Abstract
Family firms are not a homogeneous group. Thus, we theorize and empirically validate how governance differences among family firms (family governance heterogeneity) influence internationalization, considering the home country’s level of family collectivism as a moderator. Drawing on the bifurcation bias approach, we conduct a multilevel meta-analysis on 33,493 family firms. Our findings suggest that family CEOs and greater family involvement on the board hamper internationalization, but family collectivism improves these relationships. In contrast, non-founding generations promote internationalization independently of this country-level moderator. Our study reconciles the theoretical and empirical inconsistencies in the relationship between family governance and family firm internationalization.
Keywords
Introduction
With their meta-analysis, Arregle et al. (2017) dispel the myth that family firms internationalize less than nonfamily firms. However, the authors emphasize that internationalization differences may exist among family firms (Arregle et al., 2017). Thus, they call for additional research on family firm heterogeneity at the firm and country levels. This study answers their call by focusing on family governance and the home country’s culture as primary sources of this heterogeneity (Chrisman et al., 2018).
The literature emphasizes how three governance factors are not only idiosyncratic to family governance but also represent key determinants of family firm internationalization (e.g., Arregle et al., 2021; Debellis et al., 2021): the presence of a family chief executive officer (CEO), family involvement on the board, and generational involvement (i.e., whether the founding or non-founding generations lead the firm). However, it remains unclear whether these governance factors promote or hinder internationalization since extant studies show inconsistent findings. The reason for these inconsistencies is twofold. First, prior studies use divergent theoretical foundations (Arregle et al., 2017) and do not always account for the key principles of international business (IB) theory (Arregle et al., 2021). Second, most studies are conducted in a single home country. Thus, they cannot consider the role of family firm heterogeneity at the country level and its interaction with family governance at the firm level to explain internationalization (De Massis et al., 2018).
Our study addresses these shortcomings. We analyze the relationships between three family governance factors (i.e., the presence of a family CEO, family involvement on the board, and generational involvement) and internationalization, additionally showing that the home country’s level of “family collectivism” moderates these relationships (Brewer & Venaik, 2011, p. 436). We draw on the bifurcation bias approach (Verbeke & Kano, 2012) to reconcile the divergent theoretical foundations and integrate key principles of IB theory. In essence, we argue that family firms’ unique characteristics, such as long-term orientation and family social capital, promote internationalization (Hennart et al., 2019). However, family firms experience bifurcation bias, that is, an affect-based preference for family- over nonfamily-related assets, which hampers their internationalization (Kano & Verbeke, 2018). Only family firms that implement economizing mechanisms against this bias promote internationalization (Majocchi et al., 2018). Nevertheless, family governance heterogeneity alone is insufficient to explain the different levels of internationalization existing among family firms (De Massis et al., 2018) because the consequences of bifurcation bias also depend on country-level factors. Thus, we consider the cultural dimension of family collectivism as a moderator. Countries differ in the extent of pride, loyalty, and cohesion individuals feel toward their families (House et al., 2002). Family collectivism shapes the strength of the relations among family members and between family and nonfamily stakeholders, playing a crucial role in how stakeholders manage bifurcation bias (Verbeke et al., 2020).
Empirically, we conducted a multilevel meta-analysis including 44 studies with 54 independent family firm samples and 33,493 family firms. This methodology (a) accounts for the multilevel nature of our data; (b) resolves the existing empirical inconsistencies by estimating the overall effect sizes and signs (i.e., positive or negative) of relationships based on the cross-country evidence from existing quantitative primary studies; and, (c) contextualizes the relationships with country-level moderators (Borenstein et al., 2021).
Our contribution to the family firm and IB literature is threefold. First, we advance our theoretical understanding of how family firm heterogeneity affects internationalization. We reconcile the inconsistent explanations from a multitude of family firm and IB theoretical foundations on whether family governance hinders or promotes internationalization. We suggest that bifurcation bias is at the heart of family firms’ internationalization decisions (Majocchi et al., 2018). Considering family firm heterogeneity from a governance perspective complements Arregle et al.’s (2017) seminal meta-analysis on the differences between family and nonfamily firm internationalization. Second, we offer a more comprehensive understanding of family firm internationalization by considering the heterogeneity of these firms at the country level. Similar family governance can lead to different internationalization levels depending on the home country’s culture. Complementing conceptual studies on the importance of culture (Verbeke et al., 2020), we further theorize and empirically demonstrate that this country-level factor contributes to determining the consequences of bifurcation bias in the IB context. This approach extends Arregle et al. (2017) and Berrone et al.’s (2022) meta-analyses. While they focus on institutional settings, we focus on family collectivism as the cultural dimension “the most intricately linked with the integration of individuals into […] families” and firms (Sharma & Manikutty, 2005, p. 297). Third, our approach to meta-analyzing family firm heterogeneity at the firm and country levels helps reconcile the empirical inconsistencies in the relationship between family governance and internationalization. By adopting an IB perspective, our study complements and extends recent meta-analyses (e.g., Lohwasser et al., 2022; Palm et al., 2024) on the relevance of firm- and country-level factors as sources of family firm heterogeneity that affect family firms’ strategic decisions and their outcomes.
