0.222)    (0.172)      (0.172)      (0.172)     (0.502)   (0.413)        (0.443)     (0.520)     (0.642)
Tax rates are a major obstacle        0.272      0.239        0.239        0.239       0.337     0.326          0.306        0.322      0.284
Y:1 N:0                              (0.247)    (0.208)      (0.208)      (0.208)     (0.323)   (0.311)        (0.272)     (0.296)     (0.385)
Government support Y:1 N:0           -0.207     -0.146       -0.146       -0.146       0.233     0.257          0.251        0.147     -0.099

                                                                        86
                                  (0.189)    (0.137)     (0.137)          (0.137)   (0.218)    (0.246)      (0.268)     (0.271)     (0.388)
Capital city Y:1 N:0              -0.005     -0.006      -0.006           -0.006     0.045      0.052        0.025      -0.008       -0.068
                                  (0.037)    (0.027)     (0.027)          (0.027)   (0.100)    (0.110)      (0.072)     (0.065)     (0.148)
Obstacles severity                 0.002      0.002       0.002            0.002     0.004     -0.002       -0.004      -0.001       -0.015
                                  (0.037)    (0.037)     (0.037)          (0.037)   (0.079)    (0.041)      (0.089)     (0.020)     (0.299)
Sole proprietorship Y:1 N:0       -0.062      0.091       0.091            0.091     0.095     -0.427       -0.656      -0.700      -2.566*
                                  (0.611)    (0.451)     (0.451)          (0.451)   (0.852)    (0.668)      (0.582)     (0.546)     (1.349)
Shareholding company (non-         0.197      0.128       0.128            0.128     0.248      0.353        0.113      -0.059       -0.457
traded) or Partnership Y:1 N:0    (0.213)    (0.169)     (0.169)          (0.169)   (0.370)    (0.404)      (0.351)     (0.383)     (0.578)
R&D and New product/process        0.057     -0.072      -0.072           -0.072    -0.323      0.160        0.275       0.104        0.627
                                  (0.288)    (0.217)     (0.217)          (0.217)   (0.341)    (0.277)      (0.322)     (0.314)     (0.530)

                                                                          Panel B: Structural effects
                                     10th      20th         30th         40th        50th         60th        70th        80th        90th
                                  quantile   quantile     quantile     quantile    quantile     quantile    quantile    quantile    quantile
Country-Industry                    7.241*     4.754    -20.424***    19.166*** 60.071*** 58.282***        55.206***   62.730***   61.624***
                                   (3.862)    (3.085)      (4.843)      (4.232)     (6.268)      (6.282)     (7.673)     (8.811)    (16.454)
Labor productivity (logs)           -1.932    -3.184     -18.935**     -13.611       2.165       -2.481       1.595     -21.039     -17.083
                                   (6.597)    (5.884)      (9.014)     (10.832)    (16.167)     (17.730)    (15.626)    (16.149)    (32.933)
Supply disruption Y:1 N:0            0.625     1.888    20.906***     20.399*** -5.741*** -3.486**            0.355     4.058**     9.115**
                                   (1.189)    (1.181)      (1.826)      (1.766)     (1.767)      (1.412)     (1.297)     (1.760)     (3.795)
Overdraft Y:1 N:0                 1.854***   1.548***       0.894        1.110       1.022        1.429     2.252**       1.463      -3.519
                                   (0.637)    (0.479)      (0.722)      (0.691)     (0.897)      (0.918)     (0.894)     (0.982)     (2.147)
Finance obstacle (0-4)               0.645     0.554       -0.590        0.197      2.557*       2.441*       1.714     3.285**     8.062**
                                   (0.748)    (0.612)      (0.969)      (1.023)     (1.525)      (1.394)     (1.154)     (1.485)     (3.443)
Skills obstacle (0-4)              -1.652*   -1.410**    -2.260***     -1.580*      -1.135       -1.158      -1.454      -3.228      -2.288
                                   (0.980)    (0.678)      (0.796)      (0.939)     (1.638)      (1.621)     (2.259)     (2.119)     (3.508)
Young firm (<=10 years) Y:1 N:0     -0.375    -0.198       -0.393        0.045       1.450        0.681       2.263      1.791*      4.481*
                                   (0.568)    (0.446)      (0.608)      (0.583)     (1.399)      (1.540)     (1.382)     (1.051)     (2.678)
High share of women workers          1.901    2.112*        2.345        1.077       3.344       3.817*     3.864**       1.877       7.454
Y:1 N:0                            (1.534)    (1.188)      (1.628)      (1.821)     (3.077)      (2.308)     (1.807)     (2.185)     (5.206)

