See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/5158278

Perceptual Variables and Nascent Entrepreneurship

Article  in  Small Business Economics · April 2005

DOI: 10.1007/s11187-005-1984-x · Source: RePEc

CITATIONS READS

1,060 5,060

2 authors:

Pia Arenius Maria Minniti

University of Turku Southern Methodist University
25 PUBLICATIONS   3,275 CITATIONS    18 PUBLICATIONS   2,774 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Maria Minniti on 03 June 2014.

The user has requested enhancement of the downloaded file.

Small Business Economics (2005) 24: 233–247
Springer 2005
DOI 10.1007/s11187-005-1984-x

Perceptual Variables and

Pia Arenius
Nascent Entrepreneurship Maria Minniti

ABSTRACT. Using a large sample of individuals 1. Introduction

in 28 countries, we investigate what variables are
significantly correlated with an individual’s deci- In recent years, most of the economics literature
sion to become an entrepreneur. Following exist- related to entrepreneurship has focused on
ing literature in economics, we link such a employment choices and on the alternative moti-
decision to demographic and economic character- vations that cause some individuals to become
istics. In addition, we argue that perceptual vari- entrepreneurs. In this type of studies, an individ-
ables such as alertness to opportunities, fear of ual’s decision to become an entrepreneur is the
failure, and confidence about one’s own skills are result of a maximization process in which the
also important. Our results suggest that percep- individual in question compares the returns from
tual variables are significantly correlated with alternative income producing activities and
new business creation across all countries in our selects the employment opportunity with the
sample and across gender. Although our data do highest expected return (Kihlstrom and Laffont
not allow the identification of causal relation- 1979; Leazar, 2002; Minniti, 2004).
ships, our findings suggest that, when making Empirical studies in economics have also
decisions, nascent entrepreneurs rely significantly emphasized entrepreneurship as an employment
on subjective and often biased perceptions rather choice. As a result, they have established demo-
than on objective expectations of success. Thus, graphic and economic factors such as education,
perceptual variables should be included in eco- age, wealth, and work situation to be important
nomic models of entrepreneurial behavior. drivers of entrepreneurial behavior. We comple-
ment this approach by incorporating additional
KEYWORDS: Entrepreneurship, nascent entrepreneur- variables whose importance for entrepreneurial
ship, perceptual variables. behavior has been established in the psychology
and sociology literature. In other words, we
JEL CODE: M13, J24, D01
argue that, in addition to the variables described
above, when making decisions with respect to
their employment, individuals also consider a set
of subjective perceptions about entrepreneurship
that they form based on the presence of role
models, confidence in one’s skills and ability, risk
Final version accepted on February 9, 2005
propensity, and alertness to unexploited opportu-
Pia Arenius nities. After all, entrepreneurship is about people.
HEC Lausanne, BFSH-1 Thus, we add a set of variables describing per-
CH-1015 Lausanne-Dorigny sonal perceptions and judgements about the envi-
Switzerland
E-mail: pia.arenius@unil.ch
ronment that, although often biased, are
nevertheless highly correlated with an individual’s
Maria Minniti decision to start a new business. We call these
Economics Division factors perceptual variables.
Babson College, Luksic Hall The complementarities between many of the
Babson Park, MA 02457
U.S.A.
demo-economic variables and the perceptual vari-
E-mail: minniti@babson.edu ables draw attention to the fact that the decision
234 Pia Arenius and Maria Minniti

to start a new firm is a complex multi-layered incentives with the insight that subjective percep-
process contingent, to a large extent, upon the tions also influence their behavior (Cooper et al.,
context in which the actions are taken. Thus, in 1988; Busenitz and Barney, 1997; Koellinger et al.
addition to personal characteristics, we include 2004) Our paper contributes to the elimination of
the individual’s environment as a further compo- this gap in the literature by providing a more
nent of entrepreneurial behavior and consider the accurate analysis of what variables are significant
possibility of country specific effects. in the make up of nascent entrepreneurs, and by
Our results suggest that perceptual variables suggesting that perceptual variables matter and
have an impact on new business creation across should be included in analytical models of entre-
all countries in our sample. In fact, our analysis preneurial behavior.
suggests that subjective perceptions about one’s
own skills, likelihood of failure, existence of
2. Theoretical background
opportunities, and knowledge of other entrepre-
neurs, are all highly correlated to the decision to Table I provides a simple classification of several
start a new business. Although our data do not representative works on entrepreneurial motiva-
allow us to clearly establish the causal direction tion according to the variables included in the
of the relationship, they do suggest that entrepre- study.1
neurs tend to rely relatively more on subjective Our study of nascent entrepreneurship includes
perceptions than on objective expectations. demographic and economic characteristics of the
Our data are very well suited for the study of individual, perceptual variables based on subjective
nascent entrepreneurs because they provide infor- judgements of the individuals, and aggregate
mation about individuals who are in the process variables summarizing the environment in which
of starting a new business at that particular point individuals make decisions. In the following section
in time and are not, like most other data sets, the we review existing literature for each of these three
results of the ex post evaluations of past decisions groups of factors.
of successful entrepreneurs. Also, because of the
standardization of the collection process, our data
2.1. Demographic and economic characteristics
allow us to establish what factors are correlated
with entrepreneurial behavior regardless of mac- Entrepreneurship tends to be a young man’s
roeconomic conditions and to what extent coun- game. Levesque and Minniti (in press) have
try effects are significant. To our knowledge, very shown theoretically that the relationship between
few empirical studies have combined the insight age and the likelihood of starting a new business
that potential entrepreneurs react to economic picks at a relatively early age and decreases

TABLE I
Representative works grouped by variables included in the study

Demographic Aggregate Perceptual Demo-economic Demo-economic Demo-economic,

and economic variables variables and aggregate and perceptual aggregate and
variables variables variables perceptual
variables

Blanchflower Acs et al. Busenitz and Acs and Davidsson and Minniti and
(2004) (1999) Barney (1997) Audretsch (1993) Honig (2003) Arenius (2005)
Blanchflower and Carree et al. Cooper et al. Thurik et al. Delmar and
Oswald (1998) (2000) (1988) (2002) Davidsson (2000)
Chell and Weber and
Baines (2000) Milliman (1997)
Taylor (1996) Wennekers and
Thurik (1999)
 Perceptual Variables and Nascent Entrepreneurship 235