Theoretical Background
Family Governance Heterogeneity and Family Firm Internationalization
Family governance is a critical source of family firm heterogeneity at the firm level (Chrisman et al., 2018). The family and the firm’s management overlap in these firms, generating unique governance factors. The family exercises power and influence over how to deploy resources and balance family and nonfamily interests (Chrisman et al., 2018; Daspit et al., 2021). The presence of a family CEO, family involvement on the board, and generational involvement are core examples of family governance: they are idiosyncratic to family firms and key determinants of their strategic decisions, including internationalization (Arregle et al., 2021). The CEO is a firm’s primary operative decision maker (Busenbark et al., 2016) whose ultimate responsibility is to directly execute the internationalization strategy. In family firms, a family member often acts as the CEO, playing an essential role in balancing family and business interests (Chen et al., 2010; De Massis et al., 2021). A family firm’s board often consists of family and nonfamily members and has two primary tasks when internationalizing: guiding and monitoring the management. The board offers guidance by complementing the management’s expertise, knowledge, and networks (Bauweraerts et al., 2022). Moreover, the board monitors managerial actions to prevent the management from opportunistic behaviors (Hillman & Dalziel, 2003) and to ensure that family-related dynamics do not harm the firm’s well-being (Pongelli et al., 2023). Generational involvement represents the family generation leading the firm (Chrisman et al., 2018) and epitomizes the overlap between the family and the firm (Salvato et al., 2019). This governance factor influences risk tolerance and entrepreneurial orientation (Gómez-Mejía et al., 2010). As a result, generational involvement indirectly affects family firm internationalization. This study compares the founding and non-founding generations since prior literature suggests that their internationalization decisions differ (e.g., Ilhan-Nas et al., 2018; Salvato et al., 2019).
Although many previous studies investigate the relationship between family governance and family firm internationalization, its sign (positive or negative) remains unclear (Arregle et al., 2021). Some studies show that family CEOs hamper family firm internationalization (e.g., Pongelli et al., 2023), while others suggest the opposite (e.g., González & González-Galindo, 2022). Inconsistent results also prevail in the relationship between family involvement on the board and internationalization. Ramón-Llorens et al. (2017) and Ilhan-Nas et al. (2018) find no statistically significant relationship, while Sundaramurthy and Dean (2008) suggest a negative relationship. Scholars investigating generational involvement profoundly disagree on its relationship with internationalization. Karaevli and Yurtoglu (2021) find that the non-founding generations internationalize less than the founding ones. Fernández and Nieto (2005) show the opposite.
Scholars categorize the literature on the relationship between family governance and internationalization into a “restrictive” and “supportive” view (Arregle et al., 2017). The restrictive view suggests that the presence of a family CEO, family involvement on the board, and non-founding generations hinder internationalization. The agency theory and socioemotional wealth (SEW) framework agree with this view. The former highlights how family involvement in management creates agency challenges, such as nepotism and asymmetric altruism (i.e., when the family treats family employees generously despite their free-riding attitude or lacking capabilities; Chrisman et al., 2004). The latter refers to the noneconomic affect-based value that families associate with their firms (Gómez-Mejía et al., 2010). This value comes from, for example, the exercise of personal control over the firm’s operations, the identification with the firm, and the preservation of a family dynasty (Gómez-Mejía et al., 2010). The restrictive view mentions several arguments hampering internationalization. First, the lack of expertise and capabilities of family managers who are selected based on kinship rather than competence (Ramón-Llorens et al., 2017). Second, the unwillingness to hire nonfamily managers (Sundaramurthy & Dean, 2008) and accept external funding (Ilhan-Nas et al., 2018). Lastly, concerns about SEW losses (Bauweraerts et al., 2019).
By contrast, the supportive view argues that family governance promotes internationalization. The stewardship theory mainly endorses this view, suggesting that family managers act in their firm’s best long-term interest to pass it on to the next generations (Madison et al., 2016). This view relies on three main arguments. First, internationalization is a long-term and sustainable investment (González & González-Galindo, 2022). Second, family managers leverage their family social capital (i.e., family-related networks) for internationalization (Debellis et al., 2022). Third, compared to the founding generation, the non-founding generations have additional skills and competencies, essential to internationalization (Fernández & Nieto, 2005).
Family Firm Internationalization and the Bifurcation Bias Approach
We draw on the bifurcation bias approach to reconcile the opposing views on family firm internationalization. The approach (Kano & Verbeke, 2018) applies the assumptions of transaction cost theory to family firms and their internationalization, integrating family firm studies and key principles of IB theory. Ultimately, this approach identifies the conditions under which the restrictive or supportive view prevails. The approach lies on two main premises. First, family firms possess a unique set of family-related assets that constitute firm-specific advantages (FSAs; Kano et al., 2021). These assets result from the interactions between the firm and the family, as well as the environment in which the firm operates (Habbershon et al., 2003). They are characterized by high asset specificity, losing value outside the family firm. These FSAs include family managers and their loyalty to the firm, family social capital, long-term orientation, and reputation (Sestu & Majocchi, 2020). By deploying their FSAs and potentially combining them with nonfamily-related and foreign assets, family firms internationalize via export and foreign direct investments (Fourné et al., 2023; Majocchi et al., 2018). This first premise aligns with the supportive view above.
However, the approach’s second premise is that family firms also experience bifurcation bias. This bias is an affect-based heuristic postulating that family firms treat systematically and by default family-related and nonfamily-related assets differently, considering the former de facto more valuable than the latter (Verbeke & Kano, 2012). Due to this bias, family firms may make affect-based decisions that favor family over economic logic, especially in complex contexts, such as internationalization (Kano & Verbeke, 2018). This second premise aligns with the restrictive view described above. Biased family firms show suboptimal internationalization levels, predominantly leading to under-internationalization (Fourné et al., 2023; Majocchi et al., 2018). For example, biased family firms limit their internationalization by focusing on grooming family managers rather than hiring nonfamily managers with international expertise (Fourné et al., 2023). Similarly, family firms could limit investments in international opportunities to divert cash to family-related domestic operations following a SEW logic (Sánchez-Bueno & Usero, 2014).
While bifurcation bias exists by default in family firms, not all firms are equally biased. The consequences of the bias, and how well family and nonfamily stakeholders manage them, depend on firm- and country-level factors (Kano & Verbeke, 2018). At the firm level, family firms can employ economizing mechanisms to reduce bifurcation bias and leverage their family-related FSAs to internationalize. By hiring nonfamily managers, for example, family firms overcome the constraints related to the quality and quantity of family human resources, expanding knowledge, skills, and networks beyond those of the family managers (D’Angelo et al., 2016). Additionally, nonfamily managers can help family firms overcome their bias toward SEW protection by distinguishing when such protection is logical and when it is affect-based. Focusing on SEW protection does not necessarily harm family firms because the SEW can be a source of FSA for internationalization (Hennart et al., 2019). The problem is the “de facto, systemic, and dysfunctional prioritization of SEW and associated heritage assets” instead of the objective evaluation of alternatives (Kano & Verbeke, 2018, p. 165). Other examples of economizing mechanisms are the involvement of nonfamily members on the board, the introduction of formal education for family managers, and the development of formalized decision-making processes (Verbeke & Kano, 2012).