                                                                     87
Exports (proportion of sales)        -0.023       -0.045        -0.080        -0.014       0.251        0.116      -0.014        0.464        1.846*
                                     (0.448)      (0.327)       (0.379)       (0.374)     (0.467)      (0.436)     (0.459)      (0.553)      (1.095)
Foreign ownership (proportion)        0.093        0.060         0.007        -0.130      -0.161        0.023      -0.025       -0.132         0.065
                                     (0.122)      (0.106)       (0.200)       (0.231)     (0.294)      (0.243)     (0.327)      (0.367)      (0.610)
Tax rates are a major obstacle     -2.435***    -2.193***     -1.979***      -1.395**     -1.851       -1.771      -1.313       -1.595        -1.907
Y:1 N:0                              (0.846)      (0.633)       (0.685)       (0.586)     (1.180)      (1.088)     (1.118)      (1.429)      (2.522)
Government support Y:1 N:0          1.397**       1.045*       2.098***      2.100***     -0.206       -0.347      -0.620       -0.535         2.201
                                     (0.701)      (0.622)       (0.741)       (0.761)     (0.924)      (1.038)     (1.150)      (1.701)      (2.706)
Capital city Y:1 N:0                 -0.215       -0.198         0.487         0.459       0.064       -0.039       0.534        0.724         1.725
                                     (0.554)      (0.425)       (0.573)       (0.517)     (1.164)      (1.170)     (0.968)      (0.881)      (1.642)
Obstacles severity                    1.298        1.301         1.272         0.587       0.767       -0.818      -0.167       -1.561        -4.667
                                     (1.758)      (1.200)       (1.418)       (1.303)     (2.059)      (1.861)     (2.267)      (1.962)      (4.505)
Sole proprietorship Y:1 N:0          -0.058       -0.271         0.454         0.305       0.647        1.374       1.544        1.517       5.787**
                                     (0.945)      (0.776)       (0.813)       (0.645)     (1.211)      (1.083)     (0.993)      (0.984)      (2.510)
Shareholding company (non-           -1.786       -1.224         1.560         1.463      -0.385       -1.240       0.038        0.298         5.382
traded) or Partnership Y:1 N:0       (2.200)      (2.083)       (1.874)       (1.898)     (3.629)      (3.731)     (3.518)      (4.859)      (6.845)
R&D and New product/process           0.410       -0.118        -0.493        -0.712     -1.838*       -1.253      -0.927      -1.606*        -0.595
                                     (0.753)      (0.636)       (0.667)       (0.810)     (0.990)      (0.984)     (0.993)      (0.949)      (1.989)
Constant                             -0.103       -1.202       15.364**     -23.114**   -54.07***    -50.25***   -59.62***    -32.471**     -59.878*
                                     (7.208)      (6.247)       (7.350)       (9.210)    (15.115)     (17.673)    (18.197)     (14.620)     (31.925)
Observations                         2,695        2,695         2,695          2,695      2,695           2,695      2,695        2,695         2,695
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), * (10%).. Without much loss of generality, only the combined
contribution of R&D and New product/process development is shown.




                                                                          88
                               Table A10: Quantile decomposition results for LAC (Percentage points)
Dependent variable:        (1)        (2)         (3)         (4)           (5)        (6)           (7)        (8)        (9)
Sales decline (%)
                           10th       20th        30th         40th         50th       60th         70th        80th       90th
                         quantile   quantile    quantile     quantile     quantile   quantile     quantile    quantile   quantile
Total                      0.900    9.033***     -3.084     11.503*** 20.195*** 19.916***        21.622***   16.332***   13.790*
unconditional gap         (2.104)    (2.763)     (3.251)      (3.592)      (4.653)    (5.139)      (4.618)     (4.840)    (8.002)
                                                                 Panel A: Endowment effects
Country-Industry         -0.029      -1.128      -0.984       -0.984        0.804      3.854        3.073     7.782**    10.036*
                         (1.343)    (2.320)     (2.004)       (2.004)      (2.138)    (2.532)      (2.264)    (3.166)    (5.205)
Labor productivity       2.127*     2.098**     2.299**      2.299**        1.477   5.085***        1.845      0.259      0.190
(logs)
                          (1.211)    (0.935)     (1.069)     (1.069)     (1.056)      (1.557)     (2.916)     (2.463)     (3.547)
Supply disruption Y:1    2.510***   4.521***    5.062***    5.062***    10.031***    10.472***   10.877***   11.443***   10.705**
N:0
                         (0.797)    (1.158)     (1.258)      (1.258)      (2.356)     (2.398)      (2.612)    (2.702)    (4.224)
Overdraft Y:1 N:0         0.277       0.484      0.426        0.426        0.612        0.098      -0.119      0.473     -3.232
                         (0.949)    (0.995)     (1.104)      (1.104)      (0.926)     (1.122)      (1.112)    (1.620)    (3.153)
Finance obstacle (0-4)   -0.800     1.899**      0.881        0.881        2.300      3.531**      3.746*      3.838      4.531
                         (1.018)    (0.836)     (0.917)      (0.917)      (1.415)     (1.740)      (2.166)    (2.678)    (4.262)
Skills obstacle (0-4)    -0.150      -0.028     -0.042       -0.042       -0.148       -0.222      -0.200     -0.198      0.033
                         (0.486)    (0.097)     (0.144)      (0.144)      (0.476)     (0.712)      (0.646)    (0.657)    (0.212)
Young firm (<=10          0.021      -0.060     -0.038       -0.038        0.003        0.076      -0.172     -0.119     -0.106
years) Y:1 N:0
                         (0.088)    (0.253)     (0.162)      (0.162)      (0.029)     (0.322)      (0.719)    (0.504)    (0.447)
High share of women       0.009      0.011      -0.008       -0.008        0.006       0.046       -0.026     -0.013     -0.090
workers
Y:1 N:0                  (0.070)    (0.073)     (0.051)      (0.051)      (0.051)     (0.288)      (0.190)    (0.147)    (0.565)
Exports (proportion of    1.896      1.375       0.787        0.787       3.095**      1.397       -1.531     -1.875     -3.022
sales)
                         (1.467)    (1.488)     (1.462)      (1.462)      (1.422)     (0.983)      (1.524)    (1.532)    (3.368)
Foreign ownership        -0.316     -0.164       0.251        0.251       -2.248*     -2.696       -3.837     -3.401     -0.971
(proportion)

