Corruption and entrepreneurship: does gender matter?

Full text

168
8. Corruption and entrepreneurship:
does gender matter?
Claudia Trentini and Malinka Koparanova
1 INTRODUCTION
Corruption is a significant factor that determines the quality of the
“doing business” environment at large. Corrupt public practices decrease
the economic efficiency of governments, reduce incentives for public
and private investment, and weaken the institutional foundation of
development (Olken and Pande 2012). As a consequence, it is a common
practice for international donors and development agencies to devote some
of their economic assistance toward improving transparency and govern-
ance as they are viewed as essential ingredients for achieving sustainable
development. Similarly, the international development community widely
acknowledges the strong relationship between gender equality and women’s
empowerment on the one hand and economic development on the other
(Duflo 2012). In particular, gender equality is seen as a “prerequisite” for
economic growth and social justice.
This chapter explores the nexus between both issues, gender and
corruption; it investigates the role of gender in corrupt behavior and its
consequences on firm growth drawing on World Bank and EBRD firm-
level data for 31 countries in Europe and Central Asia. The work is moti-
vated by two growing strands of research documenting, on the one hand,
systematic gender differences in behavior ranging from risk aversion to
moral standards and dishonest behavior including corruption and, on the
other hand, persistent performance differences between male- and female-
owned firms.
We focus on so-called countries with economies in transition. In these
countries the process of establishing new governance institutions has been
accompanied by high levels of corruption that have adversely affected the
development of an inclusive private sector (EBRD 2005; Tonoyan et al.
2010). In such an environment, the evolution of small and medium enter-
prises (SMEs) is sluggish, high levels of unemployment persist, and poverty
and income inequalities rise. Although women account for the majority of
LINK TEXT.indd 168 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 169
the unemployed and the poor in many of those countries, female-owned
companies are rare (Aidis et al. 2007) (Appendix A Table 8A.1).
The first question we address is related to the extent to which, after
accounting for systemic differences in firms’ characteristics, gender gaps
in corruption incidence still persist. This naturally leads us to the second
question we posit as to the impact of informal payments on firm employ-
ment growth and the extent to which this could represent a gender-specific
factor hampering entrepreneurship.
Our contribution is twofold. First, the role of gender in the context of
corruption studies is virtually unexplored. Recent research has pointed out
that since women are less prone to corruption, increasing their participa-
tion in politics and in the labor force could reduce the overall level of gov-
ernment corruption and improve the business environment by increasing
business trust (Dollar et al. 2001; Swamy et al. 2001; Torgler and Valev
2006; Goetz 2007; Michailova and Melnykovska 2009). This empirical
evidence has been backed by a stream of research papers documenting sys-
tematic gender differences in economic behavior ranging from risk-taking
attitudes (Croson and Gneezy 2009) to the provision of public goods and
altruism (Andersen et al. 2008) to corruption (Rivas 2013).1 However,
experimental findings rely on artificial situations not always holding in the
real world (Frank et al. 2011). It is thus important to deepen the analysis
using micro-field data where these are available. In particular, our chapter
relies on a cross-country dataset with individual firm information allowing
us to adopt an international perspective while still taking into account firm
characteristics. We add to the literature analyzing in depth what are the
observable factors explaining lower female bribing incidence.
Second, we contribute to the debate around gender differences in entre-
preneurship. Research in this area has pointed out that firms owned by
women are smaller, concentrate in different sectors, and perform less well
than firms owned by men.2 The economic and institutional explanations for
these patterns have focused on discrimination in credit markets (Murayev
et al. 2009; Aidis et al. 2007), on the access to capital and business services,
and on women’s concentration in sectors with less growth opportunities
(Verheul and Thurik 2001; Sabarwal and Terrell 2008; World Bank 2012).
A few studies have analyzed the differential impact of institutional factors
on women entrepreneurs. For example, Estrin and Mickiewicz (2011) find
that women are less likely to undertake entrepreneurial activity in countries
where the state sector is larger, and van der Zwan et al. (2011) find that the
perceived environmental constraints that hinder entrepreneurial progress
are more prevalent in the (former) European transition economies than in
other economies. However, no paper to our knowledge has yet analyzed
the differential impact of corruption on women entrepreneurs.
LINK TEXT.indd 169 24/02/2017 15:55

170 Gender and Entrepreneurial Activity
The findings have important implications for management practice
and development programs. If gender differences in bribing behavior and
related growth performance are primarily associated with firms’ opera-
tional attributes, development programs should focus on the develop-
ment of the firm and the improvement of the business environment. For
example, a friendlier business environment to small entrepreneurs will be
more attractive to women (because they are more likely to run smaller
firms), encouraging them to become entrepreneurs and increasing in this
way their incomes and welfare.3 If, on the contrary, women entrepreneurs
are less likely to engage in corruption than men entrepreneurs, programs
to fight corruption might benefit by adopting a gender-sensitive perspec-
tive. In this case, measures and policies to fight corruption will reduce the
advantages men entrepreneurs gain from these behaviors, thereby resulting
in a reduction in the entrepreneurship gender gap.
Our results indicate that women entrepreneurs, especially those oper-
ating smaller firms, do have a significantly lower propensity to bribe as
compared to men entrepreneurs. Consistent with the vast majority of
literature on corruption and firm development, we also find that for small-
and micro-enterprises corruption has a negative impact on firm growth.
However, our findings show that for women, engaging more in bribing
activities would in fact increase their growth prospects. This is consistent
with the fact that the majority of women are micro entrepreneurs: on the
one hand, it is easier for them to escape the attention of corrupt officials,
on the other hand, it is harder to grow without the right network and expe-
rience, and hence greasing the wheels of state bureaucracy might become
necessary and facilitate their firm’s growth.4 This evidence thus provides
additional explanations for the difference in behavior across genders and
for the performance gap of women entrepreneurs in transition economies.
This can be used by policy makers to enhance their business environment
and alleviate gender inequality in their countries. In addition, it offers a
more nuanced view about corruption blurring the strict tradeoff between
the “greasing the wheels” and “sanding the wheels” of the economy
hypotheses (Méon and Sekkat 2005).
The chapter is organized as follows. The next section reviews the
literature on corruption in the context of a gender perspective. In Section
3 we describe the data and the key measures of corruption we use in the
study. In Section 4 we present the econometric models of determinants of
corruption and discuss the empirical results. In Section 5 we outline the
analytical framework of the impact of informal payments on firm growth
and provide empirical evidence on gender differentials. In Section 6 we
summarize our empirical findings and conclude.
LINK TEXT.indd 170 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 171
2 LITERATURE REVIEW AND MAIN HYPOTHESIS
The Determinants of Corruption
Classical theoretical work models corruption as a problem of moral hazard
(see for example Rose-Ackerman 1975; Shleifer and Vishny 1993; Svensson
2003). In these kinds of models a bureaucrat decides whether to be corrupt
(obtaining a bribe) or honest on the basis of the gains he or she can obtain
from the two alternative behaviors and the probability of being detected
if dishonest and punished. The higher the risk of being detected and the
punishment costs associated, the lower the probability he or she will accept
or request a bribe.5
Within this simple framework Svensson’s (2003) seminal study, develops
a series of hypotheses identifying the characteristics of firms most likely
to be the victim of corruption. In his model, public officials can extract
bribes from firms as long as these cannot easily escape from their control
and the probability of being caught and punished is low. It follows that
firms which require permits and certificates to operate, or trade abroad
and are therefore reliant on infrastructure services, are more likely to pay
bribes as they need to interact more with public officials. At the same time
the amount paid should increase if a firm’s profitability, that is, its ability
to pay, goes up, and should decrease if the firm could easily change sector
or location. These hypotheses are verified and confirmed by data from a
firms survey in Uganda.
The rising availability of firms surveys around the world allowed for a
new generation of empirical studies on corruption and business activity.
Few studies, however, focused on the determinants by firms’ characteristics
and virtually none considered gender as a possible factor. In fact empirical
studies on the relation between gender and corruption tend to focus more
on political corruption and often use macro-level estimates.6 Dollar et al.
(2001) is one of the first papers that investigate empirically the relation-
ship between women’s government participation in legislatures and the
level of perceived corruption, using as a measurement of corruption the
International Country Risk Guide (ICRG) and proxying women’s involve-
ment in government by the proportion of seats in parliament that were
held by women in the upper and lower houses. The other influential paper
investigating the relationship between gender and corruption is Swamy et
al. (2001). They use several data: in addition to macro evidence based on the
Transparency International Corruption Perception Index and the World
Value Survey, they also present micro evidence on lower female corruption
propensity using data from a World Bank study of corruption in Georgia
(survey of 350 firms).7 This part of their paper is the most related to ours.
LINK TEXT.indd 171 24/02/2017 15:55

172 Gender and Entrepreneurial Activity
In contrast, experimental findings about differences by gender in cor-
rupted behavior have been growing rapidly in the last years indicating sig-
nificant gender differences (Frank et al. 2011; Rivas 2013). These findings
suggest the main hypothesis we will test:
Hypothesis 1: women entrepreneurs are less prone to corruption after
controlling for an extensive set of firm characteristics.
We then ask what explains why women are showing a lower incidence
of bribing.
A number of papers have found differences in the incidence of corruption
by firm size and sector (Aterido and Hallward-Driemeier 2009; Beck et al.
2005; Clarke 2011; Escribano and Gualsch 2005; Seker and Yang 2012).
Some of these factors are also found to be systematically different in
female-owned firms. In fact, typically female-owned firms are smaller and
concentrate in the services sectors (Klapper and Parker 2011; Sabarwal
and Terrell 2008; Murayev et al. 2009).
This could be driving their lower bribing incidence as they more easily
escape the attention of public officers. We thus verify the following
hypotheses:
Hypotheses 2: female entrepreneurs running companies of the same size as
men and operating in the same industries are equally likely to bribe public
officials.8
As a last factor we consider risk attitudes. Engaging in bribing is a risky
business; from the point of view of an entrepreneur, paying a bribe can
be considered an investment with a high level of risk attached, especially
for small enterprises with limited resources: adding to the probability of
being detected and punished together with the public official, the expected
payoff – such as easier bureaucratic procedures – is very uncertain and
relies on the cooperation of a corrupt public officer (Freund et al. 2014;
Berninghaus et al. 2013). While entrepreneurship is historically associated
with risk bearing, risk attitude is widely considered a determining factor
on entrepreneurial activity and success. Consequently, lower female risk
attitudes (see for example Borghans et al. 2009 and Croson and Gneezy
2009) have been proposed as explanations for gender differences in entre-
preneurship (Fossen 2012). We develop this line of reasoning by asking if
women’s risk attitudes can play a role in bribing behavior.
Hypothesis 3: female entrepreneurs having the same risk attitude of men
are equally likely to bribe public officials.
LINK TEXT.indd 172 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 173
Corruption and Firm Growth
Conventional wisdom considers corruption detrimental to economic per-
formance; it distorts incentives, increases transaction costs, and aggravates
uncertainty, leading to misallocation of resources and underinvestment
(Murphy et al. 1991; Mauro 1995; Gray and Kaufmann 1998; Svensson
2005; Fisman and Svensson 2007). Of particular concern are the dis-
couraging effects of informal payments and bribe extortion on entrepre-
neurial activity. Yet the academic debate is still drawn to the question as to
whether corruption “greases” or “sands” the wheels of economic growth
and development. Supporters of the grease theory argue that corruption
can enhance efficiency by enabling entrepreneurs to circumvent burden-
some business regulation, or by incentivizing bureaucrats to work harder
through bribes (Leff 1964; Huntington 1968; Méon and Sekkat 2005;
Méon and Weill 2010). However, the majority of the cross-country lit-
erature finds no evidence in favor of this hypothesis (Campos et al. 2010).
Few researchers found evidence backing the greasing theory in highly
regulated markets (Méon and Weill 2010; Dreher and Gassebner 2013) or
in the presence of deeper distortions (e.g., Wang and You 2012 focus on
financial market failures in China, Vial and Hanoteau 2010 on a rigid and
overcentralized administration in Indonesia).
In contrast, country-level studies show convincing evidence of a nega-
tive relationship between levels of corruption and firm sales growth
(Fisman and Svennson 2007 for Uganda, Francisco and Pontara 2007
for Mauritania, Hallward-Driemeier et al. 2006 for China, Honorati and
Mengistae 2007 for India).
Evidence from studies with wider country coverage is more mixed but
points in the same direction: Seker and Yang (2012) and Gaviria (2002)
find that for Latin American firms bribing significantly reduces sales
growth; Batra et al. (2003) pool firm-level data from 81 developing and
developed countries and find that corruption has a negative and significant
impact on sales and investment growth; Aterido et al. (2011) cover over 90
countries and do not find a significant effect of corruption on employment
growth; Asiedu and Freeman (2009) find that the effect of corruption on
investment growth varies significantly across regions, having a negative
and significant effect for firms in transition countries but having no sig-
nificant impact for firms in Latin America and Sub-Saharan Africa. Beck
et al. (2005) do find that corruption affects firm growth adversely, and
especially so for small- and medium-sized firms.
While many of these studies consider the hypothesis that the impact
of corruption on a firm’s growth can vary according to the firm’s char-
acteristics (size, sector, foreign versus state ownership), none considers
LINK TEXT.indd 173 24/02/2017 15:55