At the country level, the consequences of bifurcation bias, as well as how family and nonfamily stakeholders manage them, vary significantly depending on the home country’s culture (Verbeke et al., 2020). Culture defines individuals’ beliefs and norms and, consequently, influences how family and nonfamily stakeholders interact and do business together (Aldrich & Cliff, 2003). From a bifurcation bias perspective, a crucial cultural dimension is the level of family collectivism, i.e., the degree to which individuals are proud, loyal, and cohesive in their families and organizations (House et al., 2002). This factor has “clear, actionable implications for the specific functioning of family firms vis-à-vis other types of firms,” determining how acceptable bifurcation bias is and how family and nonfamily stakeholders manage it (Verbeke et al., 2020, p. 462). Despite the relevance of culture and, specifically, family collectivism in the context of bifurcation bias, prior studies offer mainly conceptual work with limited empirical analyses (Kim & Marler, 2022; Sharma & Manikutty, 2005). We complement these studies by considering the level of family collectivism in family firms’ home country as a moderator in the relationship between family governance and family firm internationalization. We argue that this moderator can help reconcile the existing inconsistent findings on this relationship.
Hypothesis Development
Figure 1 depicts our research model with three direct and three moderator hypotheses.
Research model.
Presence of a Family CEO and Family Firm Internationalization
We argue that family CEOs are potential sources of bifurcation bias that directly limit family firm internationalization. In contrast, the presence of nonfamily CEOs is an economizing mechanism against this bias that promotes internationalization. Three theoretical arguments support this view. First, family CEOs might miss the necessary international expertise and capabilities to identify and objectively compare domestic and international opportunities, showing a default preference for the former. Introduced early on in life to their family firms (Mitter et al., 2014), family CEOs have developed strong social capital and knowledge within the home market but lack the required managerial and specialized skills for internationalization. In contrast, nonfamily CEOs are a mechanism against bifurcation bias as they increase the quantity and quality of human resources available to family firms. Nonfamily CEOs are hired for their professional competencies (Banalieva & Eddleston, 2011). For example, with their knowledge of foreign markets and international networks, nonfamily CEOs can better identify and exploit international opportunities, facilitating internationalization (D’Angelo et al., 2016).
Second, family CEOs might overprotect the family SEW to maintain family control over the business (Gómez-Mejía et al., 2010). For example, they might decline external funding or refuse to hire nonfamily (local) managers, who are often necessary when internationalizing. Both actions would dilute family ownership and control over operations abroad (Gómez-Mejía et al., 2007). Third, family CEOs put more emphasis than nonfamily CEOs on family firms’ FSAs in their home countries (Berrone et al., 2010; Gómez-Mejía et al., 2007). Family CEOs perceive the operations and relationships in the domestic markets as family-related assets and, thus, worthy of investments by default (Majocchi et al., 2018). In contrast, nonfamily CEOs consider family-related FSAs valuable assets for international expansion, representing a way to deliver firms’ strategic objectives (Sánchez Pulido et al., 2022). Thus, we posit:
Family Involvement on the Board and Family Firm Internationalization
We also argue that greater family involvement on the board negatively influences internationalization, as family board members are more prone than their nonfamily counterparts to allow bifurcation bias to occur. We ground our hypothesis on two theoretical arguments. First, a higher number of family members on the board is more likely to reinforce family firms’ preferences for domestic investments. Family board members possess a deep knowledge of the firm and its home-country environment and dynamics, because they are generally introduced to them early on in life (Mitter et al., 2014). Working together on the board, family members can leverage their common social capital, which is often domestically oriented (Kano et al., 2021). In contrast, when fewer family members are involved on the board, family firms can capitalize on the nonbiased scrutiny of nonfamily board members (Pongelli et al., 2023). Hired for their (international) expertise, experience, and networks, nonfamily board members are better equipped to escape the “home-country trap,” overcoming any affect-based preferences for domestic over international expansion (Jimenez et al., 2019).
Second, boards with more family members might be less willing than boards with fewer family members to recombine family-related FSAs with nonfamily-related and foreign assets. For example, family members on the board might be reluctant to hire professional and internationally experienced nonfamily managers to drive internationalization because they might fear that these managers would not be good representatives of the firm abroad, damaging its reputation. However, family board members might also avoid relocating family managers abroad to prevent conflicts within the family (Ciravegna et al., 2020). Moreover, family board members may refuse external sources of capital, which are often needed when internationalizing, to maintain control over the firm’s operations (Gómez-Mejía et al., 2010). In contrast, nonfamily board members are more likely to hire necessary nonfamily human resources, relocate family managers abroad, and accept external funding if needed. Hence, we posit:
Generational Involvement and Family Firm Internationalization
We argue that non-founding generations indirectly promote internationalization more effectively than the founding generation due to their superior ability to overcome bifurcation bias. Two theoretical arguments support our hypothesis. First, compared to the family firms’ founding generation, the non-founding generations are more open to internationalization. Although the former might possess a strong entrepreneurial spirit, the latter typically receive a formal education before joining the firm (Cruz & Nordqvist, 2012). Acting as an economizing mechanism against bifurcation bias (Verbeke & Kano, 2012), this education prevents the non-founding generations from an overreliance on domestic operations. These generations are more attuned to external factors, such as shifts in business conditions (Cruz & Nordqvist, 2012), and are better positioned to assess international expansion as a viable alternative to domestic growth (Kellermanns & Eddleston, 2006). Additionally, with their work and study experiences abroad, the non-founding generations possess the necessary expertise (e.g., knowledge of foreign markets’ institutional environments and customers’ preferences) and networks to identify international opportunities (Ramírez-Pasillas et al., 2021). In contrast, the founding generation may strongly prefer domestic expansion to strengthen family social capital and enhance reputation (Gómez-Mejía et al., 2007).