                                                                 89
                         (0.857)    (1.069)     (1.101)     (1.101)      (1.296)      (2.039)      (2.791)     (3.073)    (5.871)
Tax rates are a major     0.360      0.163       0.321       0.321       -0.178        0.309        0.625      -0.534      0.168
obstacle
Y:1 N:0                  (0.414)    (0.359)     (0.425)     (0.425)      (0.380)      (0.509)      (0.817)     (0.865)    (1.021)
Government support        0.028     -0.358       0.112       0.112       -0.202       -0.719       -1.314      -1.238     -5.384
Y:1 N:0
                         (0.845)    (0.645)     (0.645)     (0.645)      (0.852)      (1.073)      (1.561)     (1.801)    (3.938)
Capital city Y:1 N:0      0.193      0.176       0.717       0.717        0.396        0.445        0.041      -0.192     -0.457
                         (0.507)    (0.471)     (0.459)     (0.459)      (0.646)      (0.677)      (0.843)     (1.383)    (1.719)
Obstacles severity       -0.114     -0.054      -0.093      -0.093       -0.259       -0.347       -0.274      -0.063     -0.156
                         (0.355)    (0.179)     (0.285)     (0.285)      (0.765)      (1.032)      (0.821)     (0.296)    (0.504)
Sole proprietorship      -2.430      0.542      -0.328      -0.328       -0.643       -1.491       -2.387       2.307     -4.605
Y:1 N:0                  (2.050)    (1.705)     (1.409)     (1.409)      (1.333)      (2.221)      (2.135)     (3.472)    (7.886)
Shareholding company      0.534     -1.403      -0.188      -0.188       -0.309       -0.531        2.066      -0.202      3.150
(non-
traded) or Partnership   (1.444)    (1.231)     (1.135)     (1.135)      (0.973)      (1.444)      (1.449)     (2.055)    (5.626)
Y:1 N:0
R&D and New              -2.087*    -0.892      -1.036      -1.036       -1.020       -1.102       -2.213      -3.025     -1.074
product/process          (1.151)    (1.555)     (1.496)     (1.496)      (1.761)      (1.443)      (2.200)     (2.677)    (3.211)

                                                                 Panel B: Structural effects
                            10th      20th        30th        40th          50th         60th         70th       80th       90th
                         quantile   quantile    quantile    quantile      quantile     quantile    quantile    quantile   quantile
Country-Industry         -10.862*    -6.123    85.794***   82.934*** 76.605*** 67.892***           77.36***   112.85***   -40.783
                          (5.545)    (8.199)    (11.283)     (9.502)       (9.241)      (9.319)    (10.092)    (14.374)    (20.77)
Labor productivity        29.876     18.118     32.473**     -5.519       -28.865      36.547*      -20.170    -51.087    -97.703
(logs)                   (27.565)   (22.069)    (16.367)    (23.951)      (22.471)     (20.165)    (53.857)    (47.813)   (75.767)
Supply disruption         7.335**   7.262**    19.980***   23.595***        2.650      -6.219*    -11.41***        -      -10.537
                                                                                                              13.045***
Y:1 N:0                  (3.253)    (3.448)     (2.913)     (3.223)      (3.106)      (3.659)      (3.920)      (4.011)   (8.923)
Overdraft Y:1 N:0         0.738      1.096       2.047       1.136        0.685       -0.410       -0.074        1.079    -8.450
                         (2.080)    (1.875)     (2.665)     (3.119)      (2.417)      (2.638)      (3.112)      (3.749)   (7.526)