174 Gender and Entrepreneurial Activity
owner’s gender as a possible factor. In fact, female-owned enterprises’
lower bribing attitude could lead to a better performance due to the
avoided costs (according to the greasing the wheels theory) or, to the con-
trary (sanding wheels theory), to lower growth rates as companies cannot
deal with bureaucratic burdens that hold them back. In particular, as the
majority of women are running micro enterprises in the services sector,
where they have fewer opportunities to meet with public officials, it is likely
that, on the one hand, they find it easier to escape officials’ bribe requests
(Aterido et al. 2011) while, on the other hand, they run against additional
bureaucratic regulations whenever they decide to grow. In the latter case,
greasing the wheels could be the right strategy to pursue their growth strat-
egy in a very corrupted environment that penalizes women the most, such
as the one prevalent in transition economies (EBRD 2005; Seker and Yang
2012; Tonoyan et al. 2010; Michailova and Melnykovska 2009; Murayev et
al. 2009). These considerations lead us to our last hypothesis:
Hypothesis 4: for female entrepreneurs the impact of corruption on
employment growth after controlling for all other firm characteristics is
positive.
3 DATA AND VARIABLES DEFINITION
We use the 2005 Business Environment and Enterprise Performance Survey
(BEEPS) data produced by the World Bank and the European Bank for
Reconstruction and Development (EBRD) for 31 European and post-
socialist economies. The sample was constructed by stratified random
sampling from a national registry of firms. In each country, the industry
composition of the sample in terms of manufacturing versus services
was determined by their relative contribution to gross domestic product
(GDP). Firms that operate in sectors subject to government price regu-
lation and prudential supervision, such as banking, electric power, rail
transport, and water and waste water, were excluded from the sample. The
sample includes only registered firms (i.e., not informal firms), and their
size varies from as few as 2 employees to as many as 9,999 employees.
The BEEP surveys are consistent across countries and include a number
of questions on the firms’ characteristics. Most importantly for the
purpose of this chapter, they allow for identification of the gender of the
main owner/manager (as explained below) and provide information on
corruption incidence. Weaknesses of the database include the inability to
identify the gender of the other owners when there is more than one and
the lack of demographic information on the entrepreneurs. In addition to
LINK TEXT.indd 174 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 175
this, we cannot extend our dataset to other BEEPS waves, as earlier waves
of the survey do not provide information on the gender of the principal
owner/manager of the firm, while successive ones change the relevant
questions on the property structure of the firm used to identify female
owners and drop some of the questions related to corruption.9
Because the gender question in the BEEPS does not refer to the
manager, but to the principal owner or one of the principal owners of the
firms, we define as entrepreneur the principal owner of a firm (at least 50
percent stake in the enterprise) who is also the manager and for which we
know the gender. We thus discard data from state-owned corporations,
stock-exchange listed corporations, cooperatives, and family-owned firms
where it is not possible to identify the main owner/manager, and are left
with a sample of mostly small enterprises. There are about 9,900 such firms
in the database, with 150–600 firms per country. Excluding firms that did
not provide information about their capital stock and their sales, we end up
with a sample of 5,471 enterprises.
The countries included are as follows: five old European Union (EU)
members (Germany, Ireland, Greece, Portugal, and Spain), ten New
Member States (NMS) (Bulgaria, Czech Republic, Estonia, Hungary,
Latvia, Lithuania, Poland, Romania, Slovakia, and Slovenia),10 seven
South-Eastern Europe (SEE) states (Albania, Bosnia, Croatia, The former
Yugoslav Republic of Macedonia, Turkey, Serbia, and Montenegro), and ten
Commonwealth of Independent States (CIS) countries (Armenia, Georgia,
Kyrgyzstan, Moldova, Tajikistan, Uzbekistan, Belarus, Kazakhstan,
Russia, and Ukraine). Appendix A Table 8A.1 shows the composition of
the sample by region and by gender and the incidence of bribing behavior.
More than 40 percent of the firms in the sample come from the old EU
members, more than 32 percent are from the countries which joined the EU
in or after 2004, only about 10 percent are from SEE, and the remaining 17
percent are from the CIS. Female ownership and bribing behavior varies
drastically across groups of countries: while female ownership is about 30
percent in old and new EU countries, it accounts for only 23 percent in the
CIS and 17 percent in SEE states.
Appendix A Table 8A.2 gives the definitions of variables and basic
descriptive statistics for the entire sample.
Corruption
We adopt Svensson’s (2005) definition of corruption: “the misuse of public
office for private gain.” This definition is quite general and includes the
different forms addressed by the survey’s questionnaire and analyzed in
this chapter (described later in this section). The questions on corruption
LINK TEXT.indd 175 24/02/2017 15:55

176 Gender and Entrepreneurial Activity
were phrased indirectly to avoid implying that the respondent is engaged in
any wrongdoing. Moreover, information about the confidentiality of the
survey was explicitly communicated and repeated before all the questions
on corruption. The main question on bribes payment was “On average,
what percent of total annual sales do firms like yours typically pay in unof-
ficial payments/gifts to public officials?” Note that the question refers to
practices in their entrepreneurial industry, not to their own behavior; this
is likely to provide a more accurate answer since the entrepreneur need not
refer to their own behavior,11 but it does introduce possible distortions on
the reliability of data, likely leading to an underestimation of corruption
practices (Kraay and Murrell 2013). A number of studies make use of sub-
jective firm-level data measures of corruption arguing this an acceptable
alternative to objective measures (Gelb et al. 2007; Kaufmann et al. 2005;
Tanzi 1998). Hunt (2004) finds a high correlation (up to 0.7) between sub-
jective measures of corruption (provided by Transparency International
or World Bank surveys) and rare objective measures of corruption
(International Crime Victim Surveys). Similarly, Aterido and Hallward-
Driemeier (2009) and Tonoyan et al. (2010) underline that subjective meas-
ures are highly correlated with objective measures and also significantly
correlated with external sources, including World Bank “Doing Business”
or “Control of Corruption” indicators.
Clearly, even if objective and subjective measures are correlated, the
latter remain prone to bias. For example, Olken (2009) finds systematic
individual-level biases in subjective measures. On the other hand, Aterido
and Hallward-Driemeier (2009) show that in firm surveys, much of the
variation in subjective responses by firm types is largely due to differences
in the objective conditions across firm types. Yet concerns about response
variations by individual characteristics of the firm’s owner such as gender
remain. In particular, because the questionnaire does not discriminate
between gender, asking “what women/men in your position typically pay”
(rather than “firms like yours”), women’s lower reported corruption is
likely to be related to their expertise or knowledge of typical corruption
rates.12 To tackle this problem our benchmark model controls for two vari-
ables most related to owner’s experience and knowledge such as age of the
company and participation in formal networks (in business associations
and chambers of commerce). Further, we perform other robustness checks
by splitting the sample along these variables without finding significant
differences across subsamples’ estimates.13
We identify two measures of corruption. The first measure of corrup-
tion is related to administrative corruption. Administrative corruption
refers to the illicit and non-transparent provision of payments or other
benefits to public officials in exchange for illicit and non-transparent
LINK TEXT.indd 176 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 177
preferential treatment in the “implementation” of prescribed administra-
tive regulations, rules, and policies imposed by the state on the activities
of firms and individuals (Hellman et al. 2003). This kind of informal
payments has an incidence across the firms of about 22 percent in the old
EU countries, 33 percent in the NMS, 42 percent in the SEE states and 47
percent in the CIS (column 3 of Appendix A Table 8A.1). Column 4 shows
the amount of informal payments paid for administrative corruption in
terms of the percentage of sales. These informal payments vary consider-
ably across the sub-regions we study, with the old CIS countries leading
with the value of bribes (in percentage of the firm’s sales), reaching more
than 3 percent of sales, about twice the percentage paid (1.6 percent) in the
old EU countries.
The second measure of corruption is related to the motivation to obtain
a government contract. In this chapter we call it a gift to obtain a contract
and it refers to the payments entrepreneurs make with the objective of
obtaining an advantage in winning a government contract.14 This kind of
payments are more frequent for larger enterprises and are expected to be
positively correlated to firm growth, as entrepreneurs engaging in this form
of bribing are able to bend public rules/procedures to their own private
benefit. Here the percentages of incidence are rather similar across country
groups, involving about 30 percent of the entrepreneurs; the exception is
represented by the SEE states where the incidence is around 38 percent
(column 5 Appendix A Table 8A.1), while the percentage of contract paid
varies from a minimum of 3 percent in the old EU countries to a maximum
of 9 percent in the SEE states.
The other variables in Appendix A Table 8A.2 describe a sample of
mainly small enterprises with almost 60 percent of them having no more
than 10 permanent employees in 2002. Of these, only a fifth is exporting,
and up to 40 percent suffered interruption in some basic infrastructure
services such as power outages. Public services interruptions, trade pro-
pensity, attitude toward the local bureaucracy, and number of inspections
during the past year are directly related to the probability of an entrepre-
neur of engaging in administrative corruption: companies which need to
interact with public officials the most (for customs, permits, and public
services) are those most likely to be extorted informal payments. These
variables are summarized in the variable FormalSect. This variable is con-
structed following Svensson (2003) and is the principal component of bad
infrastructures, trade, and number of inspections by public officials and
gives a measure of the extent to which entrepreneurs have to deal with
public officials. Two thirds of the respondents were confident in the state’s
ability to protect their property rights (Property) but only less than half
thought they could act against dishonest bureaucrats (Fair_bureauc). The
LINK TEXT.indd 177 24/02/2017 15:55

178 Gender and Entrepreneurial Activity
average company age in 2005 was almost 16 years, and the majority was
based in small cities or rural areas.
Firm Growth
Regarding firm growth, the dataset contains retrospective information
for employment in 2002 (three years before the year of reference of the
survey, 2005), allowing us to construct our outcome variable of interest:
total employment growth. This measure includes permanent employment,
which is likely to reflect the long-run performance of the firm, as well as
temporary workers, which is indicative of short-run dynamics potentially
very important for micro enterprises. This chapter focuses on SMEs, where
permanent employment might not be a good proxy for growth: given the
fragility and uncertainty that very young or micro enterprises face, most
of them might not use permanent contracts until they are well established.
For the same reason, in many countries, labor regulations do not apply
to them. We thus consider using total employment growth to be a better
measure to capture growth variations for small companies. Moreover, as
argued by Aterido et al. (2007, 2011), using this measure may help avoid
inconsistencies of reporting across countries; in addition, the evolution of
employment growth is of high concern for policy makers.15 Our measure
of employment growth refers to the change of employment during the
period t and three years before, divided by the firm’s simple average of total
employment during the same period, divided by three. On average compa-
nies grew at a yearly rate of 3 percent between 2002 and 2005. Employment
growth is preferred to other measures of growth, such as sales or assets,
because these are reported in terms of percentage change from the previ-
ous three years and rather than giving the exact corresponding past value
(Appendix A Table 8A.2).16
Since the sample size is small for many countries, we define three sector
groups: manufacturing, trade, and transport and services (including hotels
and restaurants and other service sectors like real estate, renting and busi-
ness services, and construction services; Appendix A Table 8A.2 shows the
two digit ISIC codes).
Gender Differentials
Table 8.1 presents the main variables by gender. Several differences become
apparent not only in bribing behavior but also in firms’ characteristics. On
average, incidence of corruption seems to be smaller for women entrepreneurs
while there is no difference across genders in the amount paid. Confirming
literature on gender gaps in firm performance, female-owned companies
LINK TEXT.indd 178 24/02/2017 15:55

179
Table 8.1 Descriptive statistics by gender
Variable MEN WOMEN Difference
 − W
sMNWsWNW
Adm. Corr. (incidence) 0.34 0.47 3,590 0.27 0.44 1,376 0.07***
Adm. Corr. (as % of sales) 2.50 3.15 1,209 2.59 3.25 367 −0.09
Gift for contr. (incidence) 0.33 0.47 3,001 0.21 0.41 1,174 0.12***
Gift for contr. (% of contract) 5.37 4.49 990 5.31 5.08 250 0.06
Employ. growth 0.03 0.14 3,934 0.02 0.13 1,537 0.01***
micro entrepr 0.53 0.50 3,934 0.66 0.47 1,537 −0.13***
small entr 0.28 0.45 3,934 0.22 0.42 1,537 0.06***
medium entr 0.13 0.33 3,934 0.08 0.27 1,537 0.05***
large entr 0.06 0.23 3,934 0.04 0.19 1,537 0.02***
FormalSect 0.01 1.05 3,934 −0.02 1.17 1,537 0.02
Property 0.65 0.48 3,934 0.63 0.48 1,537 0.02
Fair-bureauc 0.48 0.50 3,934 0.48 0.50 1,537 0.00
Profits2003 0.90 0.30 3,934 0.89 0.31 1,537 0.01
InK 5.20 1.82 3,934 4.71 1.88 1,537 0.49***
concentration 95.73 10.23 3,934 96.40 9.75 1,537 −0.66**
memberBA 0.56 0.50 3,934 0.50 0.50 1,537 0.06***
privatization 0.08 0.27 3,934 0.07 0.25 1,537 0.01
foreign_own 0.06 0.24 3,934 0.03 0.17 1,537 0.03***
ageCompany 16.06 15.47 3,934 15.30 15.33 1,537 0.76
ageComp2 4.97 16.69 3,934 4.69 17.02 1,537 0.28
small city/rural 0.57 0.50 3,934 0.59 0.49 1,537 −0.02
LINK TEXT.indd 179 24/02/2017 15:55

180
Table 8.1 (continued)
Variable MEN WOMEN Difference
 − w
sMNwswNw
urban 0.43 0.50 3,934 0.41 0.49 1,537 0.02
manufacturing 0.36 0.48 3,934 0.30 0.46 1,537 0.06***
trade and transport 0.30 0.46 3,934 0.32 0.47 1,537 −0.02
services 0.34 0.47 3,934 0.38 0.48 1,537 −0.04***
% Workforce 90.35 17.34 3,833 92.00 15.40 1,499 −1.65***
timeReg 4.27 7.30 3,717 4.07 7.95 1,443 0.20
growth assets 0.05 0.14 3,934 0.04 0.15 1,537 0.01**
growth exports 0.01 0.10 3,934 0.01 0.07 1,537 0.01**
Notes: m stands for the mean, s for the standard deviation, and N for the number of observations. mM − mW is mean differences and t-test
significance. *significant at 10%, **significant at 5%, ***significant at 1%.
LINK TEXT.indd 180 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 181
tend to be smaller than their male-owned counterparts (both in terms of
employees and in terms of capital), they grow less (on all measures presented:
employment, assets, or exports), are less present in the manufacturing while
are concentrated in services sectors, and are less likely to participate in busi-
ness associations/chambers of commerce. These are all confounding factors
that might partly account for gender differences in bribing behavior.
4 ECONOMETRIC MODELS OF DETERMINANTS
OF CORRUPTION AND RESULTS
Specification
In this section we investigate how gender is related to entrepreneurs’
corrupt behavior. For this purpose, we adopt Svensson’s (2003) framework.
We use the following basic econometric model to investigate the link
between the gender of owners/managers and their corrupt behavior:
pi 5 a0 + a1Femalei + a2Xi + ni (1)
gi 5 g0 + g1Femalei + g2Xi + ei (2)
Equations (1) and (2) have the same structure but different dependent
variables. Equation (1) analyzes the incidence of corruption: the depend-
ent variable pi is a binary variable equal to 1 if the firm is engaged in paying
bribes and 0 otherwise. Equation (2) studies the amount paid: gi is a vari-
able measuring the amount paid as graft in terms of percentage of total
sales of the public contract. Female is a dummy variable which equals 1 if
the principal owner of the firm is a female and 0 otherwise, Xi is a vector
of variables introducing firms’ characteristics and the bargaining power of
each entrepreneur i, and ni and ei are the respective error terms.
The vector of variables Xi includes characteristics of the firm derived
from the bargaining theory applied to the corruption case (Svensson 2003).
It hypothesizes that the extraction of bribes is a two-stage game: in the first
stage corrupt bureaucrats are matched to firms, while in the second stage
a bargaining process between entrepreneurs and public officials starts to
determine the amount extracted. Accordingly, firms’ probability to be the
victim of bribe extraction (first stage of the game) and the amount they
pay (second stage) depend on measurable firm attributes: i) the extent to
which a company needs to deal with public officials, ii) its ability to pay
(profitability/size), or rather to subtract itself from public officers’ control
by moving into another sector/location.17
LINK TEXT.indd 181 24/02/2017 15:55

182 Gender and Entrepreneurial Activity
For the first category of variables, we follow Svensson (2003), construct-
ing a measure of the extent to which entrepreneurs have to deal with public
officials (FormalSect) and thus of the control public officials maintain
over a firm.18 To the second category of variables belong the following: a
dummy variable equal to 1 if the firm had positive profits in 2003 (Profits
2003), the natural logarithm of the capital stock (ln K), the size of the
enterprise measured on the basis of permanent employment in 2002, a
measure of concentration (concentration) indicating how much of the sales
come from its principal product, proxying the ability of the firm to exit the
sector/market.
To these “standard” variables we add others suggested as important by
the literature for capturing firms’ attributes influencing the interactions
with public officials: the age of the company (age) as longer established
firms could have better access to official contacts (Svensson 2003; Hellman
et al. 2003; Aterido et al. 2011, 2007; Seker and Yang 2012),19 and an indi-
cator as to whether the entrepreneur is a member of a business association
to capture their connection to networks (member BA). Participation in
networks affects the availability of information on training opportunities,
business partners, the access to new markets, and all the formal and infor-
mal practices to run a company (Tonoyan et al. 2010; Verheul and Thurik
2001). Two further variables in vector X measure the influence ownership
has on contacts with the bureaucracy: a dummy variable for privatized
firms (privatization) as these might be expected to maintain ties and access
to public officials and one for foreign-owned ones (foreign own) as foreign-
ers might – to the contrary – have more difficulties interacting with local
bureaucracies (Hellman et al. 2003; Commander and Svejnar 2007; Seker
and Yang 2012). We then add two dummy variables capturing the entre-
preneurs’ perception of their regulatory and bureaucratic environment:
property accounts for a firm’s security of property and contract rights
while Fair_bureauc reflects entrepreneurs’ trust of bureaucrats. Firms
insecure of their property rights might be tempted to seek individualized
protection through informal payments, while those confident that they can
get fair treatment from public officials will escape more easily from their
control (Hellman et al. 2003).
The expected signs of all these variables are presented in Table 8.2;
predictions are different by type of corruption and by stage of the game
(eq. 1 incidence vs eq. 2 amount paid).
To account for differences in economic conditions and productive
structure, the regressions also include a full set of sector–location–country
dummies. Our results confirm that the country-specific component is very
strong, and corruption levels and practices change considerably across
countries. Moreover, the location dummies control for potential omitted
LINK TEXT.indd 182 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 183
variables at the country level. Thus, our focus is on within-country rather
than across-country variation.20
When estimating equations (1) and (2) the question arises as how to best
control for the potential endogeneity related to some of the explanatory
variables as well as the reverse causation. In fact, bribe payments might
affect the firm’s profitability and its ability to grow (Fisman and Svensson
2007), as we will analyze in the next section. We use lagged values of
profitability and employment to mitigate the problem of endogeneity and
reverse causation.
Equation (1) is estimated using a probit model, while equation (2) is
estimated by ordinary least squares (OLS), allowing for heteroskedasticity
and clustering errors by country.
Analysis of Results
The results of the estimated models are presented in Table 8.3.
Column 1 shows the results from estimating the likelihood of being
involved in administrative corruption using the probit model. The depend-
ent variable is a dummy variable AdmCorr, which is equal to 1 if a firm
paid a bribe in the past 12 months to “have things done” and 0 otherwise.
Table 8.2 Expected signs of variables
Adm. Corr.
(incidence)
(1)
Gift for contract
(incidence)
(2)
Adm. Corr.
(% sales)
(3)
Gift for contract
(% contract)
(4)
Femaleown − − −/+ −/+
size (employment) −/+ + − −
Profits2003 −/+ −/+ + +
InK −/+ + − −
concentration −/+ −/+ − −
FormalSect + + −/+ −/+
ageCompany + + − −
memberBA + + − −
privatization + + − −
foreign_own − − −/+ −/+
Property − − − −
Fair_bureauc − − − −
manufacturing −/+ −/+ −/+ −/+
trade and transport + + + +
services −/+ −/+ −/+ −/+
Note: When no clear prediction emerges from the available literature a minus/plus is
indicated.
LINK TEXT.indd 183 24/02/2017 15:55

184 Gender and Entrepreneurial Activity
Table 8.3 Determinants of corruption
Adm. Corr.
(Incidence)
(1)
Gift for contract
(Incidence)
(2)
Adm. Corr.
(% sales)
(3)
Gift for contract
(% contract)
(4)
Femaleown −0.037
[0.016]**
−0.072
[0.012]***
−0.003
[0.244]
−0.023
[0.427]
small entr 0.082
[0.021]***
0.017
[0.025]
−0.131
[0.169]
0.237
[0.349]
medium entr 0.038
[0.030]
−0.057
[0.027]**
−0.643
[0.273]**
−0.092
[0.493]
large entr 0.016
[0.046]
−0.078
[0.033]**
−0.681
[0.357]*
−0.747
[0.900]
Profits2003 −0.005
[0.018]
0.023
[0.022]
0.168
[0.211]
−0.38
[0.314]
InK 0.012
[0.010]
0.008
[0.006]
−0.15
[0.060]**
−0.18
[0.096]*
concentration −0.002
[0.001]***
−0.002
[0.000]***
−0.001
[0.007]
−0.004
[0.013]
FormalSect 0.043
[0.009]***
0.03
[0.009]***
0.049
[0.079]
0.216
[0.069]***
ageCompany −0.002
[0.001]**
0
[0.001]
−0.006
[0.011]
−0.016
[0.021]
ageComp2 0.001
[0.001]**
0
[0.001]
0.004
[0.008]
0.018
[0.016]
memberBA 0.01
[0.031]
0.021
[0.015]
−0.304
[0.174]*
−0.148
[0.267]
privatization −0.033
[0.025]
−0.022
[0.029]
0.066
[0.307]
−0.214
[0.650]
foreign_own −0.031
[0.031]
0.008
[0.025]
−0.233
[0.305]
0.237
[0.435]
Property −0.089
[0.017]***
−0.083
[0.015]***
−0.536
[0.136]***
−0.743
[0.201]***
Fair_bureauc −0.062
[0.016]***
−0.027
[0.018]
−0.278
[0.156]*
−0.414
[0.344]
small city/rural −0.03
[0.025]
−0.06
[0.023]**
−0.144
[0.162]
−0.364
[0.252]
trade and
transport
0.058
[0.021]***
0.044
[0.016]***
−0.407
[0.275]
−0.593
[0.327]*
services 0.14
[0.024]***
0.088
[0.022]***
0.018
[0.242]
0.408
[0.216]*
Observations 4,966 3,862 1,576 1,240
Notes: Columns 1 and 2 show marginal effects after probit estimation evaluated at the
mean; columns 3 and 4 are estimated by OLS. The dependent variables are dummy variables
denoting the incidence of administrative corruption (col. 1); gifts to obtain a contract
(col.2); administrative corruption in % of sales (col. 3); and gifts to obtain a contract in %
of contract value (col. 4). The reference category is microenterprises. The regressions include
country dummies. Asymptotic cluster-robust standard errors (clustering by country) are
reported in brackets. *significant at 10%, **significant at 5%, ***significant at 1%.
LINK TEXT.indd 184 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 185
Marginal effects estimated at the mean are reported for all variables. The
coefficient on the variable Female is negative and statistically significant.
According to the estimates, female-owned/managed businesses have about
3.7 percent lower probability of bribing with respect to male-owned firms.
The estimation results also suggest that small and medium firms are likely
to be more engaged in corruption (with respect to micro firms); the latter
result is consistent with findings of Aterido et al. (2011), who report that
micro firms (less than 10 employees) tend to be proportionally less affected
by a weak environment. Consistent with our prior expectations, an exten-
sive contact with public officials (FormalSect) increases the likelihood
to be extorted bribes (Svensson 2003). In line with the existing literature
(e.g., Seker and Yang 2012), we also find that younger firms are the most
targeted by public officials. Together with the negative sign for the concen-
tration variable this would suggest that longer and better (concentrated)
established firms are less likely to be extorted informal payments. Similarly,
managers secure of their property and contract rights (Property) and able
to secure fair treatment by public officials (Fair_bureaucracy) are less
prone to administrative corruption and thus less likely to seek individual
protection by buying bureaucrats’ favors (Hellman et al. 2003; Batra et
al. 2003). Sector specificities seem to be an important determinant of the
corrupt behavior of entrepreneurs. As expected firms in services, trade,
and transport are highly impacted by corruption.
Column 2 shows the results from estimating the same model for the
variable incidence of gifts for obtaining a contract (GiftForContr). The
dependent variable takes the value of 1 if in the past 12 months a firm
made an informal payment to obtain a public contract and 0 otherwise.
Also, for this form of corruption women entrepreneurs are over 7 percent
less likely to engage in informal payments. While for most variables results
do not change relevantly, a few differences are noticeable. In particular,
medium and larger enterprises seem to be significantly less involved in
bribing, while the coefficient on age is no more significant. This is most
probably related to their established networks with public managers and
their market leading position. Similarly, being established in the country-
side lowers the probability of being involved in this kind of corruption.
Columns 3 and 4 contain OLS estimations of equation (2). The depend-
ent variables are the amount of administrative corruption (expressed in
percentage points of sales) and of gifts for obtaining a contract (in per-
centage points of the contract value). The results do show lower bribes
by female-owned enterprises relative to their male counterparts, but the
respective coefficients are not statistically significant. Consistent with the
bribe bargaining theory, large firms and members of a network pay, in
relative terms, less (note that large firms’ payment will be higher in absolute
LINK TEXT.indd 185 24/02/2017 15:55

186 Gender and Entrepreneurial Activity
terms). Conversely, the idea that firms with more refusal power pay less is
not supported by the data. The concentration rate proxies for the diversifi-
cation of a firm’s business where more concentrated businesses are believed
to have less mobility across sectors and have therefore less refusal power.
This variable is negatively, but insignificantly correlated with payments. On
the other hand, a firm’s refusal power seems to be better captured by the
entrepreneurs’ consciousness of their property rights (Property) or of their
ability to defend themselves from a corrupt bureaucrat (Fair_bureauc).
There are no significant differences in the amount paid across sectors.
In sum, the results from Table 8.3 provide some evidence that women
entrepreneurs are less likely to engage in corruption. In particular, women
have a lower probability to be involved in administrative corruption and
to offer gifts to public officials to obtain contracts but we do not find any
gender differences in the amount paid.
Gender Differences by Size, Sector, and Risk Attitude
As a second step, we extend the analysis on corruption by looking at the
bribe incidence at different firm sizes, in different sectors, and with dif-
ferent risk attitudes. Because size effects on corruption are not linear and
change depending on the type of corruption, we investigate if gender
propensities to bribe change with the company’s size. This is because a
lower propensity of women to bribe public officers might be related to the
fact that the majority of them run a micro enterprise, lack a network of
contacts, and/or have fewer occasions to meet with public officials. In this
case, the gap in moral standards could be smaller (or even reverse) in larger
enterprises where women entrepreneurs need to interact more with govern-
ment bureaucracy and fellow businesspeople to survive the competition.
Similarly, operating in the trade and transport sectors most likely involves
the acquisition of more permits and licenses with respect to other sectors,
while the services sector might be less reliant on permits and more reliant
on public infrastructures. As a consequence, the higher concentration of
women in the service sectors could be driving the different propensities.
To investigate the role of different risk attitudes on corruption we recur
to a measure of tax evasion. The variable Workforce is the percentage of
actual workforce that entrepreneurs declare to the authorities for tax pur-
poses. Entrepreneurs who underreport on this measures are evading taxes
and are, thus, if detected by the authorities, subject to fines. Hence, tax
evaders should have a high propensity to take risk in the sense that they
believe detection is unlikely. We create a dummy variable called HighRisk
taking the value 1 if the declared workforce is below the sample average
and 0 otherwise. An extensive literature presents women as more risk
LINK TEXT.indd 186 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 187
averse; as a consequence, it could be that this different attitude is driving
their bribing behavior.
To verify these additional hypotheses, we augment the benchmark
regressions with interactions of these variables with the dummy for female
ownership.
The results are reported in Table 8.4.
For space concerns we only show the coefficients on the interaction
terms. Panel A reports the results of the interactions by size. The coeffi-
cients on Female do not change significantly from previous results showing
a lower likelihood of women-led micro firms to engage in administrative
corruption and bribing activity to obtain a contract. Further, the results
indicate that on average the incidence of administrative corruption is
higher for men running bigger companies (with respect to micro enter-
prises) and that for females the differences across size categories are not
statistically significant. In contrast, the incidence of gifts for contracts
is lower for large men-owned enterprises but is significantly higher for
women-led firms of comparable size.
Panel B shows the results of the interactions by sector. The coefficient
on Female now refers to firms in the manufacturing sector. Men-owned
firms in the trade and transport and services sectors have a much higher
likelihood of being involved in bribes (with respect to the manufacturing
sector); however, this likelihood does not significantly vary by sector for
females.21
Panel C shows that while men with a high-risk attitude are involved
in corruption significantly more than their low-risk peers, for women
incidence of corruption does not seem to vary significantly according to
risk attitudes.
These results confirm that there are ample variations in corruption
across the dimensions analyzed. However, only for specific types of cor-
ruption are these relevant and statistically significant for explaining the
gender gap. There is a difference by size: for large women-led firms the
gender gap in corrupted behavior is reversed, and in the case of gifts to
obtain public contracts this difference is large and statistically significant.
This suggests that women managing companies of this size category do
not have different moral standards from their male peers.
Robustness Checks
Sample selection
In the previous results, equation (2) has been estimated by OLS. However,
findings may be affected by selection bias. The selection bias arises if
the error terms ni from the incidence equation (1) and eiv from equation
LINK TEXT.indd 187 24/02/2017 15:55

188 Gender and Entrepreneurial Activity
Table 8.4 Determinants of corruption size, sector, and risk attitude
interactions
Administrative
corruption
(1)
Gifts to obtain
contract
(2)
PANEL A: size interactions
Femaleown −0.049
[0.022]**
−0.086
[0.016]***
Fem XSE 0.025
[0.043]
0.032
[0.040]
Fem XME −0.004
[0.058]
−0.031
[0.069]
Fem XLE 0.15
[0.098]
0.273
[0.084]***
small entr 0.075
[0.025]***
0.009
[0.026]
medium entr 0.038
[0.028]
−0.055
[0.026]**
large entr −0.01
[0.041]
−0.116
[0.028]***
PANEL B: sector interactions
Femaleown −0.038
[0.034]
−0.091
[0.023]***
Fern X trade and transport −0.009
[0.040]
0.032
[0.039]
Fern X services 0.010
[0.031]
0.036
[0.038]
trade and transport 0.061
[0.023]***
0.037
[0.017]**
services 0.137
[0.026]***
0.08
[0.019]***
PANEL C: risk attitude interactions
Female owned −0.038
[0.020]*
−0.077
[0.018]***
Fem X HighRisk 0.023
[0.038]
0.026
[0.042]
HighRisk 0.201
[0.023]***
0.179
[0.021]***
Observations 4,966 3,862
Notes: Columns 1 and 2 show marginal effects after probit estimation evaluated at the
mean. The regressions include industry and country dummies. The specification is the same
as in Table 8.3; coefficients are not reported as they are basically unchanged. Asymptotic
cluster-robust standard errors (clustering by country) are reported in brackets. *significant
at 10%, **significant at 5%, ***significant at 1%.
LINK TEXT.indd 188 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 189
(2) are correlated. This occurs because in equation (2) we only analyze
informal payments for a sample of corrupt entrepreneurs who could
have very different characteristics with respect to their honest peers. To
correct for selection bias, we estimate Heckman-type selection models.22
Identification of the selection equation (1) requires a variable that deter-
mines the probability of paying bribes, but is irrelevant in the main equa-
tion of interest; in other words, it does not affect the amount paid. As
an instrument for the identification of the selection equation we use the
variable (Workforce). As we already discussed, we interpret this variable as
a measure of risk aversion. More risk-averse entrepreneurs are less willing
to operate in the shadow economy and engage in either tax evasion or
bribing activities. However, once firms select into the informal economy,
the bargaining process over the amount paid is not determined by their
risk aversion but rather by their refusal power. The results are presented in
Appendix A Table 8A.3.
In Appendix A Table 8A.3 we do not report the results of the selection
equation as this is very similar to the results in Table 8.3. However, we
report the coefficients on the instrument chosen which is highly significant
in both regressions. We find evidence of selection in both equations, and
the coefficient on Female further decreases but it is still not statistically
significant. The other coefficients also do not change significantly.
Omitted individual characteristics
Even though the BEEPS contains a rich set of firm-level variables, it con-
tains little information about characteristics of firm owners/ managers.
Because of the formulation of the relevant question about corruption
(“what a typical firm pays”), answers could be related to the experience
and knowledge entrepreneurs have of their environment rather than to
their firm’s characteristics. Consequently, women, being less experienced
as managers and participating less in formal networks, could in fact under-
report on corruption. In other words, the omission of some individual-
level variables could introduce a bias in our results. We address this
possible bias following a strategy similar to Blanchflower et al. (2003) and
Murayev et al. (2009). We use several sample splits and compare regres-
sion results for groups of firms that differ in the extent to which personal
experience and networking ability should influence bribe exposure. In par-
ticular, the sample is divided on the basis of the ages of firms, which is a
proxy of entrepreneurs’ experience, and on their participation in networks,
such as business associations and/or chambers of commerce. The idea
behind splitting the sample based on network participation is that such
participation may provide an idea about an entrepreneur’s ability to deal
with formal and informal institutions.
LINK TEXT.indd 189 24/02/2017 15:55

190 Gender and Entrepreneurial Activity
The results from estimating the basic models for the subsamples are
reported in Appendix A Table 8A.4. As in the above analysis, the depend-
ent variables measure incidence of administrative corruption and gift for
obtaining a contract. For space considerations, the table shows only the
coefficients on the gender dummy, Female, estimated for each pair of sub-
samples. Panel A shows the results for the sample split based on average
age. Despite some differences in the coefficients on the gender variable in
the two subsamples, none of the differences are statistically significant.
Similarly, there are no statistically significant differences in the coefficients
on the female dummy in the sample split based on network participation
(see Panel B). We therefore conclude that the basic results for gender dis-
crimination obtained in this study are sufficiently robust and are unlikely
to be driven by the omission of essential variables in the regressions.
Having determined the main factors influencing corrupt behavior of
entrepreneurs and the gender differentials, we next turn to the conse-
quences of corruption on firm growth with a special focus on gender
differences.
5 ECONOMETRIC MODEL OF THE IMPACT OF
CORRUPTION ON FIRM GROWTH
Specification
A growing body of literature is using enterprise surveys to investigate the
impact of the “business climate” on firm performance, often including
corruption in the factors shaping the operating environment of firms. This
literature identifies bribery and corruption as major problems for enter-
prises especially in middle-income countries (Gelb et al. 2007), and finds
a negative impact of corruption on firms’ productivity (Escribano and
Guasch 2005), sales growth (Beck et al. 2005; Fisman and Svensson 2007;
Ayyagari et al. 2008; Seker and Yang 2012; Gaviria 2002; Batra et al. 2003),
investment growth (Asiedu and Freeman 2009; Gaviria 2002; Batra et al.
2003), or employment growth (Aterido et al. 2007 and 2011).
From the bargaining theory analyzed in the previous section, it is easy to
derive support for a negative relationship between administrative corrup-
tion and firm growth (sand the wheels). Managers spending longer time
in dealing with bureaucrats penalize their enterprise development while
informal payments reduce companies’ profits and the incentives to increase
profitability. Further, bribing can subtract important resources from the
young and small companies.
On the other hand, analyzing bribing for winning contracts, expectations
LINK TEXT.indd 190 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 191
are that engaging in this activity will lead to significant competitive advan-
tages to enterprises which are big enough to be able to change or bend
rules. Moreover, taking into consideration the highly corrupt environment
prevalent in transition economies, “greasing the wheels” of bureaucracy
could alleviate the burden of public procedures and regulations (Méon
and Weill 2010). We add a gender dimension to the existing literature on
the impact of corruption on firm growth. As shown in the previous section
there are gender differences in the propensity to bribe. In this section, we
investigate if these differences are important for the economic perfor-
mance of firms.
The specification used for the empirical estimation of the impact of cor-
ruption on firm growth is common in the above-cited literature and is the
following:
DLi 5 l0 + l1gi + l2gi * Femalei + l3 Femalei + l4Xi + xi (3)
where Li is annual employment growth for firm i; g is a variable to reflect
the corruption measure – the incidence of bribery or the amount paid aver-
aged over all observations (not payers are assigned a zero); g*Female is the
interaction term of the corruption measure and the indicator for female-
owned and managed companies; and Xi is a vector of control variables
already presented in the previous section. These include dummy variables
for the following variables: foreign ownership; privatization; if the firm is
involved in trade, either importing or exporting; if the company was pri-
vatized; the firm’s size. Company’s age and its squared term are continuous
variables. Finally, a full set of sector, geographic location, and country
interaction dummy variables are included in all regressions to control for
local and sector demand and business environments and to alleviate the
problem of omitted variables. Also, in these regressions we allow for heter-
oskedasticity and cluster errors by country. Appendix A Table 8A.5 shows
the correlations among these variables.
In estimating equation (3) the natural question that arises is how to best
control for possible endogeneity related to the independent variables and
reverse causality. In particular, the relation between bribery and growth
can go two ways depending on the type of corrupt behavior. On the one
hand, public officials can assess and select firms better suited, hence with
better performance and growth potential to carry the financial burden of
a bribe. On the other hand, successful entrepreneurs can buy their market
advantage from public workers or regulators and hence bribery can
become a factor for firm growth.
As an instrument for firm bribery we use sector–location–country
cluster averages of responses. Our identification strategy thus assumes
LINK TEXT.indd 191 24/02/2017 15:55

192 Gender and Entrepreneurial Activity
that in addition to an idiosyncratic component, whether or not a firm is
bribed is determined by the underlying characteristics of the particular
sector–location–country of the firm.23 Several studies have used average
answers to identify the impact of the business climate – among which is
corruption – on firm performance measures.24 This strategy assumes that
a cluster’s average level of bribery is uncorrelated to a firm’s unobservable
characteristics that affect the probability of being solicited for a bribe.
Countries have very different corruption levels and this is uncorrelated
to unobservable firm characteristics. To minimize reverse causality, the
cluster average assigned to each firm excludes a firm’s own answer; thus,
while sector–location–country averages impact a firm’s performance, the
single firm characteristics do not influence the cluster average. To ensure
adequate numbers of firms in each cell, we drop one dimension of the cell
until an adequate number is reached.25 Also, using grouped averages allows
us to mitigate the effects of measurement error, a likely concern for bribery
data given their secretive nature.
As an additional instrument we use a measure of the control public offi-
cials maintain over the firms: the percentage of time (timeReg) managers
devote to dealing with authorities.26 Wang and You (2012) and Aterido et
al. (2011) use a similar form of this variable as an objective measure of cor-
ruption; in particular, they focus on time spent with authorities to obtain
permits and licenses. Their variable mainly captures administrative corrup-
tion, but its impact on the incidence of other forms of corruption (gift for
contract) is not warranted. In contrast, timeReg has a wider correlation
with corruption and the institutional and operational background of the
enterprises as it measures the time spent by senior managers dealing with
public officials about the application and interpretation of laws and regula-
tions and to get or maintain access to public services. Thus, while it captures
a more country- or regional-wide environmental feature (the complexity
of laws/regulations, or the accessibility of public services), it also contains
an idiosyncratic component represented by the public officials’–firms’ bar-
gaining relationship. As shown by the correlation table (Appendix A Table
8A.5), it has no direct effect on growth, while it is strongly correlated with
all corruption measures. The instruments are interacted with the dummy
Female to estimate the differential impact of corruption on genders.
Analysis of Results
In Table 8.5 we report the results from the OLS estimates of equation (3),
controlling for sector, location, and country interaction dummies.
The impact of administrative corruption or gifts to obtain a contract on
firm growth seems to follow a similar pattern with significant coefficients
LINK TEXT.indd 192 24/02/2017 15:55

193
Table 8.5 Total employment growth regressions fixed effects (country, sector, rural–urban). Dependent variable yearly
employment growth rate.
(1) (2) (3) (4)
Admin. Corr. (dummy) −0.016
[0.004]***
FemXAdmCorr (dummy) 0.033
[0.008]***
Admin. Corr. (amount) −0.004
[0.001]***
FemXAdmCorr (amount) 0.006
[0.002]***
Gift for Contract(dummy) −0.013
[0.007]*
FemXGift4Cont(dummy) 0.033[0.009]***
Gift for Contract(amount) −0.002
[0.001]**
FemXGift4Cont(amount) 0.004
[0.001]**
Femaleown −0.014
[0.005]***
−0.009
[0.004]**
−0.013
[0.005]**
−0.011
[0.005]*
small entr −0.041
[0.008]***
−0.041
[0.008]***
−0.041
[0.009]***
−0.041
[0.009]***
medium entr −0.084
[0.018]***
−0.085
[0.018]***
−0.093
[0.017]***
−0.093
[0.017]***
large entr −0.106
[0.023]***
−0.107
[0.023]***
−0.117
[0.022]***
−0.117
[0.022]***
LINK TEXT.indd 193 24/02/2017 15:55

194
Table 8.5 (continued)
(1) (2) (3) (4)
InK 0.018
[0.003]***
0.018
[0.003]***
0.019
[0.003]***
0.019
[0.003]***
foreign_own 0.006
[0.009]
0.005
[0.009]
0.008
[0.011]
0.008
[0.011]
privatization −0.03
[0.010]***
−0.03
[0.010]***
−0.026
[0.009]***
−0.026
[0.009]***
trade 0.024
[0.005]***
0.024
[0.005]***
0.025
[0.005]***
0.025
[0.005]***
ageCompany −0.002
[0.000]***
−0.002
[0.000]***
−0.002
[0.000]***
−0.002
[0.000]***
ageComp2 0.001
[0.000]***
0.001
[0.000]***
0.001
[0.000]***
0.001
[0.000]***
Constant −0.011
[0.011]
−0.012
[0.011]
−0.012
[0.012]
−0.012
[0.012]
Observations
R-squared
4,966
0.07
4,966
0.07
4,175
0.07
4,175
0.07
Notes: The regressions include sector–location–country interaction dummies. Asymptotic cluster-robust standard errors (clustering by country)
are reported in brackets. The dependent variable is yearly total employment growth. In each column we analyze the impact of a different form of
corruption: (1) incidence of administrative corruption, (2) amount paid for administrative corruption, (3) incidence of gifts to obtain government
contracts, (4) amount paid to obtain a contract. *significant at 10%, **significant at 5%, ***significant at 1%
LINK TEXT.indd 194 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 195
similar in size. This leads to the conclusion that bribing has a negative
impact on firm growth. However, the positive coefficient of the interaction
with the female dummy suggests that for women entrepreneurs informal
payments could contribute to their companies’ growth. This conclusion
could support the idea that for women entrepreneurs – often operating
micro enterprises – bureaucracy is a too heavy burden in general and
informal payments might represent a means to simplify their manage-
ment and contribute to firm growth. Surprisingly, this positive effect for
women-owned or -managed firms is found also for administrative cor-
ruption (incidence or amount paid cols 1 and 2), lending some support to
the greasing the wheels hypothesis under very heavy bureaucratic control.
Adding to this hypothesis, the coefficient on the Female dummy indicates
that women-owned firms have a growth rate on average of 1 percent lower
than men-owned companies.
The effects of administrative corruption in terms of amount paid on
firm growth as reported in column 2 provide a clear picture of the very dif-
ferent impact they have on women’s as oppose to men’s owned companies.
While for men paying 1 percent of bribes leads to a 0.4 percent drop in
firm growth rate, for female-owned companies this would actually imply a
higher growth rate of 0.2 percent. Similarly, men-owned firms paying gifts
for contracts experience a decrease in firm growth of 0.2 percent while
their women-owned peers could in fact grow 0.2 percent (column 4). This
could be consistent with the scenario described by Wang and You (2012)
in China where corruption can be positive for firm growth in the presence
of underdeveloped financial markets. Hence, the differential and positive
impact of corruption on women entrepreneurs could be linked to their
difficulties in accessing credit.27
Consistent with the literature on firm growth, the rest of the coefficients
indicate that micro, younger, trading firms as well as those with a higher
capital stock are growing faster. The negative relationship between growth
and age and size is a well-established result (see for example Mansfield
1962; Evans 1987; Hall 1987; and Dunne and Hughes 1994 more recently
and for developing countries Mead and Liedholm 1998 and Nichter and
Goldmark 2009). Similarly, literature has established that trading compa-
nies are more dynamic, have higher productivity, and grow faster as shown
in a number of studies such as Pavcnik (2002), Van Biesebroeck (2005),
Lileeva and Trefler (2010), and Aw et al. (2011). Foreign ownership has
been found to have no effect on growth, while privatized companies grow
on average 3 percent less.
We report the results of the two-stage regressions using the sector–
country averages of reported bribes and the measure of time spent in
dealing with regulations (timeReg) in Table 8.6. The coefficients and their
LINK TEXT.indd 195 24/02/2017 15:55

196 Gender and Entrepreneurial Activity
Table 8.6 Total employment growth IV regressions using sector–country
averages and time spent in dealing with regulations as
instruments. Dependent variable yearly employment growth rate.
(1) (2) (3) (4)
Admin. Corr. (dummy) −0.018
[0.006]***
FemXAdmCorr (dummy) 0.022
[0.020]
Admin. Corr. (amount) −0.007
[0.002]***
FemXAdmCorr (amount) 0.012
[0.007]*
Gift for Contract (dummy) −0.021
[0.011]*
FemXGift4Cont (dummy) 0.083
[0.034]**
Gift for Contract (amount) −0.004
[0.001]***
FemXGift4Cont (amount) 0.012
[0.005]***
Femaleown −0.011
[0.006]**
−0.014
[0.005]***
−0.025
[0.007]***
−0.02
[0.006]***
small entr −0.039
[0.008]***
−0.04
[0.008]***
−0.041
[0.009]***
−0.039
[0.009]***
medium entr −0.085
[0.017]***
−0.086
[0.018]***
−0.092
[0.016]***
−0.092
[0.016]***
large entr −0.107
[0.024]***
−0.108
[0.024]***
−0.117
[0.022]***
−0.117
[0.022]***
lnK 0.019
[0.003]***
0.018
[0.003]***
0.019
[0.003]***
0.019
[0.003]***
foreign_own 0.005
[0.010]
0.005
[0.009]
0.011
[0.011]
0.011
[0.011]
privatization −0.029
[0.010]***
−0.03
[0.010]***
−0.023
[0.008]***
−0.023
[0.008]***
trade 0.025
[0.005]***
0.024
[0.005]***
0.024
[0.005]***
0.023
[0.005]***
ageCompany −0.002
[0.000]***
−0.002
[0.000]***
−0.002
[0.000]***
−0.002
[0.000]***
ageComp2 0.001
[0.000]***
0.001
[0.000]***
0.001
[0.000]***
0.001
[0.000]***
J-Test 1.636 1.261 3.233 4.114
p-value [0.441] [0.532] [0.199] [0.128]
First stage F-test
corruption 23.97 28.1 27.64 49.53
corr.Xfemale 130.51 10.59 21.91 44.75
endogeneity test 0.459 2.767 4.786 3.346
p-value [0.795] [0.251] [0.091] [0.188]
Observations 4,713 4,713 4,056 4,056
LINK TEXT.indd 196 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 197
significance are surprisingly similar to the OLS28 results but increase in
size in most cases. For example, in the case of gifts for public contracts
(column 4), paying a bribe of 1 percent for men-owned firms is associ-
ated with a fall in growth of 0.4 percent and an increase of women-owned
firms’ employment rate of 8 percent. At the same time the coefficients on
the variable Female also increase indicating a bigger starting growth gap.
The F-statistics of our instruments’ (sector–country averages and
timeReg) joint significance in the first-stage regression are usually above
20 and are highly significant. The validity of the instruments (whether
they are uncorrelated with the error process in equation (3)) is tested, and
the null hypothesis that the instruments are valid cannot be rejected by the
Hansen J-statistic.
Robustness Tests
Some of the explanatory variables in equation (3) could potentially be
endogenous. In particular, the effect of inputs might be related to possible
unobserved heterogeneity. In Appendix A Table 8A.6 we treat capital (lnK)
and the measure indicating if a firm is engaged in trade (trade) as endog-
enous. Following Commander and Svejnar (2007) we use as instruments
for the capital input and the trade orientation of the firm the percentage
change in fixed assets and in exports in the previous three years (changeA,
changeE).29 The endogeneity test confirms that capital stock and engage-
ment in foreign trade activities cannot be treated as exogenous and their
coefficients increase several times. Also the estimates for the corruption
measures roughly double.30
Another source of potential endogeneity is self-selection, with better
Table 8.6 (continued)
Notes: The regressions include sector–location–country interaction dummies. Asymptotic
cluster-robust standard errors (clustering by country) are reported in brackets. The
dependent variable is yearly total employment growth. In each column we analyze the
impact of a different form of corruption: (1) incidence of administrative corruption, (2)
amount paid for administrative corruption, (3) incidence of gifts to obtain government
contracts, (4) amount paid to obtain a contract. *significant at 10%, **significant at 5%,
***significant at 1%. F-test on instruments is the test statistic on the F-test of the joint
significance of the instruments (3-sector–country averages and timeReg) in the first-
stage regressions, with p-values in brackets. Hansen J-statistic is the test statistic on the
overidentification test of the instruments, with p-values in brackets. The endogeneity test
is defined as the difference of two Sargan-Hansen statistics: one for the equation with
the smaller set of instruments, where the corruption measures (AdmCorr, Gift4Con) are
treated as endogenous, and one for the equation with the larger set of instruments, where
the corruption measures are treated as exogenous. The endogeneity test statistic is robust to
heteroskedasticity.
LINK TEXT.indd 197 24/02/2017 15:55

198 Gender and Entrepreneurial Activity
performing firms choosing locations based on the quality of the business
environment. We verify this by restricting the sample to micro enterprises
and SMEs who are least likely to be footloose and where the location
tends to be associated with the places of birth or residence of the principal
owner. Appendix A Table 8A.7 shows that the negative impact of corrup-
tion is even stronger for this group of enterprises.
The results so far have pooled countries together, but it could be that the
effects vary by the degree of institutional transition to market economies
of these countries. To verify this, in Appendix A Table 8A.8 we drop the
old EU countries. Contrary to Aterido et al. (2011) we find that results are
stronger for transition economies where corruption is more widespread
and the rule of law weaker.
6 CONCLUSIONS
In this chapter we analyze corrupt behavior of entrepreneurs, focusing
on gender differentials in the determinants of firms behavior and in the
impact of corruption on firm growth. Using data relating to responses
from 5,471 firms across 31 countries in Europe and Central Asia in the
2005 wave of the BEEPS, we bring empirical evidences to the hypothesis
that genders have different propensities to corruption. We also estimate
the impact of informal payments on firm employment growth and analyze
if corruption is a gender-specific obstacle to entrepreneurship develop-
ment. When estimating the econometric equations on the determinants
of corrupt behavior and on the impact of corruption on firm growth, we
make use of two measures of corruption: administrative corruption and
informal payments for obtaining a contract. We analyze both the incidence
and the amount paid as bribes for each of the cases.
Our findings confirm that female-owned firms are less prone than male-
owned to make informal payments. In contrast, we do not find any gender
difference in the amount paid. Analyzing the determinants of corruption,
we find support for Svensson’s (2003) idea that companies dealing exten-
sively with the public sector face higher probabilities of being extorted
undue payments.
Further findings suggest that factors affecting corruption in its various
measures, and hence motivation, could differ. The results highlight that
there is a specific combination of variables reflecting a firm’s characteris-
tics (size, sector, rural), its business environment, and its flexibility to adjust
(age, mobility) that determines the ways and extent of a firm’s engagement
in corruption. Among these variables, a firm’s networking capacity and
firm size affect positively the probability of bribing officials to win a public
LINK TEXT.indd 198 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 199
contract. In particular, we find some supporting evidence that these gender
bribing differentials are correlated with firm size and gender gaps in risk
attitudes. In fact, when we study gender differentials by firm size, we find
that women leading big enterprises are actually more likely to offer gifts to
public officials to obtain a public contract.
In a second group of findings, we provide new evidence on the role of
corruption in explaining employment growth. While results confirm that
bribing has a negative impact on firm growth in general, significant gender
differentials are found in terms of a positive growth impact of bribing on
female-owned firms. Starting from lower growth rates, for female-owned
firms informal payments could represent a means to smooth management
and improve performance.
In sum, we find significant differences across the moral standards of the
two genders possibly motivated by the dimension and developmental phase
of their firm and by different risk attitudes. Moreover, the differential
impact of grafting on firm growth provides a new interpretation of corrup-
tion as an explanatory factor for existing gender gaps in entrepreneurship.
The contribution of our chapter therefore consists in identifying gender
differentials in firms’ behavior regarding various measures and motivations
of corruption as well as in the different role of corruption on employment
growth of female-owned firms versus male-owned ones.
Our results could be used to identify priorities in economic policies
specifically targeting corruption. For the emerging markets in Eastern
Europe and Central Asia, these measures and policies are crucial to build the
fundamentals of a strong middle class of both women- and men-owned and
managed companies. Inclusive growth requires increasing the potential of
women and men to sustain the private sector performance through building
a friendly and less corrupt administration, an efficient system of regulation,
licenses and taxes that are not a burden, and an environment where corrup-
tion cannot thrive. Strengthening property rights and security to perform
fair business is another area where policies could directly contribute to
mitigate the adverse effects of corruption on growth and gender gaps.
NOTES
1. For a review of the literature on laboratory corruption experiments see Frank et al.
(2011).
2. For a review see Klapper and Parker (2011).
3. The development of small enterprises is a key factor for poverty alleviation and sustain-
able growth in many countries. This in turn could address the issue of “feminization of
poverty” (Chen et al. 2005). In fact, according to recent statistics, about a fifth to a third
of all small entrepreneurs in the world are women.
LINK TEXT.indd 199 24/02/2017 15:55

200 Gender and Entrepreneurial Activity
4. Aterido et al. (2007, 2011) find non-linear effects of corruption on firm growth with
micro and small firms growing faster in a more corrupt environment while this creates
bottlenecks for medium and large firms.
5. Banerjee et al. (2012) develop a model that is more flexible allowing the capturing of
specific “environment” features such as nature of the task, bureaucrats’ characteristics,
and institutional framework. Given the very typical types of corruption we focus on in
this study, we will not make use of their framework but rather follow Svensson’s (2003)
model and empirical strategy.
6. A growing body of literature analyzes the impact of female leaders on policy and politi-
cal outcomes (especially public expenditures). See for example Ferreira and Gyourko
(2014) or Chattopadhyay and Duflo (2004).
7. As pointed out by Sung (2003), macro results are likely to be driven by the countries’
constitutional liberalism (judiciary and press), where liberal political systems promote
gender equality and better governance.
8. We adopt here a classical economic view (and liberal feminism theories, see for example
Ahl 2006) that considers men and women as essentially equal; as a consequence, control-
ling for firm level and owner’s differences should account for differences in enterprise
performance and thus corruption.
9. In 2009 the BEEPS was restructured to improve cross-country comparability and to
make it compatible with the Enterprise Surveys of the World Bank. There were changes
in the questionnaire and in the methodology. For more details see http://ebrd-beeps.
com/about/.
10. Bulgaria and Romania are included in the New Member State group even if they
accessed to the EU only in 2007. However, Steves and Rousso (2003) and Budak and
Goel (2006) show that for transition economies a near-term prospect to access EU has
a positive effect on corruption fighting. On the other hand we left Croatia in the SEE
states group as it only accessed in 2013.
11. Corruption is likely to be the area where willingness to be truthful may be particularly
delicate, but Hellman et al. (2000) and Svensson (2003) find little evidence in bias in
responses to questions of corruption given the phrasing of the questionnaire and the
overall list of constraints.
12. We thank an anonymous referee for highlighting this point.
13. This approach is similar to the one followed by Blanchflower et al. (2003) and Murayev
et al. (2009) for missing individual variables.
14. When the government is a buyer or a contractor, a corrupt entrepreneur may pay to be
included in the list of qualified bidders, to have public officials structure the bidding
specifications such that he or she is the only qualified supplier, or to be selected as the
winning contractor. Once selected, he or she may pay to charge inflated prices or to
skimp on quality (Rose-Ackerman 1996).
15. Aterido et al. (2007, 2011), however, use only permanent employment growth as does
Davis and Haltiwanger (1992, 1999) as they analyze a bigger sample of companies.
16. This most likely introduces some imprecision in the measures of assets/sales growth
leading to somewhat weaker results. Similarly, past profits are not reported; instead
companies are asked if they had positive income in 2003. Findings for assets and sales
growth are available from the authors upon request.
17. More profitable/bigger companies are expected to pay more in absolute level. However,
research has shown that bigger firms pay less in relative terms (in percentage of sales or
of contract).
18. This variable is the principal component of three other variables: see Appendix A Table
8A.2. We prefer to combine the explanatory variables into a “formal sector index” to
avoid multicollinearity problems.
19. Since the age of a company is not expected to influence corruption in a linear manner,
both age and the age squared (age2) are included as variables.
20. Adding variables related to countries’ macroeconomic conditions is thus not necessary
as there is no time dimension in the dataset.
LINK TEXT.indd 200 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 201
21. While the interactions with Female are not statistically significant for AdmCorr, the
resulting coefficients for women entrepreneurs in the trade and transport and services
sectors are significant and negative, suggesting that these two sectors drive most of the
gap among genders.
22. As in standard bivariate models we assume that the errors are distributed bivariate
normal with mean zero and correlation rho.
23. Identification relies on the assumption that corruption has an additive structure with a
component firm specific and one location and sector specific. Gik
5gik+avGk where Gik is
a firm’s individual bribery statistic, avGk is the average corruption for cluster k, and gik is
the firm-specific element. Using group averages as instruments is a common technique
described in Krueger and Angrist (2001).
24. Fisman and Svensson (2007) and Seker and Yang (2012) use industry-location averages
to show that bribes negatively impact firm growth. There is then a range of studies,
including Aterido et al. (2011), Commander and Svejnar (2007), Hallward-Driemeier
etal. (2006), Ayyagari et al. (2008) and others, which use location averages to examine
the impact of business environment on firm performance. For an extensive overview of
this literature see Dethier et al. (2011).
25. The minimum size of each cell is 15. The principal results are obtained using 3 sectors-
country averages; when the cell is smaller than 15 we compute 2 sectors (just manufac-
turing and services) – country averages if the problem persists we just compute country
averages. This happens only in 1 case. Results using location (rural–urban)–country
averages do not change significantly and are available from the authors upon request. In
this case there are no cells smaller than 15.
26. This instrument is proposed and used also by Fisman and Svensson (2007).
27. For the same region and based on the same data, Murayev et al. (2009) find significant
differences across genders in accessing credit.
28. In fact, the endogeneity C-test cannot reject the null hypothesis that the various
corruption measures can be treated as exogenous.
29. Commander and Svejnar (2007) also instrument labor input; to construct our size
variable we use the lagged value of permanent employment, minimizing in this way any
possible endogeneity problem. They also treat as endogenous the ownership status of
the company. We also instrument privatization with the year of privatization, and the
results do not change significantly.
30. Results treating the corruption measures as exogenous are hardly different from the
ones shown in Appendix A Table 8A.6.
REFERENCES
Aidis, R., F. Welter, D. Smallbone, and N. Isakova (2007), ‘Female entrepreneur-
ship in transition economies: The case of Lithuania and Ukraine’, Feminist
Economics, 13(2): 157–183.
Ahl, Helene (2006), ‘Why research on women entrepreneurs needs new directions’,
Entrepreneurship Theory and Practice, 30(5): 595–621.
Andersen, S., E. Bulte, U. Gneezy, and J. List (2008), ‘Do women supply more
public goods than men? Preliminary experimental evidence from matrilineal and
patriarchal societies’, American Economic Review: Papers and Proceedings, 98(2):
376–381.
Asiedu, E., and J. Freeman (2009), ‘The effect of corruption on investment growth:
Evidence from firms in Latin America, Sub-Saharan Africa, and transition
countries’, Review of Development Economics, 13(2): 200–214.
Aterido, R., and M. Hallward-Driemeier (2009), ‘Comparing apples with . . .
LINK TEXT.indd 201 24/02/2017 15:55

202 Gender and Entrepreneurial Activity
apples: How to make (more) sense of subjective rankings of constraints to busi-
ness’, World Bank Policy Research Working Paper No. 5054, The World Bank.
Aterido, R., M. Hallward-Driemeier, and C. Pages (2007), ‘Investment climate and
employment growth: the impact of access to finance, corruption and regulations
across firms’, IZA Discussion Papers No. 3138, Institute for the Study of Labor
(IZA).
Aterido, R., M. Hallward-Driemeier, and C. Pages (2011), ‘Big constraints to small
firms’ growth? Business environment and employment growth across firms’,
Economic Development and Cultural Change, 59(3): 609–647.
Aw, B.Y., M.J. Roberts, and D.Y. Xu (2011), ‘R&D investment, exporting, and pro-
ductivity dynamics’, American Economic Review, 101(4): 1312–1344.
Ayyagari, Meghana, Asli Demirgüç-Kunt, and Vojislav Maksimovic (2008), ‘How
important are financing constraints? The role of finance in the business environ-
ment’, World Bank Economic Review, 22(3): 483–516.
Banerjee, A., S. Mullainathan, and R. Hanna (2012), ‘Corruption’, in Robert
Gibbons and John Roberts (eds), The Handbook of Organizational Economics.
Princeton, N: Princeton University Press, pp. 1109–1147.
Batra, G., D. Kaufmann, and A.H.W. Stone (2003), Investment Climate around
the World: Voices of the Firms from the World Business Environment Surveys.
Washington, DC: World Bank.
Beck, T., A. Demirguc-Kunt, and V. Maksimovic (2005), ‘Financial and legal
constraints to growth: Does firm size matter?’, Journal of Finance, 60(1):
137–177.
Berninghaus, S.K., S.Haller, T. Krüger, T. Neumann, S. Schosser, and B. Vogt
(2013), ‘Risk attitude, beliefs, and information in a Corruption Game – an experi-
mental analysis’, Journal of Economic Psychology, 34(February): 46–60.
Blanchflower, D.G., P.B. Levine, and D.J. Zimmerman (2003), ‘Discrimination
in the small-business credit market’, Review of Economics and Statistics, 85(4):
930–943.
Borghans, L., J.J. Heckman, B.H.H. Golsteyn, and H. Meijers (2009), ‘Gender
differences in risk aversion and ambiguity aversion’, Journal of the European
Economic Association, 7: 649–658.
Budak, J., and R.K. Goel (2006), ‘Corruption in transition economies: Effects of
government size, country size and economic reforms’, Journal of Economics and
Finance, 30(2): 240–250.
Campos, N., R. Dimova, and A. Saleh (2010), ‘Whither corruption? A quantitative
survey of the literature on corruption and growth’, CEPR Discussion Paper No.
8140.
Chattopadhyay, R., and E. Duflo (2004), ‘Women as policy makers: Evidence from
a randomized policy experiment in India’, Econometrica, 72(5): 1409–1443.
Chen, M., J. Vanek, F. Lund, J. Heintz, R. Jhabvala, and C. Bonner (2005).
Progress of the World’s Women 2005: Women, Work and Poverty. New York:
United Nations Development Fund for Women.
Clarke, G.R.G. (2011), ‘How petty is petty corruption? Evidence from firm surveys
in Africa’, World Development, 39(7): 1122–1132.
Commander, S., and J. Svejnar (2007), ‘Do institutions, ownership, exporting and
competition explain firm performance? Evidence from 26 transition countries’,
Institute for the Study of Labor (IZA) Discussion Paper No. 2637.
Croson, R., and U. Gneezy (2009), ‘Gender differences in preferences’, Journal of
Economic Literature, 47(2): 448–474.
LINK TEXT.indd 202 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 203
Davis, S.J., and J. Haltiwanger (1992), ‘Gross job creation, gross job destruction,
and employment reallocation’, Quarterly Journal of Economics, 107(3): 819–863.
Davis, S.J., and J. Haltiwanger (1999), ‘On the driving forces behind cyclical move-
ments in employment and job reallocation’, American Economic Review, 89(5):
1234–1258.
Dethier, J., M. Hirn, and S. Straub (2011), ‘Explaining enterprise performance in
developing countries with business climate survey data’, World Bank Research
Observer, 26(2): 258–309.
Dollar, D., R. Fisman, and R. Gatti (2001), ‘Are women really the “fairer” sex?
Corruption and women in government’, Journal of Economic Behavior and
Organization, 46: 423–429.
Dreher, A., and M. Gassebner (2013), ‘Greasing the wheels? The impact of regula-
tions and corruption on firm entry’, Public Choice, 155: 413–432.
Duflo, E. (2012), ‘Women’s empowerment and economic development’, Journal of
Economic Literature, 50(4): 1051–1079
Dunne, T., and A. Hughes (1994), ‘Age, size, growth, and survival: UK companies
in the 1980s’, Journal of Industrial Economics, 42: 115–140.
Escribano, A., and J.L. Guasch (200), ‘Assessing the impact of the business climate
on productivity using firm level data: Methodology and the cases of Guatemala,
Honduras, and Nicaragua’, World Bank Policy Research Working Paper No.
3621, The World Bank.
Estrin, S., and T. Mickiewicz (2011), ‘Institutions and female entrepreneurship’,
Small Business Economics, 37(4): 397–415.
European Bank for Reconstruction and Development (EBRD) (2005), Transition
Report 2005. London: European Bank for Reconstruction and Development.
Evans, D.S. (1987), ‘The relationship between firm growth, size, and age: Estimates
for 100 manufacturing industries’, Journal of Industrial Economics, 35: 567–581.
Ferreira, F., and J. Gyourko (2014), ‘Does gender matter for political leadership?
The case of U.S. mayors’, Journal of Public Economics, 112: 24–39.
Fisman, R., and J. Svensson (2007), ‘Are corruption and taxation really harmful
to growth? Firm level evidence’, Journal of Development Economics, 83: 63–75.
Fossen, F.M. (2012), ‘Gender differences in entrepreneurial choice and risk
aversion – a decomposition based on a microeconometric model’, Applied
Economics, 44(14): 1795–1812.
Francisco, M., and N. Pontara (2007), ‘Does corruption impact on firms’ ability to
conduct business in Mauritania? Evidence from investment climate survey data’,
World Bank Policy Research Paper No. 4439, The World Bank.
Frank, B., J.G. Lambsdorff, and F. Boehm (2011), ‘Gender and corruption: Lessons
from laboratory corruption experiments’, European Journal of Development
Research, 23(1): 59–71.
Freund, C., M. Hallward-Driemeier, and B. Rijkers (2014), ‘Deals and delays:
Firm-level evidence on corruption and policy implementation times’, World
Bank Policy Research Paper No. 6949, The World Bank.
Gaviria, A. (2002), ‘Assessing the effects of corruption and crime on firm perfor-
mance: Evidence from Latin America’, Emerging Markets Review, 3: 245–268.
Gelb, A., V. Ramachandran, M. Kedia-Shah, and G. Turner (2007). ‘What matters
to African firms? The relevance of perceptions data’, World Bank Policy
Research Working Paper No. 4446, The World Bank.
Goetz, A.M. (2007), ‘Political cleaners: Women as the new anti-corruption force?’,
Development and Change, 38(1): 87–105.
LINK TEXT.indd 203 24/02/2017 15:55

204 Gender and Entrepreneurial Activity
Gray, C.W., and D. Kaufmann (1998), ‘Corruption and development’, Finance and
Development, 35(1): 7–10.
Hall, B.H. (1987), ‘The relationship between firm size and firm growth in the US
manufacturing sector’, Journal of Industrial Economics, 35: 583–606.
Hallward-Driemeier, M., S. Wallsten, and L.C. Xu (2006), ‘Ownership, investment
climate and firm performance. Evidence from Chinese firms’, Economics of
Transition, 14(4): 629–647.
Hellman, J., G. Jones, and D. Kaufmann (2003), ‘Seize the state, seize the day:
State capture, corruption, and influence in transition’, Journal of Comparative
Economics, 31: 751–773.
Hellman, J., G. Jones, D. Kaufmann, and M. Schankerman (2000), ‘Measuring
governance and state capture: The role of bureaucrats and firms in shaping the
business environment’, EBRD Working Paper No. 51.
Honorati, M., and T. Mengistae (2007), ‘Corruption, the business environment,
and small business growth in India’, World Bank Policy Research Paper No.
4338, The World Bank.
Hunt, J. (2004), ‘Trust and bribery: The role of the quid pro quo and the link with
crime’, NBER Working Paper No. 10510.
Huntington, S.P. (1968). Political Order in Changing Societies. New Haven, CT:
Yale University Press.
Kaufmann, D., A. Kraay, and M. Mastruzzi (2005), ‘Governance matters IV:
Governance indicators for 1996–2004’, World Bank Policy Research Working
Paper No. 3630, The World Bank.
Klapper, L.F., and S. C. Parker (2011), ‘Gender and the business environment for
new firm creation’, World Bank Research Observer, 26(2): 237–257.
Kraay, A., and P. Murrell (2013), ‘Misunderestimating corruption’, World Bank
Policy Research Working Paper No. 6488, The World Bank.
Krueger, A.B., and J. Angrist (2001), ‘Instrumental variables and the search for
identification: From supply and demand to natural experiments?’, Journal of
Economic Perspectives, 15(4): 69–85.
Leff, N.H. (1964), ‘Economic development through bureaucratic corruption’,
American Behavioral Scientist, 8(3): 8–14.
Lileeva, A., and D. Trefler (2010), ‘Improved access to foreign markets raises plant-
level productivity . . . for some plants’, Quarterly Journal of Economics, 125(3):
1051–1099.
Mansfield, E. (1962), ‘Entry, Gibrat’s law, innovation, and the growth of firms’,
American Economic Review, 52: 1023–1051.
Mauro, P. (1995), ‘Corruption and growth’, Quarterly Journal of Economics, 110:
681–712.
Mead, D.C., and C. Liedholm (1998), ‘The dynamics of micro and small enter-
prises in developing countries’, World Development, 26(1): 61–74.
Méon, P.-G., and K. Sekkat (2005), ‘Does corruption grease or sand the wheels of
growth?’, Public Choice, 122(1): 69–97.
Méon, P.-G., And L. Weill (2010), ‘Is corruption an efficient grease?’, World
Development, 38(3): 244–259.
Michailova, J., and I. Melnykovska (2009), ‘Gender, corruption and sustainable
growth in transition countries’, MPRA Paper No. 17074.
Murayev, A., O. Talavera, and D. Shäfer (2009), ‘Entrepreneurs’ gender and
financial constraints: Evidence from international data’, Journal of Comparative
Economics, 37(2): 270–286.
LINK TEXT.indd 204 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 205
Murphy, K., A. Shleifer, and R. Vishny (1991), ‘The allocation of talent: Implications
for growth’, Quarterly Journal of Economics, 106: 503–530.
Nichter, S., and L. Goldmark (2009), ‘Small firm growth in developing countries’,
World Development, 37(9): 1453–1464.
Olken, B. (2009), ‘Corruption perceptions vs. corruption reality’, Journal of Public
Economics, 93(7–8): 950–964.
Olken, B.A., and R. Pande (2012), ‘Corruption in developing countries’, Annual
Review of Economics, Annual Reviews, 4(1): 479–509.
Pavcnik, N. (2002), ‘Trade liberalization, exit, and productivity improvement:
Evidence from Chilean plants’, Review of Economic Studies, 69(1): 245–276.
Rivas, M.F. (2013), ‘An experiment on corruption and gender’, Bulletin of
Economic Research, 65(1): 10–42.
Rose-Ackerman, S. (1975), ‘The economics of corruption’, Journal of Public
Economics, IV: 187–203.
Rose-Ackerman, S. (1996), The Political Economy of Corruption – Causes and
Consequences. Public Policies for the Private Sector, Note No. 74. Washington,
DC: The World Bank.
Sabarwal, S., and K. Terrell (2008), ‘Does gender matter for firm performance?
Evidence from Eastern Europe and Central Asia’, World Bank Policy Research
Working Paper No. 4705, The World Bank.
Seker, M., and J.S. Yang (2012), ‘How bribery distorts firm growth: Differences
by firm attributes’, World Bank Policy Research Working Paper No. 6046, The
World Bank.
Shleifer, A., and R.W. Vishny (1993), ‘Corruption’, Quarterly Journal of Economics,
108: 599–617.
Steves, F., and A. Rousso (2003), ‘Anti-corruption programmes in post-communist
transition countries and changes in the business environment, 1999–2002’,
European Bank for Reconstruction and Development, Working Paper No. 85.
Sung, H.-E. (2003), ‘Fairer sex or fairer system? Gender and corruption revisited’,
Social Forces, 82: 703–723.
Svennson, J. (2003), ‘Who must pay bribes and how much? Evidence from a cross
section of firms’, Quarterly Journal of Economics, 118(1): 207–230.
Svennson, J. (2005), ‘Eight questions about corruption’, Journal of Economic
Perspectives, 19(3): 19–42.
Swamy, A., S. Knack, Y. Lee, and O. Azfar (2001), ‘Gender and corruption’,
Journal of Development Economics, 64(1): 25–55.
Tanzi, V. (1998), ‘Corruption around the world: Causes, consequences, scope, and
cures’, IMF Staff Papers, 45(4): 559–594.
Tonoyan, V., R. Strohmeyer, M. Habib, and M. Perlitz (2010), ‘Corruption and
entrepreneurship: How formal and informal institutions shape firm behavior in
transition and mature market economies’, Entrepreneurship Theory and Practice,
34(5): 803–831.
Torgler, B., and N.T. Valev (2006), ‘Women and illegal activities: Gender differences
and women’s willingness to comply over time’, Working Paper No. 2006-15,
Center for Research in Economics, Management and Arts (CREMA).
Van Biesebroeck, J. (2005), ‘Exporting raises productivity in Sub-Saharan African
manufacturing firms’, Journal of International Economics, 67(2): 373–391.
van der Zwan, P., I. Verheul, and R. Thurik (2011), ‘The entrepreneurial ladder in
transition and non-transition economies’, Entrepreneurship Research Journal,
1(2). https://doi.org/10.2202/2157-5665.1012.
LINK TEXT.indd 205 24/02/2017 15:55

206 Gender and Entrepreneurial Activity
Verheul, I., and R. Thurik (2001), ‘Start-up capital: “Does gender matter?”’, Small
Business Economics, 16: 329–345.
Vial, V., and J. Hanoteau (2010), ‘Corruption, manufacturing plant growth, and
the Asian paradox: Indonesian evidence’, World Development, 38(5): 693–705.
Wang, Y., and J. You (2012), ‘Corruption and firm growth: Evidence from China’,
China Economic Review, 23(2): 415–433.
World Bank (2012), ‘Opportunities for men and women in Emerging Europe and
Central Asia’, Washington, DC: The World Bank.
LINK TEXT.indd 206 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 207
APPENDIX A
Table 8A.1 Descriptive statistics by group of countries
Country
group
Obs Female
owned
Administrative
corruption
Adm. Corr
% of sales
Gift to
obtain a
contract
Gift for
contract %
of contract
Old EU 2,244 30.5% 22.0% 1.64 29.5% 3.1%
NMS 1,763 30.9% 33.1% 2.74 27.8% 6.0%
SEE 535 16.6% 41.6% 2.62 38.2% 8.6%
CIS 929 23.7% 47.4% 3.20 29.6% 6.0%
Note: Old EU: Germany, Greece, Ireland, Portugal, Spain; New Member States (NMS):
Bulgaria, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania,
Slovakia, Slovenia; South-Eastern Europe (SEE): Albania, Bosnia and Herzegovina,
Croatia, Macedonia, Serbia and Montenegro, Turkey; Commonwealth of Independent
States (CIS): Armenia, Belarus, Kazakhstan, Kyrgyzstan, Moldova, Russia, Tajikistan,
Ukraine, Uzbekistan.
LINK TEXT.indd 207 24/02/2017 15:55

208
Table 8A.2 Definitions of variables and their descriptive statistics
Variable Definition  sN
Femaleown 1 if the manager/owner is female, else 0 0.28 0.45 5,471
Adm. Corr.
(incidence)
1 if firm reported a positive amount of payment to the question: “On average,
what percent of total annual sales do firm’s like yours typically pay in
unofficial payments/gifts to public officials?” Else 0
0.32 0.47 4,966
Adm. Corr.
(% of Sales)
Average amount paid (In % of sales) for firm that declares bribing behavior in
the question above
2.52 3.17 1,576
Gift for contract
(incidence)
1 if firm reported a positive amount of payment to the question: ‘When firms
in your Industry do business with the government, what percent of the
contract value would be typically paid in additional or unofficial
payments/gifts to secure the contract?’ Else 0
0.30 0.46 4,175
Gift for contract
(% of contract)
Average contract’s percentage paid for firm that declares bribing behavior in the
question above
5.36 4.61 1,240
Empl. growth ((Employment − Employment_2002)/((Employment + Employment_2002)/2))/3) 0.03 0.14 5,471
micro entr 1 if 1–10 permanent employees in 2002, else 0 0.57 0.50 5,471
small entr 1 if 11–49 permanent employees in 2002, else 0 0.27 0.44 5,471
medium entr 1 if 50–199 permanent employees in 2002, else 0 0.11 0.32 5,471
large entr 1 if 200 and more permanent employees in 2002, else 0 0.05 0.22 5,471
FormalSect First principal component of the follwing variables: trade (exports + Imports),
no. of inspections, bad infrastructure (no. of services suffering Interruptions:
water, electricity, telephone, emails)
1.55E–09 1.09 5,471
export 1 if firm exports its products abroad, else 0 0.20 0.40 5,451
LINK TEXT.indd 208 24/02/2017 15:55

209
import 1 if firm imports, else 0 0.25 0.43 5,471
electricity 1 if firm suffered electricity interruptions, else 0 0.41 0.49 5,471
water 1 if firm suffered water interruptions, else 0 0.14 0.35 5,471
telephone 1 if firm suffered telephone interruptions, else 0 0.20 0.40 5,471
email 1 if firm does not use email regularly, else 0 0.31 0.46 5,471
No. Inspections Number of times in the past 12 months the establishment either inspected or
required to meet with officials from these agencies: Tax inspectorate, Labor
and social security, Fire & building safety,
Sanitation/ Epidemiology, Municipal police, Environmental, Customs agency
5.60 13.41 5,471
Property 1 if firms answered 4 to 7 to the question: “I am confident that the legal system
will uphold my contract and property rights in business disputes: strongly
disagree (1), disagree in most cases (2), tend to disagree (3), tend to agree (4),
agree in most cases (5), strongly agree (7)”. Else 0
0.65 0.48 5,471
Fair_bureauc 1 if firms answered 4 to 6 to the question: “If a government agent acts against
the rules I can usually go to another official or to his superior and get the
correct treatment without recourse to unofficial payments/gifts” (“never” (1),
“seldom” (2), “sometimes” (3), “frequently” (4), “usually” (5), “always” (6)).
Else 0
0.48 0.50 5,471
Profits2003 1 if firm was profitable in 2003, else 0 0.90 0.30 5,471
InK natural logarithm of the capital stock 5.06 1.85 5,471
concentration % of sales from the main business activity 95.92 10.10 5,471
memberBA 1 if the firm is a member of business association or chamber of commerce, else 0 0.54 0.50 5,471
privatization 1 if the firm was created by a privatization of a firm-owned entreprise, else 0 0.08 0.27 5,471
foreign_own 1 if the foreigners own 10 % or more of the firm, else 0 0.05 0.23 5,471
ageCompany Age of the firm 15.85 15.43 5,471
ageComp2 Age squared divided by 100 4.89 16.78 5,471
small city/rural 1 if firm is in city of fewer than 250,000 inhabitants or in rural area, else 0 0.57 0.49 5,471
LINK TEXT.indd 209 24/02/2017 15:55

210
Table 8A.2 (continued)
Variable Definition  sN
urban 1 if firm is in city of 250,000 or more Inhabitants, else 0 0.43 0.49 5,471
manufacturing 1 if firm’s sector is ISIC 15–37, else 0 0.34 0.47 5,471
trade and transport 1 if firm’s sector is ISIC 50–52 or 60–64, else 0 0.31 0.46 5,471
services 1 if firm’s sector is ISIC 45, 55, 70–74, 93, else 0 0.35 0.48 5,471
%Workforce % of workforce reported in the question: “What percentage of total workforce
would you estimate the typical firm in your area of business reports for tax
purposes?”
90.82 16.83 5,332
timeReg Percentage of senior management’s time spent over the last 12 months in
dealing with public officials about the application and interpretation of laws
and regulations and to get or to maintain access to public services
4.22 7.48 5,160
growth assets % change of Axed assets in the last 36 months (divided by 3) 0.05 0.15 5,471
growth exports % change of exports in the last 36 months (divided by 3) 0.01 0.09 5,471
Note: m stands for the mean, s for the standard deviation, and N for the number of observations.
LINK TEXT.indd 210 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 211
Table 8A.3 Determinants of the amount paid, selection model
Adm. Corr
Heckman
(1)
Gift for contract
Heckman
(2)
Femaleown −0.043
[0.233]
−0.089
[0.406]
small entr −0.149
[0.169]
0.26
[0.338]
medium entr −0.639
[0.265]**
−0.036
[0.478]
large entr −0.682
[0.348]*
−0.686
[0.871]
Profits2003 0.14
[0.202]
−0.403
[0.306]
InK −0.146
[0.059]**
−0.177
[0.098]*
concentration −0.001
[0.007]
−0.005
[0.013]
FormalSect 0.05
[0.081]
0.213
[0.068]***
ageCompany −0.005
[0.011]
−0.015
[0.021]
ageComp2 0.004
[0.008]
0.017
[0.016]
memberBA −0.269
[0.170]
−0.144
[0.260]
privatization 0.027
[0.310]
−0.235
[0.622]
foreign_own −0.218
[0.301]
0.243
[0.433]
Property −0.538
[0.139]***
−0.733
[0.198]***
Fair_bureauc −0.323
[0.146]**
−0.449
[0.319]
small city/rural −0.149
[0.158]
−0.411
[0.257]
trade and transport −0.417
[0.268]
−0.561
[0.324]*
services −0.016
[0.243]
0.382
[0.203]*
LINK TEXT.indd 211 24/02/2017 15:55

212 Gender and Entrepreneurial Activity
Table 8A.3 (continued)
Adm. Corr
Heckman
(1)
Gift for contract
Heckman
(2)
Selection instrument
%Workforce −0.017
[0.003]***
−0.014
[0.003]***
p −0.16
[0.088]*
−0.277
[0.108]***
Observations 5,332 5,332
Notes: Columns 1 and 2 show regressions adjusted for selection bias with Heckman
correction. The regressions include country and industry dummies. Asymptotic cluster-
robust standard errors (clustering by country) are reported in brackets. *significant at 10%,
**significant at 5%, ***significant at 1%.
LINK TEXT.indd 212 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 213
Table 8A.4 Robustness checks sample split
PANEL A: sample split age
Administrative corruption Gifts to obtain
contract
Young Old Young Old
Female owned −0.036
[0.024]
−0.038
[0.018]**
−0.065
[0.026]**
−0.079
[0.012]***
Observations 1,511 3,303 1,193 2,508
PANEL B: sample split network
Ye s No Yes No
Female owned −0.034
[0.022]
−0.038
[0.024]*
−0.077
[0.018]***
−0.07
[0.017]***
Observations 2,705 2,109 1,806 1,891
Notes: The models are identical to those reported in columns 1, and 2 of Table 8.3. Panel
A reports the results for the sample split based on average age, Panel B for the sample split
based on membership in a business association (yes versus no). The regressions include
industry and country dummies. Asymptotic cluster-robust standard errors (clustering by
country) are reported in brackets. *significant at 10%, **significant at 5%, ***significant at
1%.
LINK TEXT.indd 213 24/02/2017 15:55

214
Table 8A.5 Correlations table
Empl.
Growth
Adm.
Corr.
Gift4contr Female
owned
size2002 InK age privatization foreign_own timeReg
Empl. growth 1
Adm. Corr. 0.0089 1
Gift4contr −0.009 0.5798*** 1
Femaleown −0.0361*** −0.0674*** −0.1151*** 1
size2002 −0.1045*** 0.0978*** 0.0344** −0.1109*** 1
InK 0.0308** 0.0015 0.0441*** −0.1179*** 0.5707*** 1
age −0.1328*** −0.0602*** −0.0289* −0.0222 0.2895*** 0.2990*** 1
privatization −0.0579*** 0.0420*** −0.0365** −0.0179 0.2808*** 0.0984*** 0.2190*** 1
foreign_own 0.0348** 0.0209 0.0108 −0.0644*** 0.1233*** 0.0787*** −0.0219 0.0305** 1
timeReg 0.0212 0.1400*** 0.1382*** −0.0121 0.0787*** 0.0375*** 0.0165 0.0561*** 0.0372*** 1
Note: *significant at 10%, **significant at 5%, ***significant at 1%.
LINK TEXT.indd 214 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 215
Table 8A.6 Total employment growth IV regressions using sector–
country averages and time spent in dealing with regulations
as instruments and instrumenting capital and trade status.
Dependent variable yearly employment growth rate.
(1) (2) (3) (4)
Admin. Corr. (dummy) −0.04
[0.012]***
FemXAdmCorr (dummy) 0.066
[0.031]**
Admin. Corr (amount) −0.01
[0.003]***
FemXAdmCorr (amount) 0.03
[0.011]***
Gift for contract (dummy) −0.052
[0.017]***
FemXGift4Cont (dummy) 0.133
[0.061]**
Gift for contract (amount) −0.008
[0.002]***
FemXGift4Cont (amount) 0.023
[0.007]***
InK 0.138
[0.025]***
0.136
[0.025]***
0.15
[0.028]***
0.147
[0.027]***
trade 0.143
[0.050]***
0.139
[0.049]***
0.114
[0.047]**
0.121
[0.045]***
Femaleown 0.008
[0.009]
0.005
[0.009]
−0.007
[0.013]
−0.005
[0.009]
small entr −0.192
[0.026]***
−0.192
[0.026]***
−0.208
[0.031]***
−0.203
[0.031]***
medium entr −0.386
[0.051]***
−0.382
[0.050]***
−0.416
[0.057]***
−0.409
[0.055]***
large entr −0.573
[0.076]***
−0.567
[0.075]***
−0.63
[0.089]***
−0.621
[0.087]***
foreign_own −0.039
[0.014]***
−0.037
[0.015]**
−0.018
[0.022]
−0.02
[0.022]
privatization −0.029
[0.017]*
−0.029
[0.017]*
−0.024
[0.018]
−0.023
[0.018]
ageCompany −0.004
[0.001]***
−0.003
[0.001]***
−0.003
[0.000]***
−0.004
[0.001]***
ageComp2 0.002
[0.000]***
0.002
[0.000]***
0.002
[0.000]***
0.002
[0.000]***
J-Test
p-value
2.753
[0.252]
3.612
[0.164]
1.926
[0.382]
2.665
[0.264]
LINK TEXT.indd 215 24/02/2017 15:55

216 Gender and Entrepreneurial Activity
Table 8A.6 (continued)
(1) (2) (3) (4)
First stage F-test
corruption 15.79 24.59 18.28 32.73
corr.Xfemale 87.24 7.32 14.88 35.14
InK 11.73 13.85 9.11 9.67
trade 6.34 6.41 5.34 5.36
endogeneity test 9.235 6.709 8.094 7.844
p-value [0.01] [0.035] [0.017] [0.019]
Observations 4,713 4,713 4,056 4,056
Notes: The regressions include sector–location–country interaction dummies. Asymptotic
cluster-robust standard errors (clustering by country) are reported in brackets. The
dependent variable is yearly total employment growth. In each column we analyze the
impact of a different form of corruption: (1) incidence of administrative corruption, (2)
amount paid for administrative corruption, (3) incidence of gifts to obtain government
contracts, (4) amount paid to obtain a contract.
*significant at 10%, **significant at 5%, ***significant at 1%.
F-test on instruments is the test statistic on the F-test of the joint significance of the
instruments (3-sector–country averages, timeReg, changeA, changeE) in the first-stage
regressions, with p-values in brackets. Hansen J-statistic is not shown as the equation is just
identified. The endogeneity test is defined as the difference of two Sargan-Hansen statistics:
one for the equation with the smaller set of instruments, where capital and trade (lnK, trade)
are treated as endogenous, and one for the equation with the larger set of instruments,
where capital and trade are treated as exogenous. The endogeneity test statistic is robust to
heteroskedasticity.
LINK TEXT.indd 216 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 217
Table 8A.7 Total employment growth IV regressions for only SMEs
using sector–country averages and time spent in dealing
with regulations as instruments. Dependent variable annual
employment growth rate.
(1) (2) (3) (4)
Adm. Corn (dummy) −0.045
[0.014]***
FemXAdmCorr (dummy) 0.085
[0.038]**
Adm. Corn (amount) −0.01
[0.003]***
FemXAdmCorr (amount) 0.033
[0.012]***
Gift for contract (dummy) −0.058
[0.019]***
FemXGift4Cont (dummy) 0.148
[0.057]***
Gift for contract (amount) −0.008
[0.002]***
FemXGift4Cont (amount) 0.022
[0.007]***
InK 0.158
[0.031]***
0.156
[0.030]***
0.163
[0.031]***
0.16
[0.030]***
trade 0.124
[0.047]***
0.126
[0.045]***
0.11
[0.052]**
0.114
[0.048]**
Femaleown 0.008
[0.013]
0.008
[0.014]
−0.01
[0.014]
−0.002
[0.011]
small entr −0.215
[0.034]***
−0.216
[0.032]***
−0.221
[0.035]***
−0.216
[0.034]***
medium entr −0.424
[0.064]***
−0.423
[0.061]***
−0.443
[0.066]***
−0.436
[0.063]***
large entr
foreign_own −0.043
[0.016]***
−0.044
[0.016]***
−0.037
[0.021]*
−0.037
[0.020]*
privatization −0.03
[0.016]*
−0.03
[0.016]*
−0.033
[0.019]*
−0.034
[0.019]*
ageCompany −0.004
[0.001]***
−0.004
[0.001]***
−0.004
[0.001]***
−0.004
[0.001]***
ageComp2 0.002
[0.001]***
0.002
[0.001]***
0.002
[0.000]***
0.002
[0.001]***
First stage F-test
corruption 25.7 33.54 28.91 48.47
corr.Xfemale 129.85 8.5 13.89 32.01
InK 12.48 13.82 14.94 14.98
trade 7.69 8.11 6.5 7.23
endogeneity test 17.24 17.24 16.44 16.57
p-value [0.000] [0.000] [0.000] [0.000]
Observations 4,709 4,709 3,961 3,961
LINK TEXT.indd 217 24/02/2017 15:55

218 Gender and Entrepreneurial Activity
Table 8A.7 (continued)
Notes: The regressions include sector–location–country interaction dummies. Asymptotic
cluster-robust standard errors (clustering by country) are reported in brackets. The
dependent variable is yearly total employment growth. In each column we analyze the
impact of a different form of corruption: (1) incidence of administrative corruption, (2)
amount paid for administrative corruption, (3) incidence of gifts to obtain government
contracts, (4) amount paid to obtain a contract.
*significant at 10%, **significant at 5%, ***significant at 1%.
F-test on instruments is the test statistic on the F-test of the joint significance of the
instruments (3-sector–country averages, timeReg, changeA, changeE) in the first-stage
regressions, with p-values in brackets. Hansen J-statistic is not shown as the equation is just
identified. The endogeneity test is defined as the difference of two Sargan-Hansen statistics:
one for the equation with the smaller set of instruments, where capital and trade (lnK, trade)
are treated as endogenous, and one for the equation with the larger set of instruments,
where capital and trade are treated as exogenous. The endogeneity test statistic is robust to
heteroskedasticity.
LINK TEXT.indd 218 24/02/2017 15:55

Corruption and entrepreneurship: does gender matter? 219
Table 8A.8 Total employment growth IV regressions excluding old EU
countries, using country – sector averages and time spent in
dealing with regulations as instruments. Dependent variable
annual employment growth rate.
(1) (2) (3) (4)
Adm. Corr (dummy) −0.036
[0.017]**
FemXAdmCorr (dummy) 0.075
[0.058]
Adm. Corr (amount) −0.012
[0.004]***
FemXAdmCorr (amount) 0.046
[0.013]***
Gift for contract (dummy) −0.084
[0.028]***
FemXGift4Cont (dummy) 0.275
[0.105]***
Gift for contract (amount) −0.009
[0.003]***
FemXGift4Cont (amount) 0.023
[0.008]***
InK 0.153
[0.037]***
0.152
[0.034]***
0.151
[0.035]***
0.155
[0.036]***
trade 0.129
[0.055]**
0.129
[0.052]**
0.122
[0.056]**
0.113
[0.052]**
Female owned 0.009
[0.022]
−0.011
[0.015]
−0.034
[0.030]
−0.003
[0.017]
small entr −0.2
[0.036]***
−0.204
[0.034]***
−0.199
[0.037]***
−0.199
[0.038]***
medium entr −0.398
[0.071]***
−0.399
[0.066]***
−0.405
[0.069]***
−0.411
[0.071]***
large entr −0.626
[0.109]***
−0.624
[0.102]***
−0.626
[0.106]***
−0.629
[0.108]***
foreign_own −0.052
[0.015]***
−0.052
[0.016]***
−0.045
[0.020]**
−0.043
[0.020]**
privatization −0.038
[0.016]**
−0.037
[0.016]**
−0.032
[0.017]*
−0.037
[0.017]**
ageCompany −0.004
[0.001]***
−0.004
[0.001]***
−0.004
[0.001]***
−0.004
[0.001]***
ageComp2 0.002
[0.001]***
0.002
[0.001]***
0.002
[0.001]***
0.002
[0.001]***
First stage F-test
corruption 29.71 24.94 28.12 52.69
corr. Xfemale 66.76 7.68 15.77 38.98
LINK TEXT.indd 219 24/02/2017 15:55

220 Gender and Entrepreneurial Activity
Table 8A.8 (continued)
(1) (2) (3) (4)
InK 9.62 12.34 11.65 11.62
trade 7.76 7.79 6.96 7.53
endogeneity test 14.58 14.48 14.34 14.77
p-value [0.000] [0.000] [0.000] [0.000]
Observations 2,883 2,883 2,788 2,788
Notes: The regressions include sector–location–country interaction dummies. Asymptotic
cluster-robust standard errors (clustering by country) are reported in brackets. The
dependent variable is yearly total employment growth. In each column we analyze the
impact of a different form of corruption: (1) incidence of administrative corruption, (2)
amount paid for administrative corruption, (3) incidence of gifts to obtain government
contracts, (4) amount paid to obtain a contract.
*significant at 10%, **significant at 5%, ***significant at 1%.
F-test on instruments is the test statistic on the F-test of the joint significance of the
instruments (3-sector -country averages, timeReg, changeA, changeE) in the first-stage
regressions, with p-values in brackets. Hansen J-statistic is not shown as the equation is just
identified. The endogeneity test is defined as the difference of two Sargan-Hansen statistics:
one for the equation with the smaller set of instruments, where capital and trade (lnK, trade)
are treated as endogenous, and one for the equation with the larger set of instruments,
where capital and trade are treated as exogenous. The endogeneity test statistic is robust to
heteroskedasticity.
LINK TEXT.indd 220 24/02/2017 15:55