Second, unlike the founding generation, the non-founding generations witness the simultaneous aging of the firm and the family. Over time, the family grows, and the firm ages. Thus, family members share fewer close bonds (Gómez-Mejía et al., 2007; Sciascia et al., 2014). Additionally, decision-making processes become less personalized and more structured (Cruz & Nordqvist, 2012). As a result, dysfunctional behaviors (e.g., nepotism and asymmetric altruism) and the inclination to systematically prefer family-related over nonfamily-related assets decline. Further, compared to the founding generation, the non-founding generations identify less with the firm, reducing biased preferences for SEW protection (Gómez-Mejía et al., 2007). For example, compared to the founding generation, non-founding generations might be more willing to accept external funding to finance internationalization opportunities (Mariotti et al., 2021), despite the risk of loosening control over the business (Gómez-Mejía et al., 2007). Thus, we posit:
The Moderating Role of the Home Country’s Level of Family Collectivism
The home country’s culture influences the consequences of bifurcation bias, as well as how family and nonfamily stakeholders manage this bias (Verbeke et al., 2020). Thus, we argue that higher levels of family collectivism weaken the negative relationship between the presence of a family CEO and internationalization, as well as the negative relationship between greater family involvement on the board and internationalization. Conversely, they strengthen the positive relationship between the non-founding generations and family firm internationalization.
In countries with higher levels of family collectivism, e.g., Italy and Mexico, individuals have a strong sense of obligation to family members (Grove, 2005) and prioritize the care of the family over the pursuit of economic goals (House et al., 2002). Nonfamily stakeholders (e.g., customers, suppliers, financial organizations, and business partners) consider bifurcation-biased behaviors the norm. Bifurcation bias aligns with family and nonfamily stakeholders’ cultural values (Verbeke et al., 2020). Therefore, nonfamily stakeholders are well-trained to navigate family-related logic and manage bifurcation bias, limiting conflicts with biased family members. Moreover, in these countries, individuals show pride and loyalty to their families and organizations. With the firm’s best interest in mind, nonfamily stakeholders are willing to engage with family members and bring a new and unbiased perspective on internationalization as an alternative to expanding in the domestic market. For example, nonfamily managers might frequently deal with family-related behaviors, such as nepotism or asymmetric altruism, and still help the family take a structured rather than affect-based approach to international expansion. Also, capital investors regularly engage with family-centered businesses and support them financially.
In contrast, in countries with lower levels of family collectivism, e.g., the U.K. and the USA, individuals focus more on the self (House et al., 2002), emphasizing personal needs over family obligations (Grove, 2005). Such countries are more transparent in differentiating between family- and business-related interests. Therefore, the occurrence of bifurcation bias has a stronger impact on family firms. Since bifurcation bias deviates from their values, nonfamily stakeholders do not accept it and are less equipped to manage its negative consequences properly (Verbeke et al., 2020). Tensions between family and nonfamily managers might arise due to the conflict between family-related (affect-based) and economic goals. As a result, for example, nonfamily managers may prefer to quit the firm rather than deal with the bias, limiting the benefits that family managers have from nonfamily managers’ objective scrutiny (Lippi & Schivardi, 2014). Further, external investors might refuse to support family firms, perceiving that these firms make internationalization decisions based on sentimental judgments rather than economic criteria (Kano & Verbeke, 2018). Hence, we posit:
Methodology
Literature Search and Sample
To identify the sample of primary studies for the meta-analysis, we performed a systematic literature review of published and unpublished studies until June 2023. We defined the sample of primary studies following the PRISMA approach, illustrated in Figure A1 (Page et al., 2021). Initially, we searched for studies in the EBSCO Business Source Premier and Scopus databases using the following keywords and combinations thereof: “international*,”“family firm*,”“governance,” and “family involvement.” We also used synonyms, such as “internationalization process,”“family-owned,”“family CEO,”“family board,” and “generational involvement.” We supplemented this search with four additional search measures. First, we used the snowballing technique on the references of existing relevant meta-analyses (Arregle et al., 2017; Duran et al., 2016; Gupta & Chauhan, 2023; Jaskiewicz et al., 2021) and literature reviews (Arregle et al., 2021; Debellis et al., 2021). Second, we conducted an issue-by-issue search in prominent international management and family firm journals (International Business Review, Journal of International Business Studies, Journal of Management Studies, Journal of World Business, European Journal of Family Business, Family Business Review, Journal of Family Business Management, and Journal of Family Business Strategy) as well as proceedings of international management and family firm conferences (Academy of International Business, Academy of Management, European International Business Academy, and International Family Enterprise Research Academy). Third, we asked IB and family firm authors for unpublished studies. Fourth, we performed unstructured searches in Google Scholar, Google, and ResearchGate (Cooper, 1998). The search yielded 6,704 studies.
After removing 2,441 duplicates, we screened the titles and abstracts of the remaining 4,263 studies. Next, we read the full text of 1,510 studies. We applied three inclusion criteria to minimize the “file drawer problem” (Rosenthal, 1979, p. 638) and capture the full extent of family firm heterogeneity (within the boundaries of our constructs’ definitions). First, primary studies use samples that consist solely of family firms. Since we analyze governance heterogeneity among family firms, our focus requires us to exclude studies with mixed family- and nonfamily-firm samples. Such studies do not indicate the share of family firms and/or do not report effect sizes separately for family and nonfamily firms. To define a “family firm,” we adopted the comprehensive approach by Arregle et al. (2017). All primary studies in our analysis define family firms as family-controlled firms, family-owned firms, family-managed firms, or family-owned or -managed firms (cf. Arregle et al., 2017, p. 813). Such a comprehensive approach to defining family firms reflects our aim to investigate family firm heterogeneity in the context of internationalization, as recently suggested by Arregle et al. (2024). Second, the primary studies report the sample size and effect size (i.e., the Pearson’s bivariate correlation coefficient) of at least one relationship between the focal constructs of our research model (Hunter & Schmidt, 2004). When studies did not report this information, we asked the respective authors for it (we contacted 32 authors and received the required information from 10 of them). Finally, if two primary studies use the same or overlapping samples, we retained the study that provided the most information on the relationships between the focal constructs (Schwens et al., 2018). By applying these three inclusion criteria, we retained 52 primary studies.