                                                                90
Finance obstacle           1.984    -3.920**   -0.258     0.578      -3.285      -6.919     -9.024   -9.263     -15.655
(0-4)                     (2.677)    (1.753)   (1.721)   (2.141)     (4.082)     (5.200)   (6.839)   (7.889)    (11.861)
Skills obstacle (0-4)    -9.295**    -5.256*   -5.193    -3.235     -8.530**   -10.579**   -8.871*   -8.804      12.205
                          (4.185)    (2.941)   (3.759)   (3.315)     (3.622)     (5.127)   (5.309)   (8.104)     (7.876)
Young firm (<=10          -0.007      0.773*    0.465     0.569       0.165      -0.683    1.237**    0.842       0.448
years) Y:1 N:0            (0.299)    (0.447)   (0.392)   (0.353)     (0.360)     (0.511)   (0.504)   (0.676)     (0.566)
High share of women       -0.079       1.411    1.543     0.984       2.100       5.356      0.572    1.441      -1.006
workers
Y:1 N:0                  (2.725)    (1.756)    (2.380)    (2.670)   (3.315)    (4.217)     (6.792)   (8.241)     (8.207)
Exports (proportion       -0.425    -0.245      -0.089     0.362    0.723**    0.901**      0.584     0.531      1.215**
of sales)                (0.643)    (0.466)    (0.322)    (0.490)   (0.342)    (0.371)     (0.417)   (0.373)     (0.506)
Foreign ownership          0.578    -0.124      -0.157     0.302     -0.601     -0.931     -0.920    -0.795       -0.251
(proportion)             (0.571)    (0.479)    (0.469)    (0.500)   (0.461)    (0.675)     (0.844)   (0.911)     (1.815)
Tax rates are a major    -2.904*    -0.909     -2.880*   -3.586**    -1.146     -1.566     -2.081     3.545        5.079
obstacle
Y:1 N:0                  (1.592)    (1.688)    (1.583)    (1.691)   (2.068)     (2.557)    (3.339)    (3.671)     (4.792)
Government               -0.434     -0.004      -1.830    -1.395     0.622       0.673      1.001      0.830    11.785***
support Y:1 N:0          (1.889)    (1.873)    (1.755)    (2.295)   (2.122)     (2.121)    (3.129)    (3.784)     (4.449)
Capital city Y:1         -0.170      0.573     4.210**   5.341***    3.835       5.955      4.678      3.534       2.421
N:0                      (2.616)    (2.245)    (1.807)    (1.795)   (3.836)     (4.328)    (5.419)    (7.499)     (9.508)
Obstacles severity        4.698      0.902       2.138     2.509    8.936*       9.407      4.261     -5.978     -11.573
                         (5.746)    (6.564)    (4.334)    (4.969)   (5.382)     (9.417)    (7.768)   (11.408)    (11.262)
Sole proprietorship      6.564*     -0.098       4.674     4.461     3.758       2.569      2.511     -4.778      -0.312
Y:1 N:0                  (3.540)    (2.629)    (3.171)    (3.346)   (3.591)     (3.911)    (3.686)    (6.189)    (14.748)
Shareholding company      4.138     -1.045       3.162     1.169    -1.843      -4.413     -0.448     -5.454      -3.514
(non-
traded) or Partnership   (3.396)    (2.712)    (2.891)   (3.882)    (2.939)     (3.884)    (3.357)   (4.989)    (14.723)
Y:1 N:0
R&D and New              -2.602     -0.974     -2.257    -2.610     -5.081      -4.585     -2.864    -3.628     -13.432**
product/process          (4.631)    (4.199)    (3.913)   (4.065)    (3.568)     (3.444)    (3.732)   (4.256)      (5.332)




                                                            91
Constant                 -30.260      -9.589        -             -         -44.253         -         -24.919      -20.728    174.137**
                                                155.04***     104.229**                 91.288***
                                                                  *
                         (34.989)    (26.008)    (25.232)      (28.20)      (29.498)     (24.665)     (61.687)    (66.273)     (88.642)
Observations               726         726          726          726          726          726           726         726          726
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), * (10%). Without much loss of generality, only
the combined contribution of R&D and New product/process development is shown.




