1. Introduction
Entrepreneurial activity is one of the most important factors contributing to the economic progress of a country [
1,
2]. Entrepreneurship may stimulate growth by introducing innovations, creating change, creating competition and enhancing rivalry [
3]. In the context of an evolving economy, entrepreneurs are important assets because they also contribute to social development [
4,
5], serve as agents of change by bringing new ideas to markets [
6,
7], may introduce important innovations by placing on the market new products and services [
8], and create an attractive business environment by encouraging the creation of new businesses [
9].
In the last years, many researchers have studied the importance of this phenomenon considering three main perspectives: the economic factors [
3,
8,
10], the socio-cultural factors [
5,
11], and perception determinants [
8,
12,
13,
14,
15]. Yet, several previous studies found some mixed results about the relationships between some perceptual variables and entrepreneurial activities. For instance, according to Özdemir and Karadeniz [
6], the fear of failure is not a significant factor influencing the likelihood of engaging in entrepreneurial activities. The authors explain this phenomenon by the characteristics of an individual’s national culture, as individuals may be accustomed to the economic instability or uncertainty of their country, which influences their risk perception or fear of failure. Also, other studies comparing different regions or countries reveal that entrepreneurial activities and their main determinants may differ in intensity and impact according to the country/region development level [
6,
16]. This may be due to the fact that an important feature of the most developed countries, which are innovation-focused economies, is that SMEs support innovation [
8] and represent the most important factor contributing to economic prosperity and job creation [
1,
3]. Similarly, Beynon et al. [
8] consider that developed countries prioritize the companies supporting innovation in their economic growth policies, given that the economy is based on these companies. In the same study, the authors state that in less developed countries with generally necessity-based economies, entrepreneurial activity varies according to the level of economic development (it decreases when the economy is dominated by the manufacturing industry, and increases when it is dominated by the service sector—innovation driven phase), which means that different factor combinations can affect the entrepreneurial activity.
Based on this theoretical background, the aim of this study is to focus on the psychological perspective through perceptual factors, and to analyse the role of entrepreneurial attributes as main determinants which explain early-stage entrepreneurial activity during, and after, the recent economic and financial crisis. To achieve this goal, we focused on the influence of four main entrepreneurial attributes on the propensity to engage in TEA, namely: Capability of identifying opportunity; having the skills, knowledge and experience to start up a business or businesses; fear of failure; and knowing other entrepreneurs [
17]. However, given the fact that the entrepreneurial activity determinants are sensitive to time and a country’s GDP [
5,
18,
19], the study examines two different moments in time, namely 2007 and 2014, in order to capture the fluctuations in entrepreneurial activities, and the effects of different perceptual factors, in less and high developed European member states.
Although, the specific body of research offers a long list of variables influencing the entrepreneurial activities, some studies offer mixed results. For instance, in certain analyses, the fear of failure is not always statistically significant [
20] and does not prevent people from becoming entrepreneurs in certain industries [
21]. These results are not coherent with the existing literature on entrepreneurship by Beynon et al. [
8]; Albulescu, Tamasila [
10], and do not support the fear of failure as a trigger of early-stage entrepreneurial activity. We contribute to the existing literature by expanding the knowledge on the determinants of cross-national variations in entrepreneurial activity across 18 European countries, and in two different time periods: 2007 and 2014.
Secondly, from a methodological perspective, the majority of studies on entrepreneurial activity determinants using GEM data apply logistic regression models to measure the relationship between a dichotomous (dependent) variable and a set of predictors, without taking into consideration the potential moderating effects. To bridge this gap, our analysis accounts for the potential moderating effects of gender and the set of independent variables used in the analysis. The limitation of logit models, and the fact that GEM Adult Population Survey, which included mostly “yes” and “no” answers, limiting the use of a more accurate statistical analysis, we evaluated the relative importance of predictors in logistic models through fully standardised logistic regression coefficient estimations. This method enabled us to determine the relative importance of predictors [
22].
In order to achieve its objective, this study presents the following sections: introduction, literature analysis and development of research hypotheses, research methodology, results and discussions and conclusions. The first part includes an analysis on the technical literature in the field, according to factors which influence the entrepreneurial activity. In the second part, the research methodology provides information on the research method, sampling and the description of the variables. The final part contains the results and conclusions of the research.
3. Materials and Methods
The objective of our analysis focused on studying the role of adult populations’ attitudes regarding entrepreneurship. Our goal is to measure the effects of entrepreneurial attitudes, such as the
capability to identify opportunity,
having skills, knowledge and experience to start up businesses,
fear of failing, and
knowing other entrepreneurs on the likelihood to engage in early-stage entrepreneurial activity. In order to analyse the influence of these perceptual factors and entrepreneurial activity, we used data retrieved from Adult Population Survey, Global Entrepreneurship Monitor (GEM) annual survey for 2007 and 2014, and a total of 18 European countries. We selected the countries based on data availability for the period analysed [
26,
27].
We chose Global Entrepreneurial Monitor (GEM) because it is one of the few standardised datasets on entrepreneurship. GEM enables across-national comparison, and provides a robust framework for analysing different aspects of entrepreneurships. GEM data not only provides a broad overview on entrepreneurship, but also focuses on the behaviour of individuals with respect to starting and managing a business. Also, with minor exceptions, the data for each of the selected country contains a representative sample of working-age population of 2000 or more. The number of variables used in the modelling procedure are described in
Table 2. The study uses logit regression models to estimate the probability that individuals belonging to a specific group have a high susceptibility to several cognitive biases. We selected the log-it regression analyses because the vast majority of the variables were dichotomous, including our dependent variable. This type of analysis enabled us to estimate the effect of perceptual factors, interactions and control variables on the probability of engagement in entrepreneurial activities. The total sample used in the analysis is divided into two groups:
G1—Less developed countries: Croatia, Greece, Hungary, Latvia, Portugal, Romania, Slovenia and Spain;
G2—High developed countries: Belgium, Denmark, Finland, France, Germany, Ireland, Italy, The Netherlands, Sweden and United Kingdom.
The delimitation criteria for the two groups were based on annual GDP per capita (using data provided by World Bank [
51]. Countries with income below the average mean were classified as G1 and those with the level of income above the average mean were classified as G2.
Dependent Variable
The dependent variable used in our research is:
Entrepreneurial activity (indicates how entrepreneurial societies really are). For measuring the level of entrepreneurial activity, we used as proxy the total early-stage entrepreneurial activity (TEA). The variable includes the category of population aged between 18 and 64 who is either actively trying to start a new business, or managing a business which is less than three-and-a half years old. TEA includes two categories of entrepreneurs: a)
opportunity-driven early stage entrepreneurs (the respondents aged 18-64 who are pulled to entrepreneurship by opportunity and their desire to become independent and increase their income, and who are motivated to pursue perceived business opportunities), and b)
necessity-driven early-stage entrepreneurs (individuals involved in start-ups only because of the lack of jobs) [
23]. TEA is measured as a dichotomous variable which takes the value”1” if the respondents answer affirmatively to their involvement in early stage entrepreneurial activities, and ”0” otherwise. This approach to measure entrepreneurial activities by a single item proxy has been widely accepted and used by researchers [
3,
5].
Independent Variables
The independent variables gather information on entrepreneurial attitudes: capability to identify opportunity (opport), self-confidence (suskills), fear of failing (fearfail), and networking (knowent).
Entrepreneurial opportunities exist only when people perceive them, and involve the discovery of new means-ends relationships [
12]. Individual perception of opportunities appears to be the main motivating factor triggering entrepreneurial behaviour [
11,
13,
14]. In the present research, opportunity perception (
opport) measures the individuals who consider that in the next six months there will be good opportunities to start a firm, in the area they live. The variable is dichotomous, with the value coded by ”1” for an affirmative answer, and ”0” otherwise.
Self-confidence (suskills) measures the people who consider important having skills, knowledge and experience when starting up a business. The question used for self-confidence assessment was: ”Do you have the knowledge, skills and experience required to start a new business?” The dichotomous variable takes the value ”1” if the respondent’s reply is “yes” to the question, and ”0” otherwise.
Fear of failure (
fearfail) measures a negative emotion resulting from the perception of different threats, being a constraining factor for venture creation [
10]. The individuals were asked whether fear of failure would prevent them from starting a business. If the answer is affirmative, the variable is coded by ”1”, and ”0” otherwise.
Networking proved to be stimulating for business growth, creating new opportunities for engaging in entrepreneurial actions and overcoming liabilities when entering an entrepreneurship [
46,
47]. In the analysis, networking (
knowent) is a dichotomous variable taking the value ”1” if the respondent answers affirmatively to the question: “Do you personally know someone who started a business in the past 2 years?”, and ”0” otherwise.
Control Variables
In the research we use the following control variables:
Gender—a dummy variable with the value ”1” for men and ”0” for women;
Age, the respondent was asked to provide the year of birth. We included in the analysis the respondents aged between 18 and 64.
We test our hypotheses using binary logistic models, and by predicting the likelihood of the effects of several attitudinal factors on entrepreneurial activities, using SPSS 21.0 software. The dichotomous dependent variable takes the value ”1” with a probability of success q (where 1 expresses involvement in TEA), or the value ”0” with the probability 1-q (not involved in entrepreneurial activities). We apply this type of analysis as our dependent variable and most independent variables are dichotomous. Developed as such, the analysis allows us to show the effect of the independent variables on the probability to engage in early stage entrepreneurial activities.
The model takes the following general form,
where:
is
, y is the dependent variable and represents an observable variable indicating the probability of involvement in entrepreneurial activities,
is the intercept,
represents the vector of independent variables and control variables,
represents the coefficients,
is the error term.
The above general form of the model can be rewritten in the following exponential form,
where:
denotes the explanatory and control variables, and
expresses the probability to obtain the value Y = 1 conditioned by the values of
.
We assessed the goodness-of-fit of the models using Omnibus Test, Cox & Snell Pseudo , Nagelkerke , indicating the usefulness of the explanatory variables in predicting the response variable, and −2 Log Likelihood. Also, we presented the value for Wald statistics, testing the significance of individual coefficients, where is given a single degree of freedom. In logistic regression, the odds ratio (Exp(β)) represents the constant effect of a predictor X, in the likelihood that one outcome will occur. The odds ratio is a single summary score of the effect.
In the literature, evidence shows that entrepreneurial activity determinants are sensitive to time and country GDP [
5,
18,
26]. Starting from this theoretical background, we consider that entrepreneurship research must focus on how different periods may shape the pattern of relationships between variables, in different contexts. In order to achieve this aim, several logistic regressions will be performed on two groups of countries (less developed (G1) and high developed economies (G2)), in two different time periods.
Moreover, in order to compare the relative influence of different predictors or independent variables within a logistic regression model, we standardized the predictors based on Menard [
22] approach. The fully standardized logistic regression coefficients can be estimated based on the following equation,
where
b is the sample estimate of the unstandardized logistic regression coefficient and
is the sample standard deviation of the predictor X.
may be described as a fully standardized logistic regression coefficient which incorporates the empirical variation in the dependent variable, as well as the predictors. Given the fact that
is,
where
is the standard deviation of the predicted values of
Y, it results that
=
/R for OLS, and for logistic regression
may be estimated as
/R. Therefore, to divide by
(to obtain the standardized coefficient) is equivalent to multiplying by R/
[
22].
We present the results and interpretation of the empirical analysis in the following section.
4. Results and Interpretation
Table 3 shows descriptive statistics and correlations. As we can see, the correlations among the included variables are not very strong, as all coefficients of independent variables have values ranging from 0.020 to 0.232 in 2007, and from 0.020 to 0.223 in 2014. The relationship between the independent variables and the dependent variables can be distorted, as there is a strong relationship between the items analysed. From the data available in
Table 3, we can see that in our analysis the coefficient values are not very high (this implies that the coefficient values are not 0.90 or higher), meaning that multi-collinearity is not an issue for the analysis. From the total sample, only 5% of the respondents were involved in entrepreneurial activities, 27% found good opportunities in the area where they lived, 45% considered that fear of failure would prevent them being actively involved in TEA, 47% considered that knowledge and skills were important in a business, and 33% considered networking an important factor for entrepreneurial activity.
Figure 1 provides a more in-depth overview of the distribution by country of the percentage of individuals actively involved in entrepreneurial activities in 2007 and 2014. Countries from G1 have a higher percentage of individuals involved in entrepreneurial activities in 2007 (5.85%) and 2014 (8.33%) than those from G2 (4.53% in 2007, and 5.99% in 2014). From G1, countries like Portugal (9%), Spain (7.3) in 2007 and Latvia (12.8) and Romania (11%) in 2014 recorded the highest individual involvement in entrepreneurial activities. From G2, Ireland (7.1%) and Finland (6.8%) recorded the highest values in 2007 and The Netherlands (8.1%) and United Kingdom (7.9%) in 2014 (
Table A1 in
Appendix A).
Opportunity perception (
= 0.129 and
= 0.114), self-confidence in their own skills (
= 0.232 and
= 0.220), and networking (
= 0.145 and
= 0.193) positively and significantly relate to entrepreneurial activities. As expected, fear of failure (
= −0.081 and
= −0.089) negatively relates to our dependent variable. With respect to age and gender, both control variables negatively and significantly relate to entrepreneurial activities. Males and younger respondents are more likely to be involved in entrepreneurial activities than females, and older respondents have a higher tendency than younger respondents to engage in entrepreneurship (
Table 3).
The analysis shows the results of several logistic regression analyses computed for 2007 and 2014, where we introduced different variables in subsequent steps of the logit models, following Liñán et al. [
52] approach (
Table 4,
Table 5,
Table 6 and
Table 7). In step 1, we included the control variables used in the analysis (namely, gender and age); in step 2, we added the individual characteristics (opportunity perception, fear of failure, self-confidence in their own skills and knowledge, and networking), and in step 3, the interaction effects. This approach is in agreement with the previous works on determinants of entrepreneurial intentions (see [
52]), adapted though for measuring entrepreneurial activities.
Table 4,
Table 5,
Table 6 and
Table 7 present the regression coefficients, significance level, exponential ratio, and Ward statistics, shown for each independent and control variable included in the study, as well as for the terms of interaction. As we mentioned previously, we used data retrieved from the Adult Population Survey, Global Entrepreneurship Monitor (GEM) annual survey for 2007 and 2014 and a total of 18 European countries.
Three logistic regression analyses were computed for 2007 for less developed economies to estimate the influence of individual characteristics on TEA.
Model (1) presents the control variables used in the analysis. The results reveal that gender and age show a significant and negative influence on early stage entrepreneurial activity. The negative effect of gender suggest that men are more likely to engage in early-stage entrepreneurial activities than women. Similarly, the age control variable showed a negative effect on the dependent variable, indicating that younger individuals have a higher likelihood of engaging in TEA than older respondents.
Model (2) introduces the main effects of perceptual factors (
Table 4). In model (2) self-confidence has the highest odds ratio (8.781), suggesting that this factor has the greatest influence on the likelihood of engaging in early-stage entrepreneurial activity. This indicates that individuals who are confident in their own abilities, skills and knowledge have a higher propensity to engage in TEA than the rest of the population. A similar effect on TEA had the scores for networking, taking into consideration the constant variation of the other variables included in the model. The odds ratio was 1.504 for an additional unit to the score of networking when opportunity perception, self-confidence and fear of failure remain constant. The opportunity perception has a positive and significant effect on TEA, suggesting that respondents who perceive entrepreneurial opportunities were more likely to engage in early-stage entrepreneurial activities (
Table 4). The probability to engage in TEA increases when the score for opportunity perception increases by one unit and the scores for fear of failure, self-confidence and networking remain constant. The odds ratio was 1.478 for an additional unit to the score of opportunity perception when the scores for fear of failure, self-confidence and networking remain constant. On the other hand, perceiving a higher risk of failure contributes to the decreasing entrepreneurial activities. The negative sign (
β = −0.338,
p < 0.01) indicates that the greater the fear of failure, the less likely to engage in entrepreneurial activity.
Model (3) tests the interaction effects between opportunity perception and gender, fear of failure, self-confidence, and networking, respectively. Using the variable of gender as a moderator, we can see if the influence of these variables differs between males and females. The interaction effect reveals that gender has a significant influence on the interaction with the fear of failure. The odds ratio was 17.377 when the score of fear of failure increases by one unit and gender is 1. We could not find any significant interaction effects between other perceptual factors and gender.
Other three logistic regression analyses were computed for 2007 for highly developed economies (G2). Similarly, model (4) presents the control variables for the second group of countries. The results show that both men and young people have a higher inclination towards entrepreneurial activity. Model (5) introduces the main effects of the independent variables for the second group of countries. The findings show that unlike G1, in G2 self-efficacy and opportunity perception have the highest odds ratios (6.551, and 1.933, respectively). Self-confidence is positively associated with the probability to engage in early-stage entrepreneurial activity, suggesting that the knowledge, skills and experience required for starting a new business has a positive effect on TEA. The odds ratio was 1.933 for an additional unit to the score of opportunity perception when the scores for fear of failure, self-confidence and networking remain constant, suggesting that individuals who recognize opportunities are 1.933 times more likely to engage in TEA than those who do not. Similarly, having a good entrepreneurial network increases the likelihood of engaging in entrepreneurial activity (
p < 0.01; odds ratio 1.898). A negative effect on the likelihood of engaging in entrepreneurial activity appeared from the scores for fear of failure (
p < 0.01), showing an inverse relationship between this individual characteristic and the propensity to engage in TEA (
Table 5).
Model (6) showed that all the interactions between independent variables, namely; opportunity perception, fear of failure, self-confidence, networking, and gender (G2). The first interaction showed a positive but insignificant influence on the probability to engage in entrepreneurial activity. The following interactions have a significant effect, revealing differences in individual perceptions between males and females. The interaction effect reveals that the increase by one unit of the variable self-confidence and the fact that gender is 1, while all the other variables remain constant, found an increase in the probability of involving TEA. We found significant interactions between gender and fear of failure with an odds ratio of 1.182, and networking and gender with an odds ratio of 1.211.
Table 6 replicated the analysis to show varying effects of individual characteristics on TEA across another time period. Gender shows a consistently negative and significant effect for all groups and periods of time considered. Age recorded similar results, which was negatively and statistically significant in both periods, validating
Hypotheses 1 and
2.
In terms of perceptual variables, self-confidence shows a consistently positive and significant effect across the two time periods and groups. The variable has the highest odds ratio in the logistic regression equations for all the models developed (recording the highest odds ratio for G1 in 2007
= 8.781 and
= 5.784 in 2014;
= 6.551 in 2007 and
= 6.816 in 2014, for G2, respectively), thereby, validating
Hypothesis 5. Opportunity perception showed a positive and significant effect on TEA for all groups of countries in both 2007 and 2014, suggesting that individuals who perceived entrepreneurial opportunities are more likely to engage in early-stage entrepreneurial activities. The odds ratios for opportunity perception is lower when compared to self-efficacy
= 1.478;
= 1.933;
= 1.489, and
= 1.524). The coefficient shows that opportunity perception is a strong predictor for developed countries in 2007, but its importance decreased in 2014. However, this variable positively and significantly influenced TEA for both groups of countries, validating
Hypothesis 3 (
Table 4,
Table 5,
Table 6 and
Table 7).
As we can see, fear of failure maintains the negative sign in follow-up models, showing that this variable contributes to the decrease in the probability of engagement in TEA. Thus, the hypothesis according to which the fear of failure is positively and significantly associated with the early-stage entrepreneurial activity is rejected (
H4). For the individuals from G1, the odds ratio was not that high (
= 0.713 and
= 0.683), indicating that this perceptual variable had the lowest effect on the likelihood of involvement in TEA in both periods. Although, the odds ratio increased in 2014 for G2, it still remained the lowest among the individual-perception factors (see models (8), (9), (11) and (12) (
Table 7). In 2014 the odds ratio for networking almost doubled, suggesting the strong effect of this variable on the propensity to engage in early-stage entrepreneurial activity (
= 1.504 to
= 2.869;
= 1.898 to
= 3.246). In 2007, this variable alongside self-confidence was one of the strongest predictors of TEA for countries from G1, and in 2014 for both groups of countries, showing that networking exerts a positive and significant influence on early-stage entrepreneurial activity,
validating H6. The only significant interactions for G1 and G2 are between
networking and gender (odds ratio of 1.239) and
opportunity perception and gender (odds ratio of 1.218) (
Table 4,
Table 5,
Table 6 and
Table 7). For highly developed economies, we could not find any other interaction effects between other independent variables and gender (
Table 6 and
Table 7). The odds ratios and standardized coefficients show that for less developed countries from G1, self-confidence and networking were the strongest predictors both in 2007 and 20014. In 2007, for high developed economies belonging to G2 the strongest predictors were opportunity perception and self-confidence. However, in 2014 the highest odds ratios were obtained for self-confidence and networking (
Table 4,
Table 5,
Table 6 and
Table 7).
Table 8The goodness-of-fit statistics indicates the results for Omnibus Test, Cox & Snell
, Nagelkerke
, and −2 Log Likelihood.
The Omnibus tests of model coefficients are significant (p < 0.05), confirming the causal relationship of the proposed logit models and acceptance of the hypothesis that coefficients are different from zero. Although, Cox & Snell , Nagelkerke indicate that the variables considered explain only a small fraction of the variance in entrepreneurial activity, their values increased from one model to the other.
-2 Log likelihood values decrease from Model 1 to 3, 4–6, 7–9, and 10–12 providing additional support for the adequacy of the models.
5. Conclusions and Study Limitations
There are numerous theoretical and empirical approaches investigating the importance of entrepreneurship in promoting the development process [
53,
54]. Different approaches to entrepreneurship mainly present three sets of research questions: Why, how, when or where some people and not others discover and exploit different entrepreneurial opportunities [
12]. In fact, the cognitive framework attempts to explain the main aspects of the discovery-exploitation process, not only at an individual level, but also at an aggregate level [
52]. Starting from this perspective, our main focus was on understanding the role of different perceptual factors on the probability to engage in entrepreneurial activities.
The study used logit regression models to estimate the probability that, individuals belonging to a specific group have a high susceptibility to several cognitive biases. We chose the logit regression analysis as the vast majority of variables were dichotomous, including our dependent variable. This type of analysis allowed us to estimate the effect of several independent variables on the probability of engagement in entrepreneurial activities. Specifically, the main objective of the paper is to examine the effects of demographic variables (gender and age), perceptual factors (capability to identify opportunity, having skills, knowledge and experience to start up businesses, fear of failing, and knowing other entrepreneurs), and the interaction effects between them and gender on the likelihood of engaging in entrepreneurial activities across eighteen European Union member states, based on the statistics gathered by the GEM (individual data).
Moreover, the literature shows that entrepreneurial activity determinants are sensitive to time and country GDP [
5,
18,
19]. Starting from this theoretical framework, we consider that entrepreneurship research must focus on how different periods may shape the pattern of relationships between variables, in different contexts. To that aim, our study examines two different moments in time, namely 2007 and 2014, in order to capture the fluctuations in entrepreneurial activities and the effects of different perceptual factors, in less and highly developed European member states. We measured the level of entrepreneurial activity using the total early-stage entrepreneurial activity (TEA) as a proxy, including individuals aged between 18 to 64, who were either actively trying to start a new business, or managing a business that was less than three-and-a half years old. Our results show that across the countries analysed, TEA tends to be the highest among factor-driven economies and declines in economies with higher GDP per capita. From the sample analysed, the mean average of TEA for less developed economies (G1) was 5.85% in 2007 and 7.19% in 2014, while the average mean of TEA was 4.53 in 2007 and 6.74% in 2014 for economies with higher GDP per capita (G2).
In the empirical analysis, we followed Liñán et al. [
52] approach, however adapted for measuring entrepreneurial activities. In subsequent steps, we first included the control variables in the analysis, then we added the individual characteristics (opportunity perception, fear of failure, self-confidence in their own skills, knowledge, and networking), and in the last step, the interaction effects between perceptual factors and gender.
In agreement with our hypotheses, the results reveal that gender is an important determinant which affects the probability to engage in entrepreneurial activities, men being more inclined to engage in entrepreneurship than females. Similarly, other studies reached to the same outcome, i.e., the constant differences between men and women in developing entrepreneurial activity are due to gender characterisation [
5,
6,
55,
56]. For instance, according to Brush et al. [
57] gender parity in perceived capabilities and economic participation significantly influence the relative parity to early-stage entrepreneurship. Other authors’ studies show similar results. In fact, Özdemir and Karadeniz [
6] demonstrate that, men with a higher income and education level, who are confident in themselves, have the ability to identify business opportunities, and are in contact with other entrepreneurs are twice more likely to engage in entrepreneurial activities than women.
Further findings are also consistent with the literature, consequently younger individuals are more likely to engage in entrepreneurial activities than older people [
5,
35]. The study conducted by Colovic at al. [
58] shows that third-age entrepreneurs tend to lag behind their younger counter parts in technology adoption and innovation. Further evidence underlines that entrepreneurship represents the main driver for development and economic recovery and is generally associated with young individuals [
59].
Besides gender and age, the analysis shows the importance of perceptual factors for entrepreneurial activity in EU-18 member states. The results show that three perceptual factors, namely self-confidence, opportunity perception, and networking have a positive and significant influence on early-stage entrepreneurial activities in both, groups of countries and time periods. As anticipated, the fear of failure has a negative and significant effect on entrepreneurial activity. These results underline that, despite the recent economic and financial crisis, individuals who possess these attributes are more likely to engage in new venture creation.
We can see that in 2007, the strongest determinants were self-confidence and networking for less advanced economies, while opportunity recognition and self-confidence had the highest odds ratio for more developed economies. In 2014, although opportunity recognition remained an important determinant for TEA, this influence was not as high as anticipated. This might be the case that individuals with high early-stage entrepreneurial activity engagement perceived the recognition of opportunities as a normal situation. As Liñán et al. [
52] stated, “it is possible too that the presence of cognitive biases is exerting an influence on perceptions about economic opportunities”. These results are in agreement with other author’s findings, who consider that these entrepreneurial attributes are important assets when starting a business [
5,
8,
13,
46]. On the other hand, the analysis on fear of failure reveals a negative influence on early-stage entrepreneurial activities, showing that individuals who perceive this negative emotion resulting from the perception of different threats are less likely to engage in early entrepreneurial activities. We validated these results for both groups of countries and time periods. The findings are in agreement with previous studies that consider fear of failure is a constraining factor for venture creation [
10].
In this study we also tested the interaction effects between opportunity perception and gender, fear of failure, self-confidence, and networking, respectively, for both groups and time periods. Using the variable gender as a moderator, we show that the influence of these variables differs between males and females. In 2007, the interaction effects revealed that, for less developed economies, gender had a significant influence on the interaction with the fear of failure. The odds ratio was 17.377 when the score of fear of failure increased by one unit and gender was 1. We did not find any interaction effect between other independent variables and gender. For high developed economies, the interactions showed a positive and significant effect between three perceptual factors and gender. The interaction effect reveals that the increase by one unit of the variable self-confidence and the fact that gender is 1, while all the other variables remain constant, determined an increase in the probability to involve in TEA. We found significant interactions between gender and fear of failure with an odds ratio of 1.182 and networking and gender with an odds ratio of 1.211. In 2014, the only significant interaction for both groups of countries are between networking and gender (odds ratio of 1.239), and opportunity perception and gender (odds ratio of 1.218).
We evaluated the relative importance of predictors in logistic models using fully standardised logistic regression coefficient calculated using Equations (3) and (4) [
22,
60].
For our analysis, although we cannot directly calculate the standard deviation for the observed values of logit (TEA), we estimated the standard deviation indirectly using the predicted values of logit (TEA) and the explained variance,
[
60]. In order to rank the magnitude of the influence of the predictors on the dependent variable, we applied Menard [
22] and Menard [
60] methodology to estimate fully standardized logistic regression coefficients. According to the Equations (3) and (4), we estimated the standard logistic regression coefficients (
Table 9). Using the data from the estimates we are able to predict whether a dichotomous predictor (TEA) is more or less strongly related to the outcome.
When the analysis includes only the control variables, estimates show that in both groups and periods gender was the strongest predictor, followed by age (see (1), (4), (7) and (10),
Table 7).
In 2007, in less and high developed countries, having skills, knowledge and experience to start up businesses have the strongest relationship with TEA, while gender has the weakest relationship with TEA. In less developed countries, knowing other entrepreneurs (
= 0.044) has the second importance in predicting entrepreneurial activity, followed by opportunity perception (
= 0.041), age (
= 0.040), and fear of failing (
= −0.037). Contrary to the situation from G1 countries, in high developed countries the capability to identify opportunity (
= 0.074) has the second importance in predicting TEA, followed by knowing other entrepreneurs (
= 0.069), fear of failing (
= −0.053) and age (
= −0.043) (see (2) and (5),
Table 9).
In subsequent models, we also added the interaction effects between individual attributes and gender. In 2007, in both groups, G1 and G2, self-confidence has the strongest relationship with TEA, while the standardized interaction between opportunity perception and gender has the weakest relationship with TEA. In less developed countries, gender (
= −0.054) is the second important predictor of TEA, followed by perception of opportunities (
= 0.041), networking (
= 0.038), age (
= −0.029), fear of failing (
= −0.029), and the standardized interactions between gender and each of the variables having skills, knowledge and experience to start up businesses (
= 0.019), fear of failing (
= −0.016) and networking (
= 0.010). Contrary to the situation from G1 countries, in G2 countries perception of opportunities (
= 0.071) has the second importance in predicting TEA, followed by gender (
= −0.06), fear of failing (
= −0.060), networking (
= 0.059), age (
= −0.042), and the interactions between gender and each of the variables having skills, knowledge and experience to start up businesses (
= 0.020), networking (
= 0.015) and fear of failing (
= 0.014) (see (3) and (6),
Table 9).
Similarly, in 2014, in both less and high developed countries, having skills, knowledge and experience to start up businesses had the strongest relationship with TEA. However, in both groups of countries, the second importance in predicting entrepreneurial activity was networking. In less developed countries, age (
= −0.049) has the third importance in predicting TEA, followed by opportunity perception (
= 0.065), fear of failing (
= −0.045) and gender
= −0.029). In high developed countries, fear of failure
= −0.096) ranked third in predicting entrepreneurial activities, followed by age
= −0.070), opportunity perception
= 0.054) and gender
= −0.029) (see (8) and (11),
Table 9). The following models included also the interactions effects for both groups. In 2014, for both groups of countries, having skills, knowledge and experience to start up businesses and networking had the strongest importance in predicting TEA. In 2014, for less developed economies, the follow up variables were gender
= −0.052), age
= −0.048), opportunity perception
= 0.043), fear of failing
= −0.043) and the interactions between gender and each perceptual factor. For high developed economies, the third importance in predicting early-stage entrepreneurial activity was fear of failing
= −0.094), followed by age
= −0.070), then opportunity perception
= 0.044), gender
= −0.039), and the interactions between gender and each perceptual factor (see (9) and (12),
Table 9).
Evidence suggests that a high level of economic development (e.g., high national income per capita) exerts a positive influence on the creation of new business [
61]. Nevertheless, according to our findings and based on the data provided by GEM, we found that less developed countries recorded higher rates of new businesses than most high developed economies. A possible explanation for this scenario is the fact that in less developed economies, individuals engage in entrepreneurial activities as a necessity (lack of jobs, high unemployment rates). As Carlsson [
62] stated, the start-up rate increased during the seventies and eighties, only after unemployment becomes a serious issue.
Our analysis shows that in both less and high developed economies, most individuals start an entrepreneurial activity because of self-confidence and networking. Clearly, one’s cognitive perception about skills, knowledge and abilities has an important effect on the efforts undertaken and how to persist in these efforts when potential obstacles arise. Individuals are more likely to engage in venture creation and choose entrepreneurship as a career path if they have the confidence to be more successful in their choice, based on their skills and knowledge [
63]. Similarly, an individual’s position in a social structure may influence the attitudes, behaviour and outcomes of the individuals occupying those positions. Translated into entrepreneurship, this implies that one’s personal and organisational network may influence the actor’s propensity to engage in venture creation. Outcomes show that for both groups of countries, networking have mostly the second strongest impact on an individual’s decision to engage in entrepreneurial activity. In other words, networking represents one of the most important drives in choosing entrepreneurship as a career path. It is clear that entrepreneurs need to establish connections to identify an opportunity and find the necessary resources to begin operations [
64].
Self-confidence and networking, other two perceptual factors, have an important impact on the entrepreneurial activity. Opportunity perception implies not only individuals’ opportunity recognition, but also their fit, i.e., the individual opportunity nexus [
29]. Opportunities exist only when people perceive them, and involve the discovery of new means-ends relationships. In agreement with our findings, other authors’ studies underline that individual perception of opportunities appears to be one of the main motivating factors triggering entrepreneurial behaviours [
11,
13,
14]. Fear of failure, unlike the previous perceptual factor, measures a negative emotion resulting from the perception of different threats, and it is considered an inhibitor of venture creation [
10]. In our analysis, for low income countries, both opportunity perception and fear of failure have similar values. Fear of failure is stronger in highly developed economies, i.e., the individuals who perceive a market opportunity for opening a business also express a higher fear of failure. This perceptual factor acts as a barrier in venture creation, and has a negative impact on TEA for both groups of countries. On the other hand, opportunity perception has stronger values for high developed economies, although this influence is not as high as anticipated. However, the perception of entrepreneurial opportunities could act as a precipitating factor, as it can reinforce other individual perceptions [
52].
Of course, this study has several limitations. Some mainly relate to the data used from GEM Adult Population Survey, including mostly “yes” and “no” answers, which limits the use of a more accurate statistical analysis. Secondly, the number of items related to entrepreneurial perception and activities is small, and includes mostly one item measures. Future studies embracing larger datasets by including more countries could offer more robust data and enable us to create a more accurate image of the main determinants of entrepreneurial activities before, during, and after the recent economic crisis.
Also, our analysis uses Adult Population Survey for 18 European member states in 2007 and 2014. A longitudinal study could capture a clearer image of TEA main determinants. Moreover, we only used a few determinants of the entrepreneurial activities. Thus, a possible future direction of research would be to study additional economic and non-economic factors (cultural indicators; other demographic indicators—environment of provenance) which may influence TEA. The national culture dimensions, i.e., power distance, individualism/collectivism, masculinity/femininity, long/short term orientation, uncertainty avoidance and indulgence/restraint) [
40] might provide a better understanding of the multifaceted relationship between economic development, country competitiveness and culture, and their impact on opportunity-driven and necessity-driven early-stage entrepreneurs. For future research, it may well be desirable to consider other European countries, as well as additional essential timelines to describe the evolution of these factors over time. In addition, it would be interesting to carry out a longitudinal study on the factors that could affect TEA. Likewise, applying qualitative methods to study more in depth this phenomen on might contribute to the improvement of the understanding of the factors and of their impact on opportunity-driven and necessity-driven early-stage entrepreneurs in different industries and different countries.