Next, we applied two exclusion criteria to ensure the methodological quality of our meta-analysis. First, we excluded studies not published in peer-reviewed journals (i.e., work-in-progress studies, conference papers, and book chapters). Second, we excluded studies published in journals not indexed in the main search databases (e.g., Scopus). Our final sample includes 44 studies with 54 independent family firm samples and 33,493 family firms. Table 1 reports the list of primary studies included in our analysis.
Primary Studies Included in the Meta-Analysis.
Note. APJM = Asia Pacific Journal of Management; BalJM = Baltic Journal of Management; E&RD = Entrepreneurship & Regional Development; EG = Estudios Gerenciales; EJFB = European Journal of Family Business; EJIM = European Journal of International Management; EMJ = European Management Journal; FBR = Family Business Review; IBR = International Business Review; IJHRM = The International Journal of Human Resource Management; JBR = Journal of Business Research; JEMI = Journal of Entrepreneurship, Management and Innovation; JFBM = Journal of Family Business Management; JFBS = Journal of Family Business Strategy; JIBS = Journal of International Business Studies; JMG = Journal of Management & Governance; JMO = Journal of Management & Organization; JoIFMI&M = Journal of International Financial Markets, Institutions & Money; JSBED = Journal of Small Business and Enterprise Development; JWB = Journal of World Business; MBR = Multinational Business Review; MIR = Management International Review; MOR = Management and Organization Review; RMS = Review of Managerial Science; SBE = Small Business Economics; TIBR = Thunderbird International Business Review.
Sample size refers to the number of firms (not observations) included in the primary studies’ samples.
Coding and Measurements
As the first step in our coding process, we extracted the effect sizes and sample sizes from the 44 primary studies. Effect sizes are the Pearson’s bivariate correlation coefficients between the family governance factors (i.e., the presence of a family CEO, family involvement on the board, and generational involvement) and internationalization (Geyskens et al., 2009). These scale-free coefficients are frequently reported in primary studies, are easily interpretable, and reflect the magnitude and direction (i.e., positive or negative) of the relationship between variables, independently of the design of the primary study (Borenstein et al., 2021). We included the Spearman’s correlation coefficient for only one primary study and conducted a robustness check, excluding this study. The results remained consistent. We chose not to use partial correlation coefficients as effect sizes, since they are influenced by the other variables in the primary studies’ empirical models (Combs et al., 2019).
A common challenge in meta-analyses is defining the focal constructs and ensuring that these definitions match the measures used in primary studies (Lipsey & Wilson, 2001). To this end, we used well-established definitions from the literature (Klier et al., 2017). We then developed a standard to code effect sizes and study information based on 10 randomly selected primary studies (Tihanyi et al., 2005). Next, two authors coded the remaining studies independently, resulting in an interrater reliability with a Cohen’s Kappa equal to .83 (Landis & Koch, 1977). The two raters discussed and resolved discrepancies.
Table B1 provides a detailed overview of the measurements included in the meta-analysis. The dependent variable, internationalization, is an adaptation of Arregle et al.’s (2017) approach and aggregates internationalization measures across four dimensions: financial measures (e.g., foreign sales to total sales; Alayo et al., 2022; Bauweraerts et al., 2019), level of multinationality (e.g., the number of foreign subsidiaries a firm has; Del Bosco & Bettinelli, 2020), internationalization scope (e.g., the number of foreign countries in which a firm operates; Cucculelli et al., 2016), and entropy measures (e.g., measures that incorporate the scale and scope dimensions of internationalization; Chung & Dahms, 2021).
Three independent variables gauge family governance. First, the presence of a family CEO is mainly a dummy variable that takes the value of 1 if a firm CEO is a family member (e.g., Cirillo et al., 2022). Second, family involvement on the board is the number (e.g., Calabrò & Mussolino, 2013) or percentage (e.g., Evert et al., 2018) of family board members. Finally, generational involvement is mainly measured as a dummy that takes the value of 1 if non-founding generations manage and/or own the family firms and 0 otherwise (e.g., Karaevli & Yurtoglu, 2021).
The moderator variable, home country’s family collectivism, is measured with the in-group collectivism dimension of the GLOBE project (practices dimension). The literature regards this measure as family-related (Brewer & Venaik, 2011) because it focuses on specific aspects of the relationship between children and parents. It reflects the extent to which individuals in a specific country are proud, loyal, and cohesive to their families and organizations (House et al., 2002). We integrated the measure from the GLOBE project based on the sampled firms’ home countries in the primary studies. This approach excludes multi-home country studies from the moderator analysis since no country-specific family collectivism value can be attributed.
We included several control variables at the country and study levels in the moderator analyses. At the country level, we controlled for factors related to cross-country differences (Tsui et al., 2007). First, we measured informal institutions with Berrone et al.’s (2022) Family Business Legitimacy Index (FBLI). Second, we measured a home country’s formal institutions with the anti-self-dealing index (Djankov et al., 2008). Finally, we controlled for a home country’s economic development, measured as its GDP (natural logarithm).
At the study level, we included both firm- and methodology-related controls. As a firm-related control variable, we included a dummy for listed firms (1 if a primary study includes solely listed firms) as a proxy for firm size and governance mechanisms. As methodology-related controls, we included four variables. First, we included the median year of the sample window. Second, we included a dummy variable to control for primary studies with a panel design. Third, we added an endogeneity check that takes the value of 1 if a primary study controls for endogeneity. Finally, we controlled for the quality of the journals in which primary studies are published with the journals’ 5-year impact factor. Table 2 reports the descriptive statistics for the sample of 44 primary studies included in the meta-analysis.