                                                                   92
                                       Table A11: Quantile decomposition results for MENA (Percentage points)
Dependent variable:           (1)            (2)         (3)         (4)            (5)         (6)           (7)       (8)        (9)
Sales decline (%)
                              10th           20th       30th         40th          50th        60th          70th       80th       90th
                            quantile       quantile   quantile     quantile      quantile    quantile      quantile   quantile   quantile
Total unconditional gap       0.411       19.276***    8.590*      9.428**       8.820**      7.165*       7.382**    7.973***     2.693
                             (2.318)        (5.949)    (4.386)      (3.956)       (3.487)     (3.724)       (3.724)    (2.642)    (8.787)
                                                                          Panel A: Endowment effects
Country-Industry              0.452         -1.009       -2.504    -1.597*       -1.597*      -0.956         -0.956    1.604     8.993**
                             (0.723)       (1.970)      (1.831)     (0.828)       (0.828)     (0.846)       (0.846)   (1.547)    (3.597)
Labor productivity (logs)    -0.050          0.648        0.221      0.169         0.169       0.504          0.504    0.792       0.969
                             (0.200)       (1.848)      (0.657)     (0.495)       (0.495)     (1.399)       (1.399)   (2.193)    (2.705)
Supply disruption Y:1       1.603**       8.608***     3.947***    2.169**       2.169**       1.378          1.378   2.499*       2.257
N:0                          (0.690)       (3.285)      (1.526)     (1.033)       (1.033)     (1.102)       (1.102)   (1.389)    (1.443)
Overdraft Y:1 N:0           3.293***         4.483        1.800      1.218         1.218      -1.545         -1.545   -2.713       6.011
                             (1.002)       (7.215)      (2.971)     (2.622)       (2.622)     (3.445)       (3.445)   (4.863)    (4.604)
Finance obstacle (0-4)       -0.308          0.769        0.233     -0.232        -0.232      -1.347         -1.347   -3.760      -2.609
                             (0.453)       (1.266)      (0.678)     (0.705)       (0.705)     (1.409)       (1.409)   (3.639)    (2.779)
Skills obstacle (0-4)        -0.541          2.353        1.796      1.230         1.230       0.812          0.812    1.033       1.714
                             (0.705)       (2.761)      (1.677)     (1.078)       (1.078)     (0.972)       (0.972)   (1.218)    (1.735)
Young firm (<=10 years)       0.524          1.610        0.599      0.470         0.470      -0.401         -0.401   -2.138      -0.565
Y:1 N:0                      (0.582)       (2.013)      (1.989)     (1.010)       (1.010)     (1.282)       (1.282)   (1.543)    (0.799)
High share of women           0.527         -0.348        0.762      0.774         0.774      -0.292         -0.292    0.560       0.446
workers Y:1 N:0              (0.754)       (0.978)      (0.888)     (0.716)       (0.716)     (0.483)       (0.483)   (1.673)    (1.796)
Exports (proportion of       -0.762          1.166        0.675      0.646         0.646       1.135          1.135   4.029*       2.942
sales)                       (0.587)       (2.346)      (1.304)     (0.681)       (0.681)     (1.476)       (1.476)   (2.372)    (3.085)
Foreign ownership             0.537       -4.786**    -3.104***   -3.164***     -3.164***     -1.637         -1.637   -0.594       3.142
(proportion)                 (0.700)       (1.929)      (0.834)     (1.127)       (1.127)     (1.558)       (1.558)   (2.333)    (2.918)
Tax rates are a major         0.066         -0.292       -0.169     -0.199        -0.199      -0.105         -0.105   -0.046       0.367
obstacle Y:1 N:0             (0.327)       (1.451)      (0.837)     (0.990)       (0.990)     (0.538)       (0.538)   (0.261)    (1.829)
Government support Y:1       -0.359        -2.686*      -1.558*    -1.831*       -1.831*     -1.661*        -1.661*   -0.601      -1.017


                                                                        93
N:0                          (0.252)     (1.459)     (0.861)     (0.989)      (0.989)         (0.940)      (0.940)       (0.383)      (0.939)
Capital city Y:1 N:0         -0.052      -0.143       -0.027       0.015        0.015          0.001        0.001         0.020        0.043
                             (0.629)     (1.737)     (0.326)     (0.186)      (0.186)         (0.024)      (0.024)       (0.253)      (0.530)
Obstacles severity            0.122       0.808        0.533       0.546        0.546          0.543        0.543         0.911        0.220
                             (0.436)     (2.674)     (1.712)     (1.771)      (1.771)         (1.756)      (1.756)       (2.910)      (0.912)
Sole proprietorship Y:1       0.394       4.464      5.458**     5.456**      5.456**          6.015        6.015         1.814     -10.392**
N:0                          (2.454)     (5.872)     (2.382)     (2.756)      (2.756)         (3.830)      (3.830)       (6.194)      (5.249)
Shareholding company         -0.911       0.849       -1.025      -1.459       -1.459         -1.491       -1.491      -2.899***     -3.037**
(non-traded) or
Partnership Y:1 N:0          (2.313)     (3.564)     (0.775)     (1.806)       (1.806)        (1.602)      (1.602)      (0.917)      (1.349)
R&D and New                   0.339       1.806       0.237       0.605         0.605          0.233        0.233        1.114        0.617
product/process
                             (0.646)     (1.275)     (0.543)     (0.505)       (0.505)        (0.246)      (0.246)      (0.746)      (1.389)