thereafter. Consistently, Reynolds et al. (2003) entrepreneurial decisions. In fact, it is not clear
have found empirical evidence showing individu- whether high unemployment discourages entre-
als between 25 and 34 years of age to be the preneurship by reducing its potential markets or
most likely to be nascent entrepreneurs. Also, increases it by providing an income producing
Blanchflower (2004) suggests that, although the activity for otherwise displaced workers (Blanch-
probability of being an entrepreneur is highest flower, 2004).
among older individuals, the likelihood of being Finally, economists have discussed the impor-
a nascent entrepreneur is maximized among tance of financial resources and constraints on
young individuals. entrepreneurial decisions (Kihlstrom and Laffont,
In recent years, gender differences in entrepre- 1979; Evans and Jovanovic, 1989). In their work,
neurial behavior have been also the subject of a entrepreneurial decisions are shown to be posi-
significant amount of attention. Brush (1990, tively related to individuals’ household incomes.
1992) observed that men and women entrepre- Overall, the evidence suggests that entrepreneurs
neurs differ very little with respect to demo- face liquidity constraints and that individuals
graphic and psychological variables, while more with greater family wealth are more likely to
pronounced differences seem to exist in business switch from employment to entrepreneurship.
goals and management styles. Similarly, Lango-
witz and Minniti (2005) found that the factors
2.2. Perceptual variables
influencing female and male entrepreneurship
tend to be the same. In spite of these similarities, We complement the group of variables above, by
women participation rates in entrepreneurship adding four perceptual variables. That is, factors
are systematically below those of men. Greene describing subjective perceptions and beliefs of
(2000) provides a possible answer to this puzzle the individual but not reflecting necessarily objec-
by suggesting the existence of differences in aver- tive circumstances. Unlike demographic and eco-
age human and social capital. nomic characteristics, perceptual variables and
In the theoretical literature in economics, edu- their impact on entrepreneurship have received
cation is often treated as a proxy for human capi- less attention from economists. This is in part
tal and an engine of innovation. Entrepreneurs, due to the paucity of available data, and in part
however, are often modeled as possessing a to the difficulty of embedding such variables into
broad range of talents but no advanced educa- standard neoclassical models. Nevertheless, the
tion in any specific area (Murphy et al., 1991; emphasis on perceptions and beliefs is not new in
Leazar, 2002). Consistently with the theory, the economics and an increasing number of scholars
relationship between education and new firm for- agree that opportunity perception is the most dis-
mation is uncertain, except for richer countries tinctive and fundamental characteristic of entre-
where post graduate training has been shown to preneurial behavior (Kirzner, 1973, 1979).
have positive effects on high-tech start-up rates The psychology literature has established the
(Blanchflower, 2004). The uncertainty surround- importance of confidence in one’s skills and ability
ing education is due in part to the fact that for entrepreneurial behavior. The decision to
education levels have primarily a contextual sig- start a new business has been related to the pres-
nificance and in part to the variety of meaning ence of intentionality and locus of control
attributed to the word entrepreneurship.2 (Baron, 2000). Starting a new firm is an inten-
Since in economics, entrepreneurship is often tional act that involves repeated attempts to exer-
treated as a work status choice, a sizable amount cise control over the process in order to achieve
of research exists on the relationship between an the desired outcome (Gartner, 1985). In econom-
individual’s decisions to become an entrepreneur ics, Harper (1998) argues that an internal locus
and the availability of employment options.3 In of control increases the link between entrepre-
general, there is agreement that employed indi- neurial alertness and self-efficacy and, therefore,
viduals are more likely to start new businesses. leads to the creation of more new firms.
Less agreement, instead, exists about the rela- The relationship between entrepreneurship and
tionship between aggregate unemployment and fear of failure has received some attention from
236 Pia Arenius and Maria Minniti

economists who have considered the relationship 1993). Thus, in addition to economic, demo-
between entrepreneurial decisions and risk aver- graphic and perceptual factors, we consider the
sion (Kihlstrom and Laffont, 1979). The intuition importance of the macroeconomic environment
is that since most individuals are risk averse and on entrepreneurial decisions by introducing the
since the perceived (rather than objective) fear of possibility of country effects.
failure is an important component of the risk
attached to starting a new business, a reduced
3. Data and Method
perception of the likelihood of failure should
increase the probability that an individual will Data used in the paper are from the Global
start a new business (Weber and Milliman, 1997). Entrepreneurship Monitor (GEM) project. Using
Finally, works in various social sciences have population samples, the GEM project estimates
established the importance of knowing other the prevalence rates of nascent and new busi-
entrepreneurs for entrepreneurial decisions. In nesses across several countries.5 The data used in
psychology, for example, Baron (2000) and Beg- this paper were collected during spring 2002.6 In
ley and Boyd (1987) have discussed the impor- each country, a standardized survey was adminis-
tance of role models because of their ability tered to a representative sample of adults (18–
to enhance self-efficacy. In economics, Minniti 64 years old) yielding a cross-country total of
(2004) has discussed increases in individuals’ con- 80,117 individuals.7 The sample size varied from
fidence generated by the presence of role models 972 individuals in Mexico to 12,837 in the UK.
and their ability to reduce ambiguity. Similarly, Because of individual-level missing data, 51,721
in sociology, Aldrich (1999), among others, has respondents were included in our final sample.8
discussed the role of personal networks and their In order to draw conclusions generalizable to the
ability to enhance entrepreneurial confidence by populations, raw data were given appropriate
providing advice, support and examples.4 weights to match each country sample with the
age and gender structure of the 2002 U.S. Census
International Database.9
2.3. Aggregate factors
To identify individuals in the process of start-
It has been shown that entrepreneurial decisions ing a business, respondents were asked: ‘‘You
differ also because of the environment in which are, alone or with others, currently trying to start
they are taken (Chell and Baines, 2000). Jack a new business, including any self-employment or
and Anderson (2002), for example, argue that selling any goods or services to others?’’ Respon-
new firm creation is not merely an economic pro- dents who answered ‘‘yes’’ were asked two addi-
cess but it is embedded in a specific environment. tional questions. These questions were used to
Thurik et al. (2002) provide a detailed analysis of separate those who were truly committed to a
aggregate conditions influencing nascent entre- new venture from those thinking about but not
preneurship and argue that technology, level of yet committed to it. These questions inquired (1)
economic development, culture, and institutions ‘‘Over the past twelve months have you done
all influence the demand for entrepreneurship by anything to help start a new business, such as
creating opportunities available for start-ups. In looking for equipment or a location, organizing
addition, they suggest that cultural and institu- a start-up team, working on a business plan,
tional conditions have an impact also on the sup- beginning to save money, or any other activity
ply of entrepreneurship because of their ability to that would help launch a business?’’ and (2)
influence the skills, resources and preferences of ‘‘Will you personally own all, part, or none
individuals within the population. Consistently of this business?’’ Only those respondents, who
with their analysis, empirical studies suggests that answered ‘‘yes’’ to the first question and ‘‘all’’ or
significant differences exist in the levels of new ‘‘part’’ to the second question, were coded as
firm creation across countries and over time nascent entrepreneurs. A total of 3,625 nascent
(Reynolds et al., 2003; Blanchflower, 2004) and entrepreneurs were identified in the sample. The
that country effects may be quite important for nascent entrepreneur prevalence rate averaged
entrepreneurial decisions (Acs and Audretsch, 4.5 across the 28 countries in our sample, ranging
 Perceptual Variables and Nascent Entrepreneurship 237

from a low of 1.0% in Japan to a high of 14.9% Confidence in one’s skills: Respondents were
in Mexico.10 asked whether they believed to have the knowl-
Our study consists in presenting a set of alter- edge, skill and experience required to start a new
native models estimating the likelihood of an business.
individual being a nascent entrepreneur given the Fear of failure: The respondents were asked
following independent variables: whether fear of failure would prevent them from
Age: The respondents were asked to provide starting a business.
their year of birth. For each of the four perceptual variables,
Gender: The respondents were asked to pro- responses were coded as binary variables with 2
vide their gender. indicating a yes response and 1 indicating a no
Education: The respondents were asked to response.
provide the highest degree they had earned. A correlation matrix for the variables was
Their responses were harmonized across all 28 computed and is presented in Table II. Eight of
countries into a five-category variable. The five the nine predictor variables have a strong corre-
categories are ‘‘No education’’, ‘‘Some second- lation (p < 0.001) with our nascent entrepreneur
ary education’’, ‘‘Secondary degree’’, ‘‘Post sec- measure. Table II shows that the likelihood of
ondary education’’, and ‘‘Graduate degree’’. In being a nascent entrepreneur decreases with age
the logistic regression analysis, the ‘‘Secondary (r ¼ )0.051). Also male respondents are shown
degree’’ category is used as the reference cate- to be more likely to be nascent entrepreneurs
gory. That is, we compared if those with levels than female respondents (r ¼ )0.069). In addition
of education others than secondary degrees working respondents are more likely to qualify as
were more or less likely to report trying to nascent entrepreneurs than their non-working,
start a business than those holding a secondary student, or retired counterparts (r ¼ 0.045). As
degree. expected, opportunity perception (r ¼ 0.165),
Work status: The respondents were asked to confidence in one’s skills (r ¼ 0.204), and know-
provide their occupational status. The response ing other entrepreneurs (r ¼ 0.157) are positively
categories were ‘‘Full or part time work’’, ‘‘Not related to being a nascent entrepreneur, whereas
working’’, and ‘‘Retired or student’’. In the logis- fear of failure (r ¼ )0.059) is negatively corre-
tic regression analysis, ‘‘Full or part time work’’ lated with our nascent entrepreneur measure.
is used as the reference category. That is, we Table II also alerts us to the presence of collin-
compared if those working full or part time were earity among the independent variables.
more or less likely to report trying to start a Since our dependent variable is a binary vari-
business than those reporting other situations as able, we analyze our data by employing binomi-
their work status. nal logistic regression models. The binominal
Household income: Respondents were asked to logistic regression estimates the probability of an
provide information about their household event happening. In our case, the event is being a
income. Data were classified into three groups, nascent entrepreneur. Maximum likelihood esti-
namely in the upper, middle or lower third of the mations were used to calculate the logit coeffi-
income distribution of the country of origin. In cients which denote changes in the log odds of
the logistic regression analysis, the middle income the dependent variable. We assessed the goodness
group is used as the reference category. of fit of the models using the Pearson Chi-square
Opportunity perception: Respondents were test, the rate of correct classifications and the
asked whether they thought that good opportuni- pseudo R2. The significance of individual inde-
ties for starting a business would exist in the area pendent variables was tested using the Wald sta-
where they lived in the six months following the tistics.
survey. We estimated five binominal logistic regression
Knowing other entrepreneurs: Respondents models. In each of them, the dependent variable
were asked whether they knew personally some- is whether the respondent is a nascent entrepre-
one who had started a business in the two years neur, that is if he or she was actively trying
preceding the survey. to start a business (YES ¼ 1 if the respondent
238 Pia Arenius and Maria Minniti

qualified as being involved in a new venture, 0

)0.012 ***
otherwise). In Model 1, we test the connection

0.33
between standard demographic and economic
9

characteristics and the likelihood of a person

being a nascent entrepreneur. In Model 2, we

)0.110 ***
0.253 ***
add perceptual characteristics to the variables
used in Model 1. In Model 3, we test the con-
8

nection between the likelihood of being a

0.231 *** nascent entrepreneur and perceptual variables
)0.039 ***
0.222 ***
only. Model 3 was created primarily to address,
though only indirectly, the collinearity issue. In
7

Model 4, we explore more in details the influence

of perceptual variables. In fact, Model 4 is a
)0.061 ***

)0.089 ***

0.059 ***
)0.082 ***

modified version of Model 2 in which we include

interaction terms between gender and all other
variables. By distinguishing between men and
6

women, Model 4 provides some information on

0.023 ***

0.024 ***

)0.019 ***
0.012 ***

the generality of the impact of various demo-

graphic, economic and perceptual characteristics
0.004

across our sample. Finally, in Model 5, we con-

5

sider the possibility of specific country effects by

adding dummy variables to the variables in
0.073 ***
)0.047 ***

0.083 ***

0.153 ***

0.098 ***

Model 2. All models are built using pooled data

Correlation table

0.000

(i.e. both men and women) across all 28 coun-

TABLE II

tries. Logistic regression results are presented in

Table III for all five models.
)0.200 ***

0.039 ***

)0.090 ***

)0.184 ***

0.052 ***
)0.124 ***
0.009 *

4. Results and Discussion

3

In Model 1, shown in Table III, we entered

0.015 ***
)0.027 ***
)0.085 ***

)0.039 ***

0.031 ***

)0.014 ***
)0.104 ***

variables measuring the five demographic and

0.001
39.490

economic characteristics of the respondent. Con-

2

sistently with existing literature, our results sug-

gest that an individual’s demo-economic make-up
***
***
***

***

0.165 ***

0.204 ***

)0.059 ***
0.157 ***

is quite important for understanding her likeli-

)0.051
)0.069
0.045
0.004
)0.031

hood of starting a new business. The chi-square

0.04

shows that the overall model is significant at the

1

0.000 level and it predicts 95.4% of the responses

80116
79879
80116
78892
78400
65746

67464

77204

76869
78852

correctly. All variables are significant.

Overall, we find that entrepreneurship is a
N

young man’s game. The coefficients of age and

Mean

39.49
0.04

1.50
2.34
3.47
2.06

0.26

0.40

0.33
0.34

gender show a negative and significant relation-

ship with the prevalence of nascent entrepre-
neurs. This is consistent with existing empirical
Knowing other
Fear of failure
in one’s skills

and theoretical literature showing that the rela-

Opportunity
Work status

Confidence
Household

perception
Education

tionship between age and the likelihood of start-

Nascent

entrepr.
Gender

income

ing a new business picks at a relatively early age

Age

and decreases thereafter (Levesque and Minniti,

forthcoming). The odds ratio for gender is 0.55
10

with a 95% confidence interval of [0.50, 0.59].

1
2
3
4
5
6

9
 Perceptual Variables and Nascent Entrepreneurship 239

This suggests that women are only half as likely information for new entrepreneurs (Weber and
to start a new business as men and is consistent Milliman, 1997). A different interpretation
with previous empirical findings (Reynolds et al., could takes into account that the knowing other
2003). Work status is related to the likelihood of entrepreneurs variable does not differentiate
being a nascent entrepreneur. In particular, between positive and negative examples. In
working individuals are shown to be more likely either case, however, direct observations of
to be nascent entrepreneurs than people in other other entrepreneurs reduce the ambiguity and
occupational groups.11 This is consistent with the uncertainty associated with entrepreneurial
observation that many people start new busi- activity thereby encouraging it (Minniti, 2005).
nesses while still holding on to a wage job. Opportunity perception is also positively and
Education is also positively related to the like- significantly related to being a nascent entrepre-
lihood of starting a new business. In particular, neur.16 The positive impact of this variable is
our results suggest that the likelihood of being easily reconciled with the economic theory of
nascent entrepreneurs increases steadily as indi- entrepreneurship according to which alertness
viduals have higher levels of education.12 This to unexploited opportunities is a necessary con-
result may be in part justified by the countries dition for entrepreneurial action (Kirzner, 1973,
included in our sample. It is also consistent with 1979).
Davidsson and Honig (2003), who link greater Confidence in one’s skills is positively and sig-
human capital with increased opportunity per- nificantly related to being a nascent entrepreneur.
ception and an increased likelihood of entrepre- The coefficient on the variable has a Wald statis-
neurial activity. Finally, household income is tics equal to 914.04 which is significant at the
associated with the likelihood of starting a new 0.000 level. The ‘‘odds ratio’’ for the confidence
business. As higher levels of income are consid- in one’s skills variable is 6.32 with a 95% confi-
ered, the relationship between nascent entrepre- dence interval of [5.61, 7.12]. This suggests that
neurship and household income can be depicted those who perceive themselves as possessing the
as a U-shaped curve.13 At low income levels, necessary skills are almost 6.4 times more likely
starting a new business may provide access to to be nascent entrepreneurs than those who do
high expected returns or simply the possibility of not believe to have the necessary skills. Thus,
employment when standard labor markets fail. confidence in one’s skills emerges as the single
At high income levels, instead, household income most important component of the decision to
reduces individuals’ financial barriers and start a new business (Koellinger et al., 2004). Our
increases their likelihood of starting a new busi- findings of the positive and very strong influence
ness. of this variable are consistent with the psychol-
In Model 2, four perceptual variables are ogy literature on intentionality and self-efficacy
added to the demo-economic ones included in (Baron, 2000). They are also consistent with
Model 1. According to the block chi-square sta- expectancy theory, which suggests that an indi-
tistics (p < 0.000), Model 2 is better than Model vidual’s belief about her ability to perform the
1 in explaining the probability of an individual task (being an entrepreneur) and the consequence
being a nascent entrepreneur. In Model 2, the of that task (to produce income and other non-
importance of age, gender, and work status is vir- monetary rewards) affect whether or not she will
tually unchanged.14 Both education and household undertake the task at all.
income, on the other hand, appear to be less rele- Finally, fear of failure has a negative and
vant. All four perceptual variables are highly sig- significant impact on being a nascent entrepre-
nificant. neur.17 The negative connection between this var-
In particular, knowing other entrepreneurs is iable and the likelihood of being a nascent
positively and significantly related to being a entrepreneur is consistent with Weber and Milli-
nascent entrepreneur.15 The positive impact man’s (1997) suggestion that an increased per-
of knowing an entrepreneur might be explained ception of the probability of failure reduces
by the fact that role models and being part entrepreneurial incentives by increasing the per-
of networks reduce ambiguity and provide ceived riskness of starting a business.
 TABLE III

240
Logistic regression results: Dependent variable = YES (nascent entrepreneurs)

Model 1 Model 2 Model 3 Model 4 Model 5

Coefficient Wald Coefficient Wald Coefficient Wald Coefficient Wald Coefficient Wald
(std. error) (std. error) (std. error) (std. error) (std. error)

Demographic and economic variables

Age )0.022*** 158.192 )0.023*** 152.689 )0.028*** 24.397 )0.019*** 98.815
(0.002) (0.002) (0.006) (0.002)
Gender )0.611*** 197.819 )0.212*** 21.594 )0.348 2.824 )0.247*** 28.455
(0.043) (0.046) (0.207) (0.046)
Work status Working 92.245*** 46.111*** 2.970 36.388***
Not )0.157** 7.451 0.074 1.486 0.321 2.865 )0.058 0.882
working (0.058) (0.060) (0.190) (0.062)
Retired )0.971*** 89.349 )0.685*** 42.238 )0.047 0.023 )0.649*** 36.282
or student (0.103) (0.105) (0.319) (0.108)
Education Secondary 85.281*** 3.897 8.356 13.215*
degree
None )0.875* 4.403 )0.628 2.195 2.497 1.091 )1.123** 6.838
(0.417) (0.424) (2.391) (0.429)
Some )0.135* 6.134 )0.044 0.600 )0.413* 5.857 )0.165** 7.237
secondary (0.054) (0.057) (0.171) (0.061)
Post 0.300*** 38.185 0.040 0.618 )0.026 0.029 )0.052 0.942
secondary (0.049) (0.051) (0.152) (0.054)
Graduate 0.432*** 17.175 0.110 1.027 )0.292 0.806 )0.019 0.026
experience (0.104) (0.108) (0.326) (0.114)
Household Middle 18.431*** 7.314* 3.043 1.282
Pia Arenius and Maria Minniti

income 33% tile

Upper 0.203*** 18.374 )0.009 0.030 )0.045 0.091 0.025 0.247
33%tile (0.047) (0.049) (0.148) (0.051)
Lowest 0.110* 4.025 0.137* 5.700 0.249 2.074 0.0367 1.280
33%tile (0.055) (0.057) (0.173) (0.059)
Perceptual variables
Knowing other entrepr. 0.864*** 344.770 0.939*** 420.373 1.037*** 54.322 0.872*** 334.254
(0.047) (0.046) (0.141) (0.048)
Opportunity 0.835*** 369.443 0.855*** 393.453 0.706*** 29.268 0.883*** 381.528
perception (0.043) (0.043) (0.130) (0.045)
Confidence in 1.844*** 914.041 1.838*** 931.652 1.862*** 98.058 1.773*** 809.679
one’s skills (0.061) (0.060) (0.188) (0.062)
Fear of failure )0.422*** 67.027 )0.388*** 57.523 )0.417** 7.224 )0.446*** 71.619
(0.052) (0.051) (0.155) (0.053)
Gender Interaction Terms
Age * gender 0.004 0.796
(0.004)
Work status * gender Working * 5.683
gender
Not working * )0.170 1.979
gender (0.121)
Retired or stud. * )0.483* 4.279
gender (0.234)
Education * gender Sec. degree * 11.349*
gender
None *gender )2.726 1.447
(2.266)
Some sec. * 0.338** 8.117
gender (0.119)
Post second. * 0.048 0.199
gender (0.108)
Graduate * 0.297 1.735
gender (0.226)
Income * Middle 0.867
gender 33% tile
Upper 0.028 0.071
33%tile (0.104)
Lowest )0.083 0.496
33%tile (0.118)
Knowing *gender )0.126 1.715
(0.096)
Opp. percep. * gender 0.096 1.099
(0.092)
Confidence *gender )0.011 0.008
(0.123)
Fear of fail. *gender )0.002 0.000
(0.107)
Models diagnostics
Constant )1.318*** 190.380 )3.693*** 948.761 )4.892*** 6131.511 )3.518*** 135.219 )4.338*** 778.487
Perceptual Variables and Nascent Entrepreneurship

(0.096) (0.120) (0.062) (0.303) (0.155)

N 51 721 51 721 51 721 51 721 51 721
Model v2 [df] 699.93*** 3708.78*** 3460.67*** 3733.62*** 4291.06***
[10] [14] [4] [27] [41]
Block v2 [dt] 3008.85*** 24.84* 582.28***
[4] [13] [27]
Overall % correct predictions 95.0 95.0 95.0 95.0 94.9
R2 0.041 0.210 0.197 0.212 0.242

*** significant at p £ 0.001; ** significant at p £ 0.01; *significant at p £ 0.05.

241
242 Pia Arenius and Maria Minniti

Overall we find that the perceptual variables who are less afraid of failure are more likely of
are powerful predictors of the likelihood of being nascent entrepreneurs. These results seem
being a nascent entrepreneur. In fact, Model 2 to confirm the importance of perceptual variables
suggests that the impact of the perceptual vari- as drivers of entrepreneurial behavior for both
ables may be stronger than that of standard genders. We also assessed the reliability of our
demo-economic variables. As mentioned before, results by internal replication, i.e., running sepa-
although in Model 2 all four perceptual variables rate regression analyses for each country. Across
are highly significant, the importance of educa- all 28 regressions, knowing other entrepreneurs,
tion and household income is reduced with opportunity perception and confidence is one’s
respect to Model 1.18 Although Dolinsky et al. skills were significant positive predictors of being
(1993) have argued that perceptual factors may a nascent entrepreneur. Age, having some second-
mediate the effects of education and household ary education and fear of failure were consistently
income, we can only suggest that perceptual vari- significant negative predictors of the likelihood of
ables are important but have no evidence to being a nascent entrepreneur. Work status and
rank their effects. household income were not significant.
To address, at least in part, the collinearity Finally, in Model 5, we test for the possibility
issue, and provide more evidence about the of country effects by using deviation coding. Devi-
importance of perceptual variables, we tested ation coding allows us to compare each individ-
Model 3 and Model 4. In Model 3, only the four ual country against the mean for all countries.
perceptual variables are included in the regres- We constructed a dummy for each individual
sion. All four are highly significant. A compari- country (e.g., U.S.A. ¼ 1 if country is U.S.A.;
son of the coefficient on these four variables in otherwise U.S.A. ¼ 0) and selected Singapore as
Model 2 and Model 3 suggests that they are very the reference country and coded it as )1 on all
similar and that the addition of demo-economic other country dummies. Singapore was selected
variables in Model 2 has very little effect on because its nascent prevalence rate is 4.5% which
them. In Model 4, instead, we consider whether is also the average across all 28 countries in our
gender changes the relationship between our sample. We then entered all the country dummies
dependent variable and the independent variables at the same time into the regression analysis.20 In
by adding gender interaction terms. Our results Model 5, the effects of age and gender remain
in Model 4 suggest that the relationships between significant and negative. However, working and
the likelihood of becoming a nascent entrepre- non-working respondents are equally likely to
neur and age, household income, and all percep- engage in nascent entrepreneurship. Furthermore,
tual variables do not depend on gender. Thus, the positive effects of post secondary and graduate
consistently with existing literature (Brush, 1990, experience disappear. All perceptual variables
1992; Langowitz and Minniti, 2005), our results remain highly significant. The coefficients for
suggest that there are only minor individual level opportunity perception, confidence in one’s skills
differences between men and women, and that and knowing other entrepreneurs are all significant
environment and institutional factors should be and positive, whereas, as expected, fear of failure
examined to explain the lower entrepreneurship is negatively related to being a nascent entrepre-
rate for women.19 neur. Finally, for 21 of the 28 countries, the
More important, Model 4 suggests that the country effect dummies are significant.21 Overall,
relationship between the likelihood of becoming Model 5 suggests that the introduction of coun-
an entrepreneur and all perceptual variables does try specific effects, though significant, does not
not depend on gender. Both men and women reduce the cross-country importance of percep-
who know other entrepreneurs, who perceive the tual variables.
existence of good opportunities, and who feel The fact that in 21 cases, the country effect
they have the ability, skills, and knowledge dummies are significant suggests that local influ-
required for starting a new business are more ences are at work. Consistently with existing lit-
likely to do so than their counterparts with differ- erature (Acs and Audretsch, 1993; Thurik et al.,
ent perceptions. Finally, both men and women 2002; Reynolds, 2003), our results, although very
 Perceptual Variables and Nascent Entrepreneurship 243

preliminary, suggest that, given their characteris- ceptual variables reflect subjective perceptions
tics, the macroeconomic environments of some rather than objective conditions. As a result, they
countries are more conducive to entrepreneurial are likely to be biased. There exists some evi-
behavior while others penalize it. Clearly, the dence that distortions in perceptions are com-
relationship between cross-country and country mon among individuals in general, and among
specific drivers of entrepreneurial behavior is a entrepreneurs in particular (Cooper et al., 1988;
complex and important one and much more Busenitz and Barney 1997). The importance of
work is needed in this area. perceptual variables, and their associated bias, in
Finally, even after all the country dummies the decision to start a new business may explain
are included, about 75% of the variance remains some of the observable inconsistencies between
unexplained. Although this is not an unusual expected and actual returns to entrepreneurship
result in empirical economic research, it is worth and entrepreneurial decisions found in the litera-
further discussion. First, it is most likely that our ture (Hamilton, 2000) as well as the fact that
study does not include all relevant variables. In many new businesses fail shortly after inception
particular, given the data available, our consider- (Dunne et al., 1988, Geroski, 1995).
ation of country specific effects is limited to the The data, unfortunately, does not allow us to
inclusion of dummy variables. This is far from establish the causal direction of the relationship.
satisfactory since we are not able to specify the In other words, we cannot establish whether con-
nature of important cross-country differences. fidence in one’s own skills, knowledge and ability
Second, perhaps the linear additive model is not is a driver of entrepreneurial behavior or if,
the most appropriate conceptual structure for the instead, the start up process increases an individ-
study of nascent entrepreneurship. The latter ual perception of her own ability. Noticeably,
issue has been partly addressed, among others, however, the percentage of individuals who
by Minniti (2005) who has shown the existence responded yes to questions concerning perceptual
of non-linearities in the decision to start a new variables varies significantly when alternative
business.22 groups are considered. Table IV shows that nas-
cent entrepreneurs in our sample, as well as new
(babybus) and more established entrepreneurs
5. Conclusions and extensions
(experienced), are much more likely to respond
In this paper, we use original data to estimate bi- affirmatively to questions about perceptual vari-
nominal logistic regression models based on a ables than respondents who are not active in
standard economic approach of gender neutrality starting or managing a business (not active).
in which economic and demographic variables Also, using 2001 GEM data for 18 countries,
are considered. We then add a set of relevant Koellinger et al. (2004) confirm our result and
perceptual variables and, finally, country dum- show that the confidence associated with one’s
mies to account for macroeconomic differences. own skills and ability declines as more experi-
Our results suggest that the addition of percep- enced entrepreneurs are considered. It is our
tual variables improves the statistical fit of the hope that other studies will follow our explor-
model. Entrepreneurship is, after all, about peo- ative investigation and lead to an unambiguous
ple and, not surprisingly, subjective and often understanding of how perceptions influence
biased perceptions emerge as being highly corre- entrepreneurial behavior.
lated to nascent entrepreneurship. To our knowl- Should they be confirmed by further research,
edge, no previous model had incorporated findings that perceptual variables are highly cor-
perceptual variables at the individual level. Our related to the likelihood of being a nascent entre-
results suggest that it is appropriate to do so. preneur are also important because of their
Our most significant result is the finding that, potential policy implications. Perceptual factors
across all countries and across genders, percep- such as opportunity recognition, knowing other
tual variables and, in particular, the perception entrepreneurs, and fear of failure, are all charac-
that individuals have of their own entrepreneurial teristics of the individual. They cannot be easily
abilities are very important. Unfortunately, per- changed by exogenous interventions such as, for
244 Pia Arenius and Maria Minniti

TABLE IV
Responses to questions about perceptual variables (percentage reporting YES)

Not active Nascent Babybusa Experiencedb

Opportunity perception 24.0% 58.2% 50.5% 37.9%

Confidence is one’s skills 38.3% 86.5% 84.6% 81.7%
Fear of failure 33.6% 20.3% 19.0% 20.8%
Knowing other 32.4% 68.4% 66.7% 53.4%
entrepreneurs

a
Babybus includes start-ups that have been in existence for more than 3 but less than 42 months.
b
Experienced includes start-ups that have been in existence for more than 42 months.

example, government policies. While the latter industries. The second type is policy aimed at
can alter an individual’s incentives, inherent fac- providing newly created firms, irrespective of
tors depend on her specific history. They are their industrial classification, with the financial,
path-dependent and, as a result, do not change organizational and technological resources
or change very slowly. Although perceptual vari- needed to grow in both domestic and foreign
ables do change over time, to alter the way in markets. This type of generic policy in particu-
which individuals think about themselves and lar promotes variation among new businesses,
their role in society takes a long time. In fact, creating the basis for a selection process that
should the causal direction of the relationship be may result in new products and approaches.
unambiguously established, that is, should fur- Finally, several extensions of our project are
ther studies show that optimistic perceptions possible. First, although our empirical models
increase the likelihood of starting a new business, are rooted in extant theoretical and empirical
our results would imply that the entrepreneurial evidence in economics and other social sciences,
history of a community is important in determin- it would be desirable to develop an analytical
ing a policy’s effectiveness or lack thereof. The and comprehensive model of how, and why,
adjustment mechanism produced by government individuals choose to start a business. Such a
interventions is, itself, a path-dependent process. model could be very useful in studying the
Therefore, the same policy may have very differ- architecture of entrepreneurial decisions. In
ent results in different communities. other words, in understanding how perceptual
Differences in aggregate conditions such as variables interact with the demo-economic
technology, level of economic development, profile of an individual and contribute to her
institutions, and culture all cause differences in decisions. Efforts in this direction could address
opportunities, resources, skills and preferences theoretically the issues of non-linearities and of
with regard to entrepreneurship. Policies pro- causality in entrepreneurial decisions. As a
viding incentives schemes for new and small result, they could lead to more accurate empiri-
firms by removing impediments for business cal estimates of nascent entrepreneurship and
start-ups may suffer from decreased probabili- its drivers, and contribute to improving the
ties of new firm survival without achieving predictive ability of our model.
much transformation towards an entrepreneurial Second, an important component of our
economy. On the other hand, the many small study is the introduction of country effects.
start-ups created in this way may function as a Entrepreneurs do not make decisions in a vac-
seedbed for a small number of successful and uum. A growing body of recent literature has
innovative new firms. Within this context, addressed explicitly the relationship between the
Thurik et al. (2002) argue that there is room entrepreneurial sector and the macroeconomy
for at least two types of policy intervention. (for example, Acs et al., 1999; Wennekers and
The first type is policy aimed at promoting the Thurik, 1999; Carree et al., 2000). In our mod-
creation of new technology-based firms in new els, the current treatment of country effects is
 Perceptual Variables and Nascent Entrepreneurship 245

limited to the introduction of dummy variables. Notes

It would be desirable to include data on the 1
role played by macroeconomic factors such as Works listed in Table I are only a small sample of the
vast literature existing on the causes and motivations of
level of development, technological sophistica- entrepreneurial behavior. Also, the scope of some of the
tion, institutions and culture (Thurik et al., listed works spans across more than one group of vari-
2002). The institutional framework, for exam- ables. In such a case they have been listed as part of the
ple, is crucial in determining the quantity and group in which they provide the strongest contribution.
2
quality of entrepreneurial behavior as it defines If entrepreneurship refers to the introduction of techno-
logical innovations, we would expect a positive correla-
individuals’ incentives to transform perceived tion between post-secondary education and start-up
opportunities into actions. Baumol (1990) rates. On the other hand, if entrepreneurship refers more
argued explicitly that institutional arrangements broadly to the exploitation of market opportunities, we
affect the quantity and type of entrepreneurial may expect people with little or no education to be penal-
efforts. Technology and level of development ized by traditional labor markets and to be more likely
to start their own business.
are also important. New technologies lead to 3
In this context, however, entrepreneurship is often
new markets, thereby creating opportunities for identified with self-employment though the two are
new start-ups. Very often, new technologies clearly not identical (Amit et al., 1993).
4
also lead to reductions in transaction costs Clearly, knowing other entrepreneurs is not only a per-
whose resulting scale effects allow smaller and ceptual variable. In addition to its contribution to the
subjective perception that an individual forms about
newer firms to compete more effectively with entrepreneurship, it may be also considered as an indica-
incumbent ones. Finally, the level of develop- tor of social capital. In the context of this paper, how-
ment influences environmental opportunities by ever, the latter is captured in the third group of variables.
5
creating the needs for deeper markets and the Additional details about the GEM project may be
more resources for nascent entrepreneurs. found at www.gemconsortium.org
6
For our purposes, complete sets of data were available
Overall, our study suggests that individual per- for 28 countries. Countries included in our sample are
ceptions matter and that richer descriptions of Argentina, Australia, Belgium, Canada, Chile, Croatia,
local institutions and public practices should be Denmark, Finland, France, Germany, Hong Kong, Hun-
introduced in microeconomic models of entrepre- gary, Israel, Japan, Korea, Mexico, Netherlands, Nor-
neurial behavior. It is our hope that our work way, Poland, Russia, Singapore, Switzerland, Slovenia,
South Africa, Sweden, Thailand, UK, U.S.A.
will encourage more accurate descriptions of 7
All surveys were administered by phone, except in Rus-
individuals’ choices in which decisions with sia, South Africa, Hungary, Mexico, and Thailand where
respect to entrepreneurship are based not only on they were conducted face-to-face.
8
objective demo-economic factors but, also, on An analysis of the individuals excluded from our final
subjective and possibly mistaken perceptions. sample showed no significant differences between the two
groups of respondents.
9
Details about the procedures used to collect and har-
Acknowedgements monize GEM data can be found in Reynolds et al.
(2005).
10
We thank the GEM Consortium for granting us For the nascent entrepreneurship prevalence rates in
the use of the data. Particular thanks go to Paul each country see Reynolds et al. (2003).
11
The odds ratio for the not working variable is 0.86
Reynolds and Steve Hunt for their assistance with a 95% confidence interval of [0.78, 0.96]. This sug-
with the data. We also thank Nancy Carter, Rolf gests that those who are not working are only nine tenth
Sternberg, Sander Wennekers, and two referees as likely to be nascent entrepreneurs as those working
for helpful comments and suggestions. Maria full-time. The odds ratio for the retired or student vari-
Minniti gratefully acknowledges financial support able is 0.39 with a 95% confidence interval of [0.32,
0.47]. This suggests that those who are retired or students
from the A. Blank Center for Entrepreneurship, are only two fifth as likely to be nascent entrepreneurs as
the W. Glavin Center for Global Management, those working full-time.
12
and the Center for Women’s Leadership at Bab- The odds ratio for the no education variable is .41
son College. Pia Arenius gratefully acknowledges with a 95% confidence interval of [0.19, 0.86]. This sug-
financial support from the Ministry of Trade and gests that those who have no education are only two
fifth as likely to be nascent entrepreneurs as those with
Industry Finland and Tekes, and the Academy of a secondary level degree. The odds ratio for the some
Finland. All errors are ours.
246 Pia Arenius and Maria Minniti

secondary variable is 0.90 with a 95% confidence inter- country and repeated the analysis. In both regressions,
val of [0.81, 0.99]. This suggests that those who have the demo-economic and perceptual variables have the
some secondary level education are only nine tenth as same betas and the model’s diagnostics are unchanged.
likely to start a new business as those with a secondary The beta coefficients of some individual country dummies
level degree. The odds ratio for the post secondary vari- change slightly, their significance, however, remains
able is 1.36 with a 95% confidence interval of [1.24, unchanged.
21
1.49]. This suggests that those who have post secondary To save space, Table IV does not include the coefficients
education are 1.4 times more likely to become nascent for the dummy variables representing each country. How-
entrepreneurs than those with a secondary level degree. ever, Argentina, Canada, Chile, Germany, Korea, Mexico,
The odds ratio for the post secondary variable is 1.58 Singapore, South Africa, Thailand and U.S.A. are coun-
with a 95% confidence interval of [1.31, 1.90]. This sug- tries where the likelihood of becoming a nascent entrepre-
gests that those who have a graduate degree are 1.6 neur is, on average, higher than in other countries.
times more likely to become nascent entrepreneurs than Thailand and Singapore emerge as countries where the
those with a secondary degree. likelihood of entrepreneurial activity appears to be particu-
13
Those belonging to the highest income group are 1.2 larly high. Countries, where on average, the likelihood of
times more likely to start a new business than those becoming a nascent entrepreneur is lower than the aver-
belonging to the middle income group. The odds ratio ages are Australia, Belgium, Denmark, Finland, France,
for the Upper 33% variable is 1.23 with a 95% confi- Hungary, Israel, The Netherlands, Slovenia, Sweden, and
dence interval of [1.13, 1.34]. Those belonging to the low- UK. Country effects are not significant for Croatia, Hong
est income third are 1.1 times more likely to become Kong, Japan, Norway, Poland, Russia, and Switzerland.
22
nascent entrepreneurs than those belonging to the middle Clearly, other reasons can account for the value of
income third. The odds ratio for the Lowest 33% is 1.14 our R2. The composition of our sample (which includes
with a 95% confidence interval of [1.03, 1.26]. both men and women) may be one of them. In fact, the
14
The only exception is not working whose moderate variables investigated may be poor predictors of nascent
importance is further reduced and for which we observe entrepreneurship for women. Delmar and Davidsson
a sign switch. This result may indicate that the effects of (2000), for example, considered age, education, gender,
perceptual variables may counteract some insecurity parents employment status, management experience and
caused by a not working status. attitude toward self-employment and found that while
15
The odds ratio for knowing other entrepreneurs is 2.37 researchers have a reasonably good understanding of the
with a 95% confidence interval of [2.17, 2.60]. This sug- characteristics associated with men going into business,
gests that those who know an entrepreneur are 2.4 times the same variables do not succeed in predicting nascent
more likely to become entrepreneurs than those who do entrepreneur status for women.
not know an entrepreneur.
16
The coefficient of the opportunity perception variable
has a Wald statistics equal to 369.45 which is significant References
at the .000 level. The odds ratio for opportunity percep-
tion is 2.31 with a 95% confidence interval of [2.12, 2.51]. Acs, Z. J. and D. B. Audretsch (eds.), 1993, Small Firms
This suggests that those who report perceiving opportuni- and Entrepreneurship: An East–West Perspective, Cam-
ties are 2.3 times more likely to become nascent entrepre- bridge, UK: Cambridge University Press.
neurs than those who do not perceive opportunities. Acs, Z. J., B. Carlsson and C. Karlsson, 1999, ‘The Link-
17 ages among Entrepreneurship, SMEs and the Macroec-
The odds ratio for the fear of failure variable is .66 with
a 95% confidence interval of [0.59, 0.73]. This suggests onomy’, in Z. J. Acs, B. Carlsson and C. Karlsson
that those who fear failure are only two thirds as likely to (eds.), Entrepreneurship, Small and Medium-Sized
become entrepreneurs as those who do not fear it. Enterprises and the Macroeconomy, Cambridge, UK:
18 Cambridge University Press.
This confirms the presence of collinearity already indi-
cated by the correlation coefficients in Table III. Aldrich, H., 1999, Organizations Evolving, London: Sage
19 Publications.
Only two interactions variables are significant. We find
that retired women and female students are two thirds as Amit, R., L. Glosten and E. Muller, 1993, ‘Challenges to
likely to start a new business as retired men and male Theory Development in Entrepreneurship Research’,
students, and that women with some secondary level edu- Journal of Management Studies 30, 815–834.
cation are more likely to start a new business than men Baron, R., 2000, ‘Psychological Perspectives on Entrepre-
with a similar level education. This is consistent with Lef- neurship: Cognitive and Social Factors in Entrepre-
kowitz (1994) who shows that men and women react sim- neurs’ Success’, Current Directions in Psychological
ilarly to the work environment when one controls for Science 9, 15–19.
spurious effects caused by systematic differences in types Baumol, W. J., 1990, ‘Entrepreneurship: Productive,
of job and job payments. Unproductive, and Destructive’, Journal of Political
20 Economy, 98, 893–921.
To provide information about Singapore, since the
country was used as the reference in the deviation coding, Begley, T. and D. Boyd, 1987, ‘Psychological Characteris-
we recoded the data using U.S.A. as the new reference tics Associated with Performance in Entrepreneurial
 Perceptual Variables and Nascent Entrepreneurship 247

Firms and Small Businesses’, Journal of Business Ven- Harper, D. 1998, ‘Institutional Conditions for Entrepre-
turing 2, 79–93. neurship’, Advances in Austrian Economics 5, 241–275.
Blanchflower, D. G., 2004, Self-Employment: More May Jack, S. L. and A. R. Anderson, 2002, ‘The Effects of
Not Be Better, NBER Working Paper No. 10286. Embeddedness on the Entrepreneurial Process’, Journal
Blanchflower, D. G. and A. J. Oswald, 1998, ‘What of Business Venturing, 17, 467–487.
Makes an Entrepreneur?’ Journal of Labor Economics Kihlstrom, R. and J. Laffont, 1979, ‘A General Equilib-
16(1), 26–60. rium Entrepreneurial Theory of Firm Formation Based
Brush, C. G., 1990, ‘Women and Enterprise Creation: on Risk Aversion’, Journal of Political Economy, 87,
Barriers and Opportunities’, in Gould, S. and J. Parzen 719–740.
(eds.), Enterprising Women: Local Initiatives for Job Kirzner, I. M., 1973, Competition and Entrepreneurship,
Creation, OECD, Paris. Chicago, IL: University of Chicago Press.
Brush, C. G., 1992, ‘Research on Women Business Own- Kirzner, I. M., 1979, Perception, Opportunity, and Profit,
ers: Past Trends, a New Perspective and Future Direc- Chicago, IL: University of Chicago Press.
tions’, Entrepreneurship Theory and Practice 16, 5–30. Koellinger, P., M. Minniti and C. Schade, 2004, I Think I
Busenitz, L. W. and J. B. Barney, 1997, ‘Differences Can, I Think I Can: A Cross-Country Study of Entre-
between Entrepreneurs and Managers in Large Organi- preneurial Motivation, Working paper, Humboldt Uni-
zations: Biases and Heuristics and Strategic Decision- versity, DE.
Making’, Journal of Business Venturing 12, 9–30. Langowitz, N. and M. Minniti, 2005, Gender Differences
Carree, M., A. van Stel, R. Thurik and S. Wennekers, 2000, and Nascent Entrepreneurshipm Working paper.
Business Ownership and Economic Growth in 23 OECD Leazar, E. P., 2002, Entrepreneurship, NBER Working
Countries. Tinbergen Institute Discussion Paper TI Paper N.9109.
2000–1/3, Tinbergen Institute, Rotterdam, NL. Lefkowitz, J., 1994, ‘Sex Related Differences in Job Atti-
Chell, E. and S. Baines, 2000, ‘Networking, Entrepreneur- tudes and Dispositional Variables: Now You See Them_s’,
ship and Micro-Business Behavior’, Entrepreneurship Academy of Management Journal 37(2), 323–349.
and Regional Development, 12, 195–215. Levesque, M. and M. Minniti, ‘The Effect of Aging on
Cooper, A. C., C. Y. Woo and W. C. Dunkelberg, 1988, Entrepreneurial Behavior’, Journal of Business Ventur-
‘Entrepreneurs’ Perceived Chances for Success’, Journal ing, forthcoming.
of Business Venturing 3, 97–108. Minniti, M., 2004, ‘Entrepreneurial Alertness and Asym-
Davidsson, P. and B. Honig, 2003, ‘The Role of Social metric Information in a Spin-Glass Model’, Journal of
and Human Capital Among Nascent Entrepreneurs’, Business Venturing 19(5), 637–658.
Journal of Business Venturing 18, 301–331. Minniti, M., 2005, ‘Entrepreneurship and Network Exter-
Delmar, F. and P. Davidsson, 2000, ‘Where Do They nalities’, Journal of Economic Behavior and Organiza-
Come From? Prevalence and Characteristics of Nas- tion, in press.
cent Entrepreneurs’, Entrepreneurship and Regional Murphy, K. M., A. Shleifer and R. Vishny, 1991, ‘The
Development 12, 1–23. Allocation of Talent: Implications for Growth’, Quar-
Dolinsky, A. L., R. K. Caputo, K. Pasumarty and terly Journal of Economics 106, 503–530.
H. Quazi, 1993, ‘The Effects of Education on Business Reynolds, P. D., B. Bygrave and M. Hay, 2003, Global
Ownership: A Longitudinal Study of Women’, Entre- Entrepreneurship Monitor Report, Kansas City, MO: E.
preneurship Theory and Practice 18(1), 43–53. M. Kauffmann Foundation.
Dunne, T., M. J. Roberts and L. Samuelson, 1988, ‘Pat- Reynolds, P. D., N. S. Bosma and E. Autio et al., 2005,
terns of Firm Entry and Exit in U.S. Manufacturing ‘Global Entrepreneurship Monitor: Data Collection
Industries’, Rand Journal of Economics 19(4), 495–515. Design and Implementation 1998–2003’, Small Business
Evans, D. and B. Jovanovic, 1989, ‘An Estimated Model Economics, 24(3), 233–247 (this issue).
of Entrepreneurial Choice Under Liquidity Con- Taylor, M. P., 1996, ‘Earnings, Independence or Unem-
straints’, Journal of Political Economy, 97, 808–827. ployment: Why Become Self-employed?’ Oxford Bulle-
Gartner, W. B., 1985, ‘A Conceptual Framework for tin of Economics and Statistics 58(2), 253–265.
Describing the Phenomenon of New Venture Crea- Thurik, A. R., L. M. Uhlaner and S. Wennekers, 2002,
tion’, Academy of Management Review 10, 696–706. ‘Entrepreneurship and Its Conditions: A Macro Per-
Geroski, P. A., 1995, ‘What Do We Know About Entry?’ spective’, International Journal of Entrepreneurship
International Journal of Industrial Organization 13, 421–440. Education 1(1), 25–64.
Greene, P., 2000, ‘Self-Employment as an Economic Weber, E. U. and R. A. Milliman, 1997, ‘Perceived Risk
Behavior: An Analysis of Self-Employed Women’s Attitudes: Relating Risk Perception to Risky Choice’,
Human and Social Capital’, National Journal of Sociol- Management Science 43, 123–144.
ogy 12(1), 1–55. Wennekers, S. and A. R. Thurik, 1999, ‘Linking Entrepre-
Hamilton, B. H., 2000, ‘Does Entrepreneurship Pay? An neurship and Economic Growth’, Small Business Eco-
Empirical Analysis of the Returns to Self-Employ- nomics 13, 27–55.
ment’, Journal of Political Economy, 108, 604–631.

View publication stats