Descriptive Statistics of the Sample of Primary Studies Included in the Meta-Analysis.
Note. ES = Effect size; Mean = Simple mean of ES (not weighted); SD = Standard deviation; CEO = Chief executive officer; GDP = Gross domestic product.
Analytical Procedure
We implemented our analysis in four steps using Stata 18 (StataCorp LLC, College Station, TX, USA) and R (R Core Team, R Foundation for Statistical Computing). First, we conducted an outlier analysis and assessed potential publication bias (Steel et al., 2021) with R’s metafor package (Viechtbauer, 2010). Second, we performed a three-level random-effects meta-analysis to test the direct hypotheses (H1–3). That is, we estimated the weighted average effect size of the relationships (
In steps two and three, we applied the random-effects procedure with restricted maximum likelihood. This more conservative procedure considers the differences between the primary studies and populations’ effect sizes as consequences of sampling error and other sources of random variation (Lipsey & Wilson, 2001). We chose the restricted maximum likelihood method, as it provides approximately unbiased estimations (Viechtbauer, 2010).
Results
Data Availability
All data, codebooks, syntax files, and results for all analyses (including robustness checks) are available from a repository at the Open Science Foundation accessible here: https://osf.io/a5n24
Outlier Analysis and Assessment of Publication Bias
We conducted an outlier analysis using the influential case and outlier diagnostics by Viechtbauer and Cheung (2010). The results indicate that our sample includes an influential effect size, that is, a Pearson’s bivariate correlation coefficient between the presence of a family CEO and family firm internationalization, obtained directly from the authors of the study Del Bosco and Bettinelli (2020). Following a conservative approach, we excluded this effect size from our analysis to test H1.
To test for publication bias, we employed a comprehensive approach (Ferguson & Brannick, 2012) using the full sample of effect sizes, including the outlier. The approach comprises three procedures: the trim-and-fill technique, Egger’s regression test, and Begg and Mazumdar’s test. To minimize Type I error, we considered the impact of publication bias as problematic only if all three tests indicate the existence of bias (Ferguson & Brannick, 2012). Table 3 indicates publication bias based on the Egger’s regression test for the relationship between family involvement on the board and internationalization. Thus, according to the comprehensive approach, publication bias is not problematic in our sample.
Assessment of Publication Bias.
Note. Dependent variable = internationalization; p-value in parentheses; results refer to the sample with all effect sizes, including the outlier (refer to the subchapter “Outlier analysis and assessment of publication bias” in the Results section). B&M = Begg and Mazumdar’s test; CEO = Chief executive officer.
p < .05.
Hypothesis Testing
First, we analyzed the direct hypotheses between family governance and family firm internationalization employing the three-level random-effects meta-analysis. Table 4 displays the results. The presence of a family CEO is negatively and significantly related to internationalization (
Relationships Between Family Governance Factors and Family Firm Internationalization.
Note. Dependent variable = internationalization; p-value in parentheses; results exclude the outlier. CEO = Chief executive officer; CI = Confidence interval.
p < .05. **p < .01. ***p < .001.
Before implementing the three-level random-effects meta-regression, we examined the variance. The Q-statistics are overall significant, and the Higgins–Thompson I2-statistics are moderate to high for all three direct hypotheses (Cheung, 2014). Moreover, we calculated the intraclass correlation coefficients (ICC; Cheung, 2014) using the formulas reported by Gooty et al. (2021). These coefficients indicate the extent to which the variance occurs at Level 2 (between-study within-country variance) and Level 3 (between-country variance). The Level 3 ICCs related to the direct hypotheses are 65% (H1), 74% (H2), and 11% (H3), respectively. Thus, for example, 65% of the between-study variance in effect sizes for the relationship between the presence of a family CEO and family firm internationalization (H1) occurs among countries (Level 3), while 35% occurs within countries (Level 2). Overall, these results indicate that true variance in the effect size distributions exists (Bliese et al., 2018). To further explain this variance (Borenstein et al., 2021), we included the home country’s level of family collectivism as a moderator.
Table 5 reports the results of the three-level random-effects meta-regressions. The table reports six models, two for each moderator hypothesis (H4a-c). Models 0a-c include only the control variables, while Models 1a-c add the moderator. The findings support H4a, indicating that the home country’s level of family collectivism positively and statistically significantly moderates the relationship between the presence of a family CEO and internationalization (B = .170; p-value = .005). Thus, although family CEOs tend to hamper internationalization, they do so less in countries with higher levels of family collectivism. Moreover, results support H4b (B = .114; p-value = .011). That is, a greater presence of family members on the board generally discourages internationalization. However, this tendency occurs to a lower extent in family firms from home countries with higher levels of family collectivism. We do not find empirical support for H4c, postulating a positive moderating influence of family collectivism on the relationship between generational involvement and internationalization (B = −.056; p-value = .289).
Meta-Regressions of the Home Country’s Level of Family Collectivism on the Relationships Between Family Governance Factors and Family Firm Internationalization.
Note. Unstandardized beta coefficients (B); p-value in parentheses; results exclude the outlier. CEO = Chief executive officer.
p < .05. **p < .01. ***p < .001.
Robustness Checks
We performed several additional analyses to assess the robustness of our empirical findings. First, we re-estimated the models, including the outlier study (Steel et al., 2021). The results remain consistent, suggesting that the outlier does not distort the results of the hypothesis testing. Second, we re-estimated all models, including unpublished studies. In the meta-regression models, instead of the impact factors of the journals in which studies were published, we included a dummy variable that takes the value of 1 if a primary study is published. The results do not change, suggesting that the “file drawer problem” (Rosenthal, 1979, p. 638) is not critical for our sample. Third, we conducted a meta-analysis, including only studies that control for endogeneity. The results remain qualitatively consistent, suggesting that endogeneity should not be a concern in our analysis.