                                                                          Panel B: Structural effects
                                10th       20th        30th        40th           50th           60th         70th         80th        90th
                              quantile   quantile    quantile    quantile       quantile       quantile     quantile    quantile     quantile
Country-Industry              -10.903     -7.644      -4.080     -9.830**     -18.303***     -14.845***   -14.845***   -27.727***     -5.434
                               (6.863)    (7.410)     (6.230)     (4.628)        (4.886)        (2.925)      (2.925)     (6.720)     (15.518)
Labor productivity (logs)   -27.582***    37.807      11.282       6.991         10.883       46.860***    46.860***   77.674***     76.093*
                               (6.262)   (41.352)    (21.516)    (14.267)       (12.609)        (8.960)      (8.960)    (14.586)     (40.126)
Supply disruption Y:1        29.102***     2.955    12.919***   27.370***      13.552***       12.231*      12.231*       -3.286      -1.169
N:0                           (10.516)    (8.847)     (4.533)     (5.787)        (5.155)        (6.726)      (6.726)     (7.984)      (9.252)
Overdraft Y:1 N:0               1.753      0.921      -0.490      -0.934         -0.629         -2.874       -2.874       -2.738       3.677
                               (1.762)    (5.164)     (2.127)     (1.787)        (1.636)        (2.306)      (2.306)     (3.169)      (2.662)
Finance obstacle (0-4)         -3.199     -7.523      -7.610      -3.024          0.922       15.399***    15.399***   34.476***     21.357**
                               (5.984)   (12.934)     (6.942)     (8.597)        (8.160)        (4.475)      (4.475)     (4.834)     (10.272)
Skills obstacle (0-4)          -0.216     12.574       7.306       3.490         -1.243          3.092        3.092        9.683    18.459***
                               (7.054)   (10.709)     (5.328)     (2.827)        (4.837)        (5.212)      (5.212)     (6.673)      (6.584)
Young firm (<=10 years)        -0.366     -1.780      -0.946      -0.699         -0.273          0.710        0.710      6.063**       2.445
Y:1 N:0                        (1.732)    (4.055)     (4.077)     (2.228)        (1.844)        (2.166)      (2.166)     (2.874)      (1.750)


                                                                       94
High share of women           3.497         -1.618        3.801      3.871        5.102*         -1.259         -1.259         6.814         4.838
workers Y:1 N:0              (3.742)       (4.539)      (2.331)     (2.464)       (2.773)        (2.606)        (2.606)       (9.241)       (9.747)
Exports (proportion of       -0.119         -0.110       -0.301     -0.312         0.094         -0.027         -0.027         0.593         0.254
sales)                       (0.589)       (0.796)      (0.578)     (0.374)       (0.444)        (0.442)        (0.442)       (0.464)       (0.435)
Foreign ownership             0.791       -2.068**     -1.418**    -1.442*       -1.143*         -0.170         -0.170        -0.001         2.636
(proportion)                 (1.015)       (0.912)      (0.569)     (0.797)       (0.588)        (0.776)        (0.776)       (1.379)       (1.834)
Tax rates are a major      -3.687***         5.815       1.898*      2.698         3.110          3.778          3.778         2.171        -7.457
obstacle Y:1 N:0             (1.374)       (4.350)      (1.056)     (2.099)       (2.386)        (2.949)        (2.949)       (2.320)       (6.349)
Government support Y:1       -0.412      -8.180***    -3.486***   -4.415***     -4.700***      -4.809***      -4.809***       -1.035        -1.202
N:0                          (0.583)       (1.142)      (1.099)     (0.950)       (1.029)        (1.606)        (1.606)       (1.622)       (2.508)
Capital city Y:1 N:0         -0.324         -2.796       -0.261      0.751         0.758          0.510          0.510         0.721        2.487*
                             (0.667)       (2.024)      (0.734)     (1.193)       (1.260)        (0.670)        (0.670)       (1.497)       (1.350)
Obstacles severity           -1.244        -32.682      -11.243    -11.825       -19.195       -28.499**      -28.499**     -46.474***      -9.874
                             (9.678)      (31.787)     (10.748)    (15.730)      (16.897)       (14.277)       (14.277)      (11.896)      (23.953)
Sole proprietorship Y:1       2.092        -10.803    -12.68***   -12.68***     -12.672**        -8.383         -8.383        -3.220       25.757**
N:0                          (4.182)       (9.889)      (3.851)     (4.414)       (5.394)        (8.695)        (8.695)      (13.383)      (10.953)
Shareholding company         -0.828          0.437     -4.265**     -5.045        -3.641         -1.831         -1.831        -2.640        -0.157
(non-traded) or
Partnership Y:1 N:0          (3.683)       (4.347)     (1.722)     (3.273)       (2.540)        (3.659)        (3.659)        (2.997)       (3.510)
R&D and New                   0.442        1.634*       0.686      0.954**       1.075**        0.719**        0.719**         0.627         0.333
product/process
                             (0.672)       (0.986)      (0.570)     (0.449)       (0.427)        (0.333)       (0.333)        (0.575)        (1.222)
Constant                      6.740        14.036        9.603       8.686       30.303*        -14.622       -14.405       -45.351***    -140.45***
                             (9.859)      (49.586)     (22.400)    (19.957)      (17.963)       (13.576)      (13.576)       (12.853)       (45.677)
Observations                  444           444          444         444           444               444          444             444           444
Huber-White robust standard errors clustered on country-industry pairs. *** (1%), ** (5%), * (10%). Without much loss of generality, only the combined
contribution of R&D and New product/process development is shown.




                                                                           95
                                                                         Appendix B: Decomposition methodology

B1: Twofold mean KOB decomposition

For the twofold mean KOB decomposition, consider a variable Y of interest such as the percentage

decline in sales. Assume that it depends linearly on a set of covariates, X. Let G stand for the

relevant group, which equals L for large firms and SME for the remaining firms. Let i stand for the

ith firm. We can express the value of Y for the ith firm in group G as follows:



                                                                                                           ������������
                                                                       ������������������������������������ = ������������������������0 + ∑������������=1 ������������������������������������ ������������������������������������������������ + ������������������������������������                                          (1)



where ������������������������������������ is the error term with an expected value, conditional on X, of 0. Next, define the

difference in the mean or expected value of Y in the two groups as follows:



                                                                           ������������ = ������������ [������������������������������������������������ ] − ������������ [������������������������ ]                                 …….                              (2)



Equations (1) and (2) can be combined to express the difference in the expected value of Y

between the two groups as follows:



               ������������
                                                                    ∗
 ������������ = �������������� (������������������������������������������������������������ ) − ������������ (������������������������������������ )�������������������������
        �����������������
        ������������=1
                       ������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������
                                                                                                                                                                                                       (3)
                                                                                                 ������������                                                                                           ������������
                                                                                                       ∗                                           ∗
                                     + ������������������������������������������������0 − ������������������������0 + ��������������������������������������������������������������� − ������������������������ �������������������������������������������������������������������������� �� + ��������������������������� − ������������������������������������ �������������������������������������������������� ��
                                       ���������������������������������������������
                                                                       ������������=1                                                             ������������=1
                                                                                                                           ������������������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������




                                                                                                                            96
The ������������������������������������������������ , ������������������������ coefficients in equation (3) are the OLS coefficient estimates from SME and large

firm subsamples, respectively. ������������ ∗ denotes coefficients from a “non-discriminatory” model, i.e., a

benchmark against which both groups are evaluated. The goal of the decomposition analysis is to

estimate the two components shown in equation (3) for each determinant (Xj) of Y. As indicated in

equation (3), the first component gives the aggregate “endowment” effect or the part of the

difference in Y between the two groups that is due to differences in the level of covariates X. The

second component in equation (3) gives the aggregate “structural” effect or the part of the

productivity gap that is due to differences between the two groups in the returns to the covariates

X. The “endowment” and “structural” effects for the individual determinants of Y are as shown in

equation (3) inside the summation sign.

           A common practice in the literature is to use for ������������ ∗ the coefficients estimated from either

of the two groups or from the combined or pooled sample of both groups. We follow this practice.

In our baseline model, we use large firm sample as the reference groups. That is, we replace ������������ ∗ with

������������������������ . Thus, the baseline mean KOB decomposition is performed using the following equation:



                 ������������                                                                                                              ������������

   ������������ = �������������� (������������������������������������������������������������ ) − ������������ (������������������������������������ )������������������������������������������������������������� + ������������������������������������������������0 − ������������������������0 + ��������������������������������������������������������������� − ������������������������������������ �������������������������������������������������� ��
                                                                                                                                                                                         (4)
          �������������������
          ������������=1                                                                       ���������������������������     ������������=1
                             ������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������                           ������������������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������




For robustness, we perform two other decompositions using the SME sample and pooled sample

as the reference groups. The decomposition with the SME and the pooled samples as the reference

groups are given by equation (5) and equation (6), respectively:




                                                                                               97
                      ������������                                                                                                                                  ������������

     ������������ = �������������� (������������������������������������������������������������ ) − ������������ (������������������������������������ )������������������������������������� + ������������������������������������������������0 − ������������������������0 + ��������������������������������������������������������������� − ������������������������������������ �������������������������������������������������������������������������� ��
                                                                                                                                                                                                                       (5)
            �������������������
            ������������=1                                                               �����������������������������   ������������=1
                                     ������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������                                                 ������������������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������




               ������������
                                                                    ������������
 ������������ = �������������� (������������������������������������������������������������ ) − ������������ (������������������������������������ )�������������������������
        �����������������
        ������������=1
                             ������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������
                                                                                                                                                                                                                       (6)
                                                                                                     ������������                                                                                        ������������
                                                                                                             ������������                                          ������������
                                           + ������������������������������������������������0 − ������������������������0 + ��������������������������������������������������������������� − ������������������������   �������������������������������������������������������������������������� �� + ���������������������������   − ������������������������������������ �������������������������������������������������� ��
                                             ���������������������������������������������
                                                                             ������������=1                                                               ������������=1
                                                                                                            ������������������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������




            ������������
where ������������������������   is the vector of coefficients estimates obtained from the pooled sample of SMEs and

large firms.



B2: Threefold mean KOB decomposition

One issue with equations (4), (5), and (6) is that the sample used to compute the weights is different

for the “endowment” and “structural” effects. An alternative is to use the same sample for

computing both these weights. This is obtained by expressing equation (2) as follows:




                                                                                                            98
                  ������������                                                                                                                                                            ������������

     ������������ = ��������������������������������������������������������������������������� � − ������������������������������������������������� �������������������������������������� + ������������������������������������������������0 − ������������������������0 + ��������������������������������������������������������������� − ������������������������������������ �������������������������������������������������� ��
            �������������������
            ������������=1                                                             ���������������������������     ������������=1
                            ������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������                                                                           ������������������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������

                                                       ������������                                                                                                                                                                             (7)
                                        + ��������������������������������������������������������������������������� � − ������������������������������������������������� ���������������������������������������������������������������� − ������������������������������������ ��
                                          ���������������������������
                                          ������������=1
                                                                                            ������������������������������������������������������������������������������������������������������������������������������������ ������������������������������������������������������������������������




Equation (7) is known as the KOB threefold decomposition with respect to large firms, which is

the reference group. That is, large firms’ mean outcome (level of the dependent variable) is viewed

as the baseline, and equation (7) tells us what it would take for the large firms’ mean outcome to

converge to that of the SMEs.

         The first term in equation (7) shows the part of the gap related to group differences in the

explanatory variables or the “endowment” effect. It is weighted by the vector of coefficients of

large firms. The second term shows the portion of the gap stemming from the difference in the

group coefficients or the “structural” effects. This component is weighted by again by large firms’

vector of mean explanatory variables. The final term denotes the portion of the total gap that exists

due to the interaction of differences in endowments and coefficients between the two groups. In

other words, the interaction term indicates the (incremental) portion of the gap that occurs when

both the endowments and coefficients change simultaneously; or, alternatively, the portion of the

gap that remains after controlling for the “endowment” and “structural” effects.



B3: Quantile decomposition methodology

The traditional KOB approach discussed above works only for the mean value of the variable of

interest. It cannot be extended to other distributional statistics such as the median and other


                                                                                                                                             99
quantiles. To perform decomposition analysis by quantiles, a new technique developed by Firpo

et al. (2007, 2009) is employed. This technique involves first estimating the Influence Function or

more specifically the Recentered Influence Function (RIFs) for each quantile and for each

distribution considered (������������������������ , ������������������������ ). RIFs are statistical tools that can be used to estimate the impact

that changes in the distribution of explanatory variables (X) have on the unconditional distribution

and therefore quantiles of the variable of interest Y. Following Firpo et al. (2007) and Fortin et al.

(2011), the RIF we use is as follows:



                                                                    ������������−������������ {������������≤������������������������ }
                 ������������������������������������ (������������������������ , ������������������������ ) = ������������������������ +        ������������������������ (������������������������)
                                                                                                       G=SME, L   (8)




where I{} is an indicator function and ������������������������ is the marginal density of ������������������������ . ������������������������ is the population τ-

quantile of the unconditional distribution of ������������������������ . The RIF is obtained by estimating the population

quantiles and the density function using the available data on ������������������������ . The estimated RIF values can be

used in the standard OLS and other regressions.

        With the RIFs estimated, we can proceed to the decomposition analysis. The

decomposition analysis can be either like the traditional KOB decomposition discussed above or

a slightly modified “hybrid” version such as the one suggested by Firpo et al. (2007). In this paper,

we use the traditional KOB method which involves repeating the exercise in equations (4) above

with the difference that Y values are replaced by the RIF values. This gives the “endowment” and

“structural” effects for difference in the level of the relevant quantile value of Y between the two

groups. The entire exercise is repeated for each quantile of interest.




                                                                                                 100