Educational heterogeneity of the founding team of innovative start-ups: confirmations and denials
- Open Access
- 10.08.2024
Erschienen in:
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by (Link öffnet in neuem Fenster)
Abstract
Introduction
Innovation is always more central in business and sustainable development, and policymakers support the establishment of innovative start-ups (Audretsch et al., 2020; Modina et al., 2023). Moreover, although recent studies suggest that innovation pays and can contribute to growth and economic development (Berger & Köhn, 2020; Fiorentino et al., 2020), several empirical investigations show that start-ups perform poorly and/or do not survive long (Gonzalez, 2017; Hyytinen et al., 2015). Consequently, analyzing the value drivers affecting the growth of innovative start-ups (ISUs) is essential to understand their potential societal benefits, given these companies’ capacity to reactivate and invigorate industries (Matricano, 2020). For these reasons, research on ISUs success factors has experienced exponential growth (Colombelli et al., 2016; Ferretti et al., 2020).
Most of the existing contributions on ISUs tend to concentrate on the characteristics of the external context or external environment to explain ISUs performance (Cavallo et al., 2020; Mustar et al., 2008). However, analyzing the internal drivers is equally important and the features of the founding teams have received increasing attention in the entrepreneurial literature (Kamm et al., 1990; Kollmann et al., 2017). A specific strand of research focuses on founding team composition (Colombo & Grilli, 2005; Das et al., 2021; Pinelli et al., 2020; Ratzinger et al., 2018) in terms of age and background (Del Bosco et al., 2021; Franco et al., 2021; Seebeck & Wolter, 2022), previous experience (Furlan, 2019; Hashai & Zahra, 2022; Honoré, 2022), levels of cohesion (Bjornali et al., 2016; Ensley et al., 2002; Knapp et al., 2015) and team cognitive features (Hoogendoorn et al., 2017; Simmers, 2021). While the social categorization perspective argues that homogeneous entrepreneurial teams are more effective than heterogeneous ones (van Knippenberg & Schippers, 2007); the alternative information processing perspective posits a positive effect of founders heterogeneity (Finkelstein & Hambrick, 1996; Hambrick & Mason, 1984). Among the most controversial elements of analysis, emerges, on one side, the impact of team heterogeneity on start-up performance and, on the other side, the educational background of founders (Pinelli et al., 2020; Ratzinger et al., 2018; Visintin & Pittino, 2014). Studies from four main literature streams– upper echelon theory, cognitive and social psychology theories, resource-based view, and human capital theory– provide starting evidence on the effect of founding team heterogeneity on start-up performance. Unfortunately, previous investigations have pointed out contrasting results regarding both the relevance and direction of the impact on performance (Ensley et al., 1998; Pinelli et al., 2020). Furthermore, while specific studies directly focused on ISUs are lacking, analyzing the distinctive features of founding teams’ heterogeneity can provide valuable insights into the key determinants of ISUs’ early performance. Furthermore, research on an experience-based variable, such as education, can contribute to entrepreneurial literature.
Anzeige
This paper aims to fill the latter gap by contributing to the literature on the value drivers of ISUs adopting a demographic approach (Hambrick, 2007; Hambrick & Mason, 1984). Identifying and analyzing the factors that favor or limit success is essential to leverage the growth of ISUs and to steer policymakers’ decisions. Notably, in this study, the following research question is posed: does the educational heterogeneity of the founding teams impact the early performance of ISUs?
Based on a panel of Italian innovative start-ups (defined according to L. 221/2012), the relationship between founding teams’ heterogeneity is investigated through a growth function estimated by Ordinary Least Squares (OLS) and unconditional quantile regression models. An original and ad hoc dataset that merges data from the Italian Ministry of Economic Development and financial information extracted from AIDA is used.
The article is structured as follows. Section 2 scrutinizes previous literature. Section 3 focuses on hypotheses development. Section 4 details the methodology applied. Section 5 offers the results, while Section 6 summarizes the findings’ discussion and the main conclusions.
Theoretical background
Start-ups are ventures at the early stage that are increasingly recognized as a relevant pillar of the modern economy (Eliakis et al., 2020; Modina et al., 2023). Among new ventures, ISUs are a relatively novel concept. They are characterized as small-medium enterprises, young (less than eight years of activity) with great potential to exploit innovations (with innovative products, processes, or projects) and create value for society (Cavallo et al., 2021).
Anzeige
Studies have suggested that one of the most relevant factors affecting the growth and success of start-ups is the quality of the entrepreneurial team (Eisenhardt & Schoonhoven, 1990; Hambrick & Mason, 1984; Heirman & Clarysse, 2004). Specifically, the founding team concept was developed by Kamm et al. (1990), who emphasized the involvement of people in new venture creation. Based on the literature review, which shows several conceptualizations and descriptions of founding teams, the founding team can be defined as an “umbrella concept” including employees, founders, and top management (Das et al., 2021). Subsequently, an extensive literature stream has suggested that the founding team plays a critical role in the success of start-ups (Clarkin & Rosa, 2005; Cooper & Daily, 1997) and can influence the performance of new ventures (Amason et al., 2006; Ensley & Pearce, 2001).
The entrepreneurial team’s investigation is frequently linked with team diversity and heterogeneity. The heterogeneity can be associated with founding teams regarding the variance of founders’ characteristics (Harrison & Klein, 2007; Jin et al., 2017). Team heterogeneity is one of the strands of research investigated through different perspectives. Heterogeneity, indeed, can increase the set of skills and abilities of the organization, improving creativity, innovativeness, and networking. At the same time, it could generate conflicts among teammates, resulting in poor performance (Chowdhury, 2005; Franco et al., 2021; Schoss et al., 2022).
Scholars employ the Upper Echelons Theory (UET), along with theories from cognitive and social psychology, the resource-based view, and human capital theory, to analyze how heterogeneity within founding teams affects startup performance.
Upper echelon theory considers the association of entrepreneurial team characteristics and behaviors with performance, providing a valuable lens to investigate the impact of founder teams on firm performance literature (Hambrick, 2007; Hambrick & Mason, 1984). The upper echelons argument emphasizes cognitive demographic attributes of the entrepreneurial team such as age, education, experience, or background. Most research focuses on relationships between inputs (e.g., founders’ characteristics) and firm-level performance (e.g., revenue growth) (Barrick et al., 2007; Ilgen et al., 2005; Smith et al., 1994).
Cognitive and social psychology theories advance the idea that team diversity improves team performance (Phillips & O’Reilly, 1998), suggesting that distributional differences can serve as indicators of available knowledge and differing perspectives. A founding team that is more diverse in demographic variables related to a given task may be more successful than a homogeneous team because the former team can draw on a more excellent pool of knowledge and different perspectives (Chattopadhyay, 1999; Hambrick & Mason, 1984).
From the resource-based view (Hamel & Prahalad, 1994), heterogeneity in founders’ background can impact start-up performance, as it provides a diverse stock of knowledge, capabilities, and expertise (Milliken & Martins, 1996; Randel & Jaussi, 2003). Since the competitive advantage of start-ups depends on the exploitation of founders’ capabilities (Chowdhury, 2005; Colombo & Grilli, 2010), a large number of studies focused on the relationship between the performance of start-ups and the characteristics of the entrepreneurial team (Visintin & Pittino, 2014).
Based on human capital theory (Gimmon & Levie, 2010), founding teams constitute an essential success determinant of young firms (Box & Larsson Segerlind, 2018; Van Der Sluis et al., 2008) or, as noted by Rogers and Larsen (1984), human capital can be the leading cause of corporate failure. Studies suggest that not only the level of human capital but also the variance across founders can affect performance (Hambrick et al., 1996).
Literature offers contributions that examine the relationship between different aspects of the founding team and firm outcome, reaching mixed and still uncertain results (Bell et al., 2011; Fuel et al., 2022). The field of education is an “achieved’ characteristic” (Forbes et al., 2006, p. 232) and heterogeneity in education is a measure of variety, the “composition of differences […] among team members” (Harrison & Klein, 2007, p. 1203). Education is a well-studied founding team characteristic in the heterogeneity existing literature, and its importance for start-ups is well-documented (Bell et al., 2011; Harrison & Klein, 2007; Horwitz & Horwitz, 2007).
In line with the literature, education can be considered a task-related characteristic (Pelled, 1996; Simons et al., 1999; Webber & Donahue, 2001). Instead, the qualification (education degree) can be related to status concerns among founders (Harrison & Klein, 2007). For instance, Henneke and Lüthje (2007) explore the relationship between the level of interdisciplinary heterogeneity in entrepreneurial teams and the level of product innovativeness in high-tech ventures. Aspelund et al. (2005) test if a greater degree of heterogeneity in the functional background within the founding team leads to a greater probability of survival for new technology-based firms; Franco et al. (2021) try to understand if the degree of educational heterogeneity favors founder teams in funds raising.
Nevertheless, while research has progressed by investigating team diversity and introducing studies on education heterogeneity, the question of whether and how to optimize founder team composition remains. Previous research has yet to find any beneficial effects of founding team heterogeneity on the success of start-ups. Most studies find a weak positive but often a statistically insignificant link between skill heterogeneity and performance (Bell et al., 2011). Overall, empirical evidence supports arguments for and against team heterogeneity since results are pretty mixed. Moreover, previous studies have not analyzed the relationship between founding team composition and performance in ISUs.
Hypotheses formulation
Therefore, the analysis focuses on the educational field (horizontal) heterogeneity and educational level (vertical) heterogeneity, which have received little attention from academics so far by adopting a static approach and concentrating on what happens after the legal establishment of a company (foundation stage). The first refers to the mix of subject areas (e.g., management, humanities, hard science, and social science). In contrast, the second refers to the combination of the educational level obtained (e.g., high school, degree, Ph.D.). As noted by Pinelli et al. (2020), the effects of ‘founders’ education on start-ups’ future success require going beyond how long a founder studied a specific subject; it also requires considering the extent to which a group of cofounders studied different disciplines and with different level of education achieved.
As Tasheva and Hillman (2019) suggested, education should be mainly related to looking at the human capital as a source of diversity. Human capital derives from experience and knowledge gained among others in the educational setting and results from individuals’ investments in specific education (Johnson et al., 2014). Moreover, the education-related individuals’ cognitive schemas affect decision-making (Hambrick & Mason, 1984). Indeed, the human capital theory is adopted as the favored theoretical lens for hypothesis formulation.
Educational field (horizontal) heterogeneity
Educational field heterogeneity in the top management team is an interesting field of study approached by various authors (Cannella et al., 2008; Kim & Rasheed, 2014; Murray, 1989; Pinelli et al., 2020). This horizontal heterogeneity can affect ISU’s performance. The human capital literature assumes that variety or diversity (Harrison & Klein, 2007) can improve performance overall (Homberg & Bui, 2013).
The performance can be partially predicted by considering the background characteristics of its top management team. The team’s diverse educational backgrounds serve as a valuable source of different perspectives, enriching the collective knowledge and enhancing their ability to process information. Additionally, this diversity stimulates constructive debates and contributes to effective decision-making processes (Hambrick & Mason, 1984).
Overall, as noted by Cai et al. (2013), previous studies offer mixed results. Focusing on the ability to acquire funds, Franco et al. (2021) show that highly educated founding teams specialized in analogous fields reach the goal similarly to teams with a lower level of education where various backgrounds coexist. Instead, Cui et al. (2019), referring to a sample of 172 Chinese listed companies in the IT industry, state that academic background (that is related to the learning experiences, research projects, published papers, published monographs, major topics of participation, and the formulation of participation standards) heterogeneity is negatively correlated with financial performance.
Moreover, as suggested by previous empirical evidence (Cohen & Levinthal, 1990; Fiorentino et al., 2022; Sapienza et al., 2004), the effect may be relevant up to a certain point but be less so beyond certain levels of heterogeneity, configuring a U-shaped function.
Consequently, the following hypothesis is formulated:
H1:
The educational field (horizontal) heterogeneity among founders affects the ISU’s performance.
Educational level (vertical) heterogeneity
On the other hand, in founding teams, the horizontal heterogeneity among members matches with vertical heterogeneity that differences in education levels can measure (Franco et al., 2021; Millán et al., 2014; Ratzinger et al., 2018) and hence a higher educational level could be linked to better firm performance, thanks to the capacity of knowledge to generate competitive advantage. Furthermore, as described by some authors (Pegkas & Tsamadias, 2014; Pinelli et al., 2020), individuals who invested more in their studies, gaining a higher educational level, likely have more substantial incentives to pursue business success to repay their investment best; additionally, a higher educational level improves the entrepreneurial alertness, helping to identify less visible business opportunities and then maximize their return (Cooper et al., 1997).
With these premises, the educational level is one of the observable characteristics that researchers investigate to verify the effect on start-up performance (Franco et al., 2021; Gimmon & Levie, 2010; Maidique, 1985; Pinelli et al., 2020; Ratzinger et al., 2018; Zhang et al., 2017), testing if the higher educational level of founding teams is related to better entrepreneurial results such as survival, growth rate and earnings.
However, the overwhelming majority of studies on team heterogeneity have focused on horizontal differences, though there are few clear and consistent themes in empirical findings within vertical heterogeneity and even less for educational vertical heterogeneity in ISUs. Among the various studies, Ratzinger et al. (2018)– using a sample of 4,953 digital start-ups– noted that increased formal business and technical education within founding teams grows the probability of reaching investment milestones. Similar results are achieved by Franco et al. (2021), analyzing a sample of 1,078 start-ups in the United States: high levels of education are associated with better performance of firms. The study of Zhang et al. (2017) points out that educational level heterogeneity negatively affects the performance of technology start-ups. Similar results emerge from the exploration of Subramanian et al. (2016) that shows a negative innovation performance (measured as the number of patent applications) when team members have diverse educational levels. Indeed, the U-shaped effect is also associated in previous literature with horizontal heterogeneity (Fiorentino et al., 2022; Visintin & Pittino, 2014).
Consequently, the following hypothesis is formulated:
H2:
The educational level (vertical) heterogeneity among founders affects the ISU’s performance.
The effect of founders’ level of education on ISU’s performance
Moreover, entrepreneurial team studies also highlight that the relationship between vertical and horizontal heterogeneity can impact ISU’s performance because vertical heterogeneity can increase the “distance” among founders (Franco et al., 2021; Pinelli et al., 2020). The mediating or moderating effect of vertical heterogeneity can help explain the mixed effects of horizontal heterogeneity on performance (Bunderson & Van der Vegt, 2018; Finkelstein, 1992). For instance, Visintin and Pittino (2014) suggest that vertical heterogeneity lessens the positive effect of horizontal heterogeneity.
Indeed, it can be helpful to overcome some limitations of previous research in entrepreneurship, focused on relationships between team member characteristics and firm performance, by also investigating critical mediating mechanisms and moderating factors (Klotz et al., 2014). Pinelli et al. (2020) suggeste that the joint presence of horizontal and vertical heterogeneity negatively moderated the relationship between educational level and performance.
Consequently, the following hypotheses are formulated:
H3a:
The founders’ level of education affects the ISU’s performance.
H3b:
The founders’ level of education moderates the relation between the founders’ educational heterogeneity (vertical and horizontal) and the ISU’s performance.
Data and methods
Data collection process
The sample refers to innovative Italian start-ups. The definition of innovative start-up has been introduced in Italy with law 221/2012; in particular, to be defined as “innovative start-up” and hence have the chance to be enrolled in the special section of the Companies Registry, entities must meet several requirements among these: not having more than 60 months of activity, not being listed, at least 15% of the company’s expenses related to R&D activities. A multi-stage data collection process is carried out to analyze the impact of educational heterogeneity, resulting in an innovative dataset that combines multiple administrative and non-administrative sources.
In the first stage of this process, data from the Italian Register of Innovative Start-ups1 are merged with data extracted from the AIDA Bureau Van Dyck dataset, which provides comprehensive information on the economic, financial, demographic, and commercial aspects of all capital companies operating in Italy. The second stage of the process aimed to gather data on the educational qualifications of the members of the founding teams for each innovative start-up. This information is not available in any downloadable dataset but can be accessed from the web pages of the ISU registry site.
To capture this data, a web-scraping approach on the search page of the Italian Register of Innovative Start-ups website is used to extract data on each founder’s level and field of study. This allowed the organization of the information into a statistical dataset merged with the dataset obtained in the first phase of the data collection process2.
The result is a dataset comprising information on 1,201 innovative start-ups, including both financial data and details on the composition of their founding teams.
Variables
The dependent variable is the revenue growth rate. Indeed, as highlighted by Klotz et al. (2014), one of the three outcome variables mostly applied is growth in sales. Although growth has a complex and multidimensional nature that is difficult to address using any single measure, revenues from sales are one of the most appropriate and diffused financial measures for start-ups (Autio et al., 2000; Modina et al., 2023). The revenue growth rate from sales is a proxy for the degree of market acceptance of a new venture (Clarysse et al., 2011). Revenue data are often a preferred measure of start-up firm growth because, first, they are relatively accessible, and second, they are used for all firms (Hoy et al., 1992). Indeed, revenues are relatively insensitive to the capital intensity and the degree of integration (Delmar et al., 2003).
A three-year growth rate is computed as the difference between the log of a firm’s total revenue from sales in the first three years according to the following formulation:
$$\:\text{l}\text{n}\left(\frac{{Revenue}_{lastyear}-{Revenue}_{-3years}}{{Revenue}_{-3years}}\right)$$
(1)
The degree of educational heterogeneity in a start-up is assessed through two variables that capture the heterogeneity related to the area (“educational field heterogeneity”) and to the level of study achieved (“educational-level heterogeneity”) of its founding team. In this study, the classification proposed by Pinelli et al. (2020) is followed, grouping the educational fields into four clusters: management (economics, business and management, finance, marketing), humanities (history, philosophy, literature, arts, etc.), hard science (physics, engineering, mathematics, computer science, chemistry, medicine, biotech, etc.), and social science (law, political science, information systems, design, etc.). At the same time, the educational level is classified according to the International Standard Classification of Education (ISCED)3 into four groups identified as: upper secondary education (ISCED 3); bachelor’s degree (ISCED 5–6); master’s degree (ISCED 7); and doctoral degree– Ph.D.– (ISCED 8).
The normalized Gini heterogeneity index4 is computed to assess the heterogeneity in terms of educational level (GINIlevel) and field of studies (GINIarea). This index ranges from 0 (no heterogeneity) to 1 (maximum heterogeneity) and, for a given categorical variable, is expressed by:
$$\:G=\frac{m}{m-1}\left(1-\sum\:_{j=1}^{m}{{f}_{j}}^{2}\right)$$
(2)
where m is the number of categories and fj is the relative frequency of each category for finite samples of size n.
In the sample, most start-ups show a medium-low degree of heterogeneity concerning the area and the level of study. The relationship between these two types of heterogeneity is not very high (p = 0.39). Although the distribution of the two indices is very similar, the heterogeneity in terms of level reaches higher values than that expressed in terms of area of study.
Focusing on the educational level, to test hypotheses H3a and H3b, the dummy variable “high_educ” is included in the proposed models. This dummy expresses the presence of highly educated personnel5, and it is equal to 1 whether at least 1/3 of the company’s personnel have earned a research doctorate (or are earning a research doctorate); alternatively, at least 2/3 of the workforce must have obtained a master’s degree; 0 otherwise.
Subsequently, the effect of educational heterogeneity and the educational level is examined by controlling for several factors (Table 1) that prior studies suggest mediating ISU growth (Colombelli et al., 2016; Hyytinen et al., 2015; Rosenbusch et al., 2011).
Table 1
Descriptive statistics of the explanatory variables
Category | Variable | Label | Obs | Mean | Std. Dev. | Min | Max |
|---|---|---|---|---|---|---|---|
Target | lngrow | Growth of Revenues from sales (ln) | 1,201 | 2.306 | 2.252 | -6.908 | 8.552 |
Educational heterogeneity | gini_level | Heterogeneity for the educational level | 1,201 | 0.233 | 0.314 | 0.000 | 1.000 |
gini_area | Heterogeneity in the educational area | 1,201 | 0.237 | 0.319 | 0.000 | 0.889 | |
Educational level | high_educ | Start-up with highly educated staff | 1,201 | 0.299 | - | 0 | 1 |
Company size | curr_ass | Current assets(2) | 1,201 | 162.784 | 294.647 | 0.231 | 3809.199 |
employees | Number of employees | 1,201 | 1.271 | 4.333 | 0.000 | 129 | |
sh_cap | Share capital (1) | 1,201 | 28.114 | 112.391 | 0.000 | 2300 | |
R&D activities | patents | Patents (1) | 1,201 | 6.516 | 80.551 | 0.000 | 1738.913 |
R_D | R&D costs (1) | 1,201 | 7.965 | 58.070 | 0.000 | 961.467 |
(1)
Data in thousands of euros.
Two kinds of explanatory variables are identified. The first group of covariates includes the variables related to company size (Holmes et al., 2010). From this perspective, the following predictors are included in the model: current assets and share capital (both expressed in thousands of euros) and number of employees. The second group of variables pertains to R&D activities, including expenses for industrial patent rights and R&D costs (both expressed in thousands of euros), which are incorporated as control factors in the model.
Methods
The effect of educational heterogeneity on start-up performance is investigated by a growth function estimated by an OLS regression model. In detail, a growth function is adopted to obtain an adjusted estimate of the effect of heterogeneity, controlling for a set of K covariates focused on several firms’ characteristics. In addition, consistent with Colombelli (2016), regional dummy fixed effects (\(\:{\theta\:}_{j}\)) are included to account for the effect of the local context on firm growth, while a vector of sectorial dummies (\(\:{\delta\:}_{r}\)) captures the difference between the economic sector of activity (according to NACE rev.2):
$$\:{\varDelta\:Y}_{i}={\beta\:}_{0}+{\beta\:}_{1}{G}_{i}^{level}+{\beta\:}_{2}{G}_{i}^{area}+{\sum\:_{k=2}^{K}{\beta\:}_{k}{x}_{ik}+{\tau\:}_{i}+{\theta\:}_{j}+{\delta\:}_{r}+\:\epsilon\:}_{i}$$
(3)
where \(\:{G}_{i}^{level}\) and \(\:{G}_{i}^{area}\) are the Gini heterogeneity indexes related to the educational level (GINIlevel) and the area of studies (GINIarea) and \(\:{\tau\:}_{i}\) is the age of the firm expressed in years.
In addition, to analyze whether and how the effect of innovativeness (see Eq. 3) varies over the whole distribution of firm growth, an unconditional quantile regression (UQR) is employed (Firpo et al., 2009).
This method requires the estimation of a recentered influence function (RIF) for every quantile of interest \(\:{Q}_{\tau\:}\):
$$\:RIF\left(\varDelta\:y;{\widehat{Q}}_{\tau\:}\right)={\widehat{Q}}_{\tau\:}+\frac{\tau\:-F(y\le\:{\widehat{Q}}_{\tau\:})}{{\widehat{f}}_{y}\left({Q}_{\tau\:}\right)}$$
(4)
where \(\:{\widehat{Q}}_{\tau\:}\) is the sample τ-th quantile, \(\:{\widehat{f}}_{y}\left({Q}_{\tau\:}\right)\) is a standard non-parametric density estimator (i.e., a kernel), and F is an indicator function.
The estimated RIF is then regressed on the chosen covariates for every quantile using a standard OLS estimator. The estimated coefficients capture the marginal impact of the covariates on the quantiles of the unconditional growth rate distribution. In other words, they provide information on the effect of educational heterogeneity among low-performing firms (at the lowest quantiles) as well as among high-performing firms (at the highest quantiles). In contrast, a classical OLS regression gives information on the impact of the covariates only for an average firm. At the same time, the UQR allows for assessing whether any difference in ISUs’ growth remains constant across growth levels or if it shrinks or increases.
Results
OLS and unconditional quantile models to assess the effect of innovativeness
The first set of results concerns estimating growth models using classical OLS regression for the entire sample of firms. The growth determinants are analyzed in a stepwise fashion. The first baseline model includes the variables related to the educational heterogeneity and the educational level accounting for the size of ISUs and a set of control covariates (regional fixed effects, sectorial dummies, and age of start-ups). In model 2, the set of variables related to R&D activities is added, while in the third model, the hypothesis of the quadratic effect of heterogeneity indexes is tested. Model 4 assesses the third research hypothesis related to the mediation effect of educational level on educational heterogeneity. The final, preferred specification is model 5, which presents the results from the OLS model, including the most significant variables (the index of heterogeneity of the educational field is excluded) and accounting for sectorial and territorial fixed effects.
The results of the OLS regression (Table 2) show that the effect of vertical educational heterogeneity is significant and negative, while the heterogeneity of the field (horizontal) is statistically not significant (model 1); at the same time, the presence of highly educated staff is significantly related to better ISUs performance. Model 2 shows that the variables relating to R&D are both significant and positively affect the growth of start-ups. At the same time, the significance and magnitude of the variables expressing heterogeneity remain unchanged.
In the third step, a quadratic specification of the model is explored; the quadratic terms of Ginilevel and Giniarea are added to model 2. The results highlight that both the coefficients associated with the heterogeneity of field studies remain not significant, while the squared index of educational level heterogeneity is significant and positive. In this light, the negative coefficient of Ginilevel and the negative coefficient associated with its quadratic terms configure a U-shaped relationship between the educational level heterogeneity and the firms’ growth. Model 4 aims to test the research hypothesis H3b which assesses the significance of a mediation effect between the educational level and horizontal and vertical heterogeneity. With this in mind, interactions between the variables Ginilevel and Giniarea and the dummy related to the presence of highly qualified personnel were included in the model. The results confirm a direct effect of vertical heterogeneity (Ginilevel) but, at the same time, deny a mediating effect between educational level and educational heterogeneity since the coefficients of both interactions are statistically insignificant. In the last model, the variables without statistical significance are excluded; the coefficients related to the educational level and educational level heterogeneity, and its quadratic term, are held while the heterogeneity of educational area is removed because its effect is insignificant in any model.
Table 2
Results from OLS regression models by macro-category of the explanatory variable
Category | Variable | Model 1 | Model 2 | Model 3 | Model 4 | Model 5 | |||||
|---|---|---|---|---|---|---|---|---|---|---|---|
Coef. | Std.err. | Coef. | Std.err. | Coef. | Std.err. | Coef. | Std.err. | Coef. | Std.err. | ||
HETEROGENEITY | gini_level | -0.351* | 0.178 | -0.360** | 0.179 | -2.401*** | 0.819 | -2.247*** | 0.773 | -2.308*** | 0.805 |
gini_area | 0.269 | 0.206 | 0.215 | 0.216 | 1.417 | 1.061 | 1.481 | 1.164 | |||
gini_level (squared) | 2.969** | 1.156 | 2.931*** | 1.088 | 2.976** | 1.155 | |||||
gini_area (squared) | -1.719 | 1.400 | -1.809 | 1.443 | |||||||
EDUCATIONAL LEVEL | high_educ | 0.313* | 0.161 | 0.340** | 0.161 | 0.348** | 0.171 | 0.463*** | 0.159 | 0.359** | 0.168 |
high_educ*gini_level | -0.419 | 0.393 | |||||||||
high_educ*gini_area | -0.0572 | 0.628 | |||||||||
SIZE | curr_ass | 0.002*** | 0.000 | 0.002*** | 0.000 | 0.002*** | 0.000 | 0.002*** | 0.000 | 0.002*** | 0.000 |
Employees | 0.077 | 0.060 | 0.075 | 0.058 | 0.074 | 0.058 | 0.074 | 0.058 | 0.075 | 0.058 | |
sh_cap | -0.001 | 0.001 | -0.001** | 0.001 | -0.001** | 0.001 | -0.001** | 0.001 | -0.001** | 0.001 | |
R&D | Patents | 0.002*** | 0.001 | 0.002*** | 0.001 | 0.002*** | 0.001 | 0.002*** | 0.001 | ||
R_D | 0.002** | 0.001 | 0.002** | 0.001 | 0.002** | 0.001 | 0.002** | 0.001 | |||
Costant | 1.426** | 0.677 | 1.543** | 0.676 | 1.529** | 0.681 | -0.295 | 0.394 | 1.803*** | 0.618 | |
CONTROLS | Economic sector fixed effects | Yes | Yes | Yes | Yes | Yes | |||||
Regional fixed effects (NUTS 3) | Yes | Yes | Yes | Yes | Yes | ||||||
Age (in years) | Yes | Yes | Yes | Yes | Yes | ||||||
Adj-R2 | 0.162 | 0.171 | 0.174 | 0.175 | 0.173 | ||||||
Observations | 1,201 | 1,201 | 1,201 | 1,201 | 1,201 | ||||||
By focusing on the final specification of the econometric model, the relationship between the heterogeneity index of educational level and the ISUs growth is plotted in Fig. 1. Overall, the effect of the heterogeneity level is significant and positive; it follows a u-shaped pattern, then it decreases with respect to increasing heterogeneity for low levels of heterogeneity, but it strongly increases starting from GINIlevel values greater than 0.5. This relationship suggests that firms with a high degree of heterogeneity relative to educational levels may have a greater likelihood of growth than those with low to medium heterogeneity.
Fig. 1
Relationship between heterogeneity of educational level and ISUs’ growth. Predictive Margins with 95% confidence intervals
In addition to the OLS estimates, an unconditional quantile regression approach is employed to analyze the effect of heterogeneity at different percentiles of the growth distribution. Starting from the final specification (model 4), Fig. 2 shows the unconditional effect expressed by the coefficients of the heterogeneity of educational level in the full UQR model.
To assess the effect of the heterogeneity level at different points along the growth of revenue distribution, the coefficients of the unconditional quantile regression model for the 10th, 25th, 50th, 75th and 90th percentiles (standard error bars are reported for all UQR estimates) are reported. The results show that the effect of heterogeneity appears not to be significant at the 10th percentile. From the 25th percentile onwards, it is significant and positive.
Fig. 2
Relationship between heterogeneity level of education and ISUs’ growth. Predictive Margins with 95% confidence intervals at the 10th, 25th, 50th, 75th, and 90th percentiles
As expected, at the center of the distribution (50th percentile), the magnitude of the effect is very similar to the mean effect estimated through OLS regression. The impact of vertical heterogeneity is stronger at higher percentiles (75th and 90th ) and at the same time, the quadratic effect (u-shape) is less accentuated than at the 50th percentile.
On the one hand, the quantile regression corroborates that the heterogeneity of the founders’ educational levels positively impacts the start-up’s growth. On the other hand, it provides further evidence that its marginal effects are more accentuated at the highest percentiles. Thus, high-performing firms would benefit more than others from the heterogeneity of the educational level.
Discussion and conclusions
This study concentrates on analyzing “internal” value drivers to explain the performance of start-ups. Some confirmations, but also some denials, are provided to several literature streams, so enriching the studies on the effect of the entrepreneurial founding team’s composition on the performance of new ventures by concentrating on “innovative” new ventures, focusing on the early stage of the firms’ life cycle; investigating mainly the role of educational heterogeneity; adding empirical evidence from the previously unexplored Italian context. First, the research on start-ups’ value drivers is contributed to in several ways. Recent calls to develop analysis at downstream levels, including the founders’ team, are addressed. By introducing variables and measures for capturing heterogeneity with reference to education, the study domain for researchers is broadened.
Second, a contribution is made to entrepreneurial team research. The call for research on the relations between the founders’ team homogeneity or heterogeneity and performance in the entrepreneurship domain (Kollmann et al., 2017) is answered. The relevance of founding teams’ composition associated with team diversity and heterogeneity is confirmed.
Although not surprising, the test of the second hypothesis confirms that the effect of educational level (vertical) heterogeneity on performance is significant. Specifically, the impact of vertical heterogeneity is positive, but the effect increases for higher levels of heterogeneity. At the same time, the test of H3a confirms educational level as a driver for ISUs performance.
This result could be questionable when considering the many examples of ISUs’ success with apparently low founders’ education levels, such as Bill Gates, Steve Jobs, and Mark Zuckerberg. However, based on empirical evidence, these cases must be exceptions rather than rules. Moreover, they have yet to complete their studies (also because they were involved in managing successful ISUs) but have dropped out of prestigious institutions such as Harvard or Stanford. They have probably followed a different learning strategy (learning without obtaining a degree). However, these reflections posit new methodological challenges for scholars in the search for the most effective proxies for measuring education levels by overcoming the traditional link between “education qualifications” and “education levels”.
The denials from this study’s findings are very interesting. The first hypothesis on the impact of educational field (horizontal) heterogeneity among founders on the ISU’s performance is not significant and any moderating effect between educational heterogeneity and performance is played by educational level. Therefore, human capital theory assumptions (Gimmon & Levie, 2010), such as that the variance across founders can affect performance, are not so strongly supported. Explanations are suggested from different literature streams developing the idea that the cognitive distance among members could activate unconstructive group dynamics (misunderstanding and potential disagreements), negatively impacting on efficiency and effectiveness (Cai et al., 2013; Franco et al., 2021). The heterogeneity can lead to communication and coordination problems when the cognitive distance among founders is too high (Díaz-Fernández et al., 2020; Harrison & Klein, 2007).
Organizational behaviors can offer helpful insights into research arguing that, while the horizontal heterogeneity focuses on an “individual level”, the vertical heterogeneity regards, at the opposite, the “team level”. In line with Tasheva and Hillman (2019), since diversity is a complex, multifaceted concept, future studies should integrate its different sources (e.g., demographic, human capital, and social capital) and levels of analysis (e.g., team and individual levels). The focus should be moved from “whether” diversity is relevant to “how” and “why” issues. The interaction between individual and team levels may be the missing link explaining how and why diversity impacts team outcomes. Organizational behavior studies evidence that individual and team-level diversity are complementary for some sources of diversity, whereas for others, they are substitutes. The implications of the tested model are important for further work on diversity, team effectiveness, and public policy efforts to promote organizational and upper echelons diversity.
Third, light is shed on the literature related to the educational background of the founders (Ratzinger et al., 2018; Visintin & Pittino, 2014). Coherently with previous investigations, findings show contrasting results regarding the relevance and direction of the effect on performance (Pinelli et al., 2020). Moreover, to the best of the authors’ knowledge, there are no specific studies focused on Italian ISUs. Overall, the educational “driver” of advancing understanding of the formation and collaboration of founder teams in start-ups research is acknowledged.
Overall, this study’s controversial (and potentially equivocal) findings suggest that the entrepreneurial team research should experiment with new streams besides the most diffused ones (upper echelon theory, cognitive and social psychology theories, resource-based view, and human capital theory). In this sense, based on entrepreneurial teams’ specificities, such as the need to commonly confront nonroutine problems (van Knippenberg & Schippers, 2007), future research should apply the groups’ theory for in-depth investigations on behavioral mechanisms in founding teams, as human groups (Carroll & Payne, 1976).
Moreover, since the question of whether heterogeneity or homogeneity is beneficial for ISUs performance is highly relevant for both external stakeholders, such as investors and internal stakeholders, such as entrepreneurs, the findings show relevant implications for practitioners. Indeed, since potential differences in the education mix (horizontal and vertical) are revealed, this study has practical significance for ISUs performance. The main insights are that including founders with high educational levels and good heterogeneity among the levels of study achieved could benefit ISUs performance.
The paper has some limitations, mainly related to the data collection process. The scraping process reduced the sample size to obtain data regarding the independent variables associated with each company’s founding team. Indeed, regarding the dependent variable, three years is a relatively short period to assess the performance, even if sufficient to assess the early-stage growth. Moreover, regarding data analysis, while the effective education level could not be directly connected to qualification, education qualification is used as a proxy for measuring education levels. Future research can complement the findings by developing methodologies that can “weigh” the different aspects of heterogeneity and extend performance analysis beyond the early stage of start-ups.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