Fourth, we conducted an additional analysis for H1 and H3 to assess potential confounding issues. That is, if a family CEO belongs to a non-founding generation, there could be a confound between H1 and H3. Thus, we estimated a moderator analysis in which the variable “founding-generation CEO” moderates the direct relationships in H1 and H3. This dummy variable takes the value of 1 if the CEOs belong to the founding generation and 0 if they belong to a non-founding generation (see Table B1 for additional details). The results show that the variable “founding-generation CEO” does not significantly moderate either of the two relationships. Thus, the potential confounding issue between H1 and H3 should not be problematic in our study. Fifth, we estimated the meta-analysis by nesting all primary studies conducted in multiple home countries into a “mixed home-country” cluster. In addition, we repeated the analysis without this cluster (Wong & Cheng, 2020). In both cases, the results do not change, suggesting that our results are unlikely to be biased by cross-country settings.
Sixth, as we aggregated the internationalization variable across four dimensions, we tested whether the different dimensions influence the direct relationships in H1–3. To this end, we created a dummy variable for each internationalization dimension (see Table B1 for additional details) and estimated 12 meta-regressions in which each internationalization dimension is the only moderator of the relationships in H1–3. The results suggest that the moderator dummies are not statistically significant. Thus, we conclude that aggregating internationalization across four dimensions does not distort our main findings. Finally, we performed three distinct two-stage meta-analytical structural equation models with R’s metaSEM package (Cheung, 2023). This methodology allows the inclusion of control variables when testing the direct hypotheses H1–3 (Cheung, 2023; Kraft et al., 2022). We included firm size and firm age as control variables (see Table B1 for additional details). The results remain consistent with the original models.
Discussion
Contributions and Implications
Our study investigates family firm heterogeneity in the context of internationalization. We draw on the bifurcation bias approach (Kano & Verbeke, 2018) and meta-analyze the relationship between family governance and internationalization, considering the home country’s level of family collectivism as a boundary condition. The results overall support our hypotheses. Family CEOs hamper internationalization, and so does greater family involvement on the board. By contrast, the non-founding generations support internationalization more than the founding generation. Moreover, family collectivism weakens the negative relationships between the presence of family CEOs and internationalization, as well as between family involvement on the board and internationalization. In contrast, this country-level factor does not influence the relationship between generational involvement and internationalization.
Our study offers two theoretical contributions and one empirical contribution to the family firm and IB literature. First, we provide a more fine-grained understanding of family firm internationalization by investigating family governance heterogeneity. We theoretically reconcile the restrictive and supportive views on the relationship between family governance and internationalization by claiming that bifurcation bias is at the heart of family firms’ internationalization decisions (Fourné et al., 2023). In line with the restrictive view, we argue that this bias leads family firms to limit their internationalization (Majocchi et al., 2018). However, firms that implement economizing mechanisms can overcome their biased preferences for family- over nonfamily-related assets (Verbeke & Kano, 2012). Hence, they promote internationalization (D’Angelo et al., 2016), consistent with the supportive view. Nonfamily CEOs, nonfamily board members, and non-founding generations act as economizing mechanisms, bringing new expertise, perspectives, and networks to family firms (Fang et al., 2018; Fernández & Nieto, 2005; Gómez-Mejía et al., 2010; Pongelli et al., 2023). Our study complements Arregle and colleagues’ (2017) meta-analysis on the differences between family and nonfamily firm internationalization. The authors find that family and nonfamily firms do not have significantly different internationalization levels and call for a more fine-grained analysis of family firm heterogeneity from a governance perspective (Arregle et al., 2017). We contribute to resolving the enduring debate surrounding the family governance factors that either promote or hinder internationalization (Arregle et al., 2021). We also demonstrate that the factors hindering internationalization tend to diminish across generations. We attribute this trend to a gradual shift away from a family-centric orientation.
Second, we advance the literature on family firm heterogeneity by investigating how the interaction between firm- and country-level factors contributes to this heterogeneity. We theorize and empirically show that family firms implementing similar family governance can differ in their internationalization levels depending on their home country’s cultural setting. In countries with lower levels of family collectivism, nonfamily stakeholders hold unfavorable views of family-related preferences and respond negatively to them. For example, more experienced nonfamily managers might avoid working in family firms, and capital providers may limit their financial backing. As a result, biased family firms might lose essential sparring partners that help overcome biases against internationalization. By contrast, in countries with higher levels of family collectivism, biased family firms are considered the norm. Nonfamily stakeholders know how to manage the detrimental effects of bifurcation bias. Family and nonfamily stakeholders work together cooperatively toward international growth.
These findings underscore the importance of country-level factors and have critical implications for both theory and practice. From a theoretical perspective, we contribute to the bifurcation bias approach by extending the notion of bifurcation bias in relation to culture, a critical but largely neglected factor in explaining family firm internationalization (De Massis et al., 2018). We complement the recent conceptual studies (e.g., Arregle et al., 2019; Kim & Marler, 2022; Verbeke et al., 2020) that emphasize the role of cross-country cultural differences in explaining how family members and nonfamily stakeholders manage bifurcation bias and its negative consequences. Additionally, we extend the literature by theorizing and empirically validating that family firm heterogeneity at the firm level alone is not sufficient to explain family firms’ heterogeneous internationalization. Specifically, we extend Arregle et al. (2017) and Berrone et al.’s (2022) seminal meta-analytical findings. Their studies focus on the formal and informal institutional settings of family firms’ home countries and, specifically, how institutions influence family and nonfamily firms’ internationalization differently. In contrast, our study focuses on family collectivism as the cultural dimension “the most intricately linked with the integration of individuals into […] families” and firms (Sharma & Manikutty, 2005, p. 297) and investigates its moderating influence on a sample of only family firms. Interestingly, we find that the home country’s informal institutions (i.e., FBLI) do not significantly relate to family firm internationalization when family collectivism is included in the meta-regression (Table 5). Likewise, the home country’s formal institutions do not significantly moderate the relationship between the family governance factors and internationalization beyond family collectivism in two of three empirical models. These results underscore the relevance of family collectivism as a boundary condition in the relationships between family governance and internationalization. Additionally, we extend prior studies suggesting the relevance of the home country’s GDP as an indicator of economic development for family firm internationalization (Eddleston et al., 2019). Our findings highlight that a country’s economic development has no moderating influence in two of three empirical models beyond family collectivism.
From a practical perspective, our results suggest that culture influences family firms’ day-to-day operations, especially in countries with lower levels of family collectivism. In these contexts, nonfamily stakeholders are less tolerant of biased behaviors toward family-related assets. Thus, family stakeholders need to be mindful of this bias, as neglecting it might jeopardize support from nonfamily stakeholders. The latter, however, are critical human resources for mitigating the detrimental consequences of bifurcation bias (Verbeke & Kano, 2012) and supporting family firms in their international growth. Country-level culture is relatively stable across generations due to institutional integration and socialization (Pagda et al., 2021; Schwartz, 1999). Thus, family firms cannot change the level of family collectivism at the country level. However, they may raise awareness concerning the detrimental consequences of bifurcation bias and implement respective economizing mechanisms by adopting professional human resource practices. For example, family firms should avoid treating family and nonfamily employees asymmetrically and hire employees based on their qualifications rather than family membership (Madison et al., 2018).
Finally, we offer an empirical contribution in that our meta-analysis helps reconcile the empirical inconsistencies in the relationship between family governance and internationalization. We contribute to answering the question of whether family governance promotes or hinders internationalization by considering both firm and country-level factors. This approach complements and extends recent meta-analytical findings on family firms. On the one hand, our meta-analysis is in line with prior meta-analytical findings on the relevance of firm- and country-level factors as sources of family firm heterogeneity (e.g., Debicki et al., 2020; Hansen & Block, 2020; Hansen et al., 2020; Lohwasser et al., 2022; Palm et al., 2024). These factors contribute to shaping family firms’ strategic decisions, explaining the differences among these firms. On the other hand, we complement prior meta-analyses by adopting an IB perspective. While most existing studies investigate how various sources of family firm heterogeneity lead to different performance outcomes (e.g., Lohwasser et al., 2022), we theorize and empirically validate that this heterogeneity also influences internationalization.
Limitations and Avenues for Future Research
Our meta-analysis is not free from limitations. First, the generalizability of our results regarding how the level of family collectivism in the family firms’ home countries influences the relationships between family governance and internationalization might be limited. The multilevel meta-regression analysis to test the respective moderator hypotheses relies on effect sizes from mainly European home countries. The reason is that the scope of the research field only recently expanded to non-European contexts (Debellis et al., 2021). In our analysis, this means that we cannot compare the results of regional subsamples (e.g., Europe vs. Asia) due to the limited effect size variance at the country level (i.e., within regional subsamples) and the limited number of studies from outside Europe. Given our findings on the relevance of culture to family firm internationalization, we encourage researchers to conduct more studies on family firm internationalization from non-European countries and in additional European countries. Second, we aggregated different measures for the same construct to conduct the meta-analysis, which might distort the mean effect sizes (Klier et al., 2017). However, we mitigated the issue by grounding our constructs in well-established definitions, using widely accepted measures, employing correlation coefficients as standardized effect sizes, estimating the interrater reliability score (Lipsey & Wilson, 2001), and conducting a robustness check (refer to the sixth check in the subchapter “Robustness checks” in the Results section). However, future studies may investigate whether family governance also influences other aspects of internationalization, such as foreign location choices. Third, our meta-analysis uses correlation coefficients as effect sizes. Therefore, we cannot claim causality based on our empirical findings. Finally, we considered family collectivism at the home-country level due to data available in the sampled primary studies. Future research could focus on cultural differences within countries (regional level) or among countries within a specific area (supraregional level, e.g., Latin America).
Our study’s theoretical implications and findings suggest four additional avenues for future studies. First, while we meta-analyzed three specific family governance factors due to their relevance in internationalization decisions, future research might investigate other factors at the firm and individual levels. For example, it would be interesting to investigate founder versus non-founder family CEOs and their relationship with family firm heterogeneity in the internationalization context (Arregle et al., 2021; Zaandam et al., 2021). Second, we did not investigate the mechanisms that limit the bifurcation bias in non-founding generations in detail. We attribute this phenomenon to a higher level of professionalism coupled with a diminished family centricity (Cruz & Nordqvist, 2012). A deeper examination may enhance our understanding of intergenerational family firm internationalization from a bifurcation bias perspective. Third, we investigated the home country’s culture in the form of family collectivism since we are interested in how the relationships among family members and between family and nonfamily stakeholders affect the consequences of bifurcation bias. Future research may delve into other relevant boundary conditions. At the family level, future studies could focus on family values. An example is intergenerational authority (Björnberg & Nicholson, 2007), as this family value shapes the relations among family members and, consequently, it might influence how they manage bifurcation bias in the IB context. At the firm level, we invite scholars to investigate family firms’ business models as a boundary condition in the relationship between family governance and family firm internationalization. Such an approach would complement the findings by Hennart et al. (2019), whose study compares family and nonfamily firms. At the country level, we suggest two directions for future research. On the one hand, other country-level boundary conditions relating to family firms’ home countries. An example is the geopolitical environment, such as the presence of influential families with substantial political power that may influence how family governance relates to internationalization. Another example is the prevalent family structure in the home country. Arregle et al. (2019) theorize on the importance of family structures in the context of IB. Future studies might extend the authors’ theorization based on the bifurcation bias approach and adopt an empirical methodology that also considers family governance heterogeneity. On the other hand, other boundary conditions relating to the host countries in which family firms operate, as these conditions remain overlooked (Debellis et al., 2021). For instance, social trust remains underexplored in the context of family firm internationalization (Plakoyiannaki et al., 2025). However, it plays a fundamental role in establishing relationships (Saparito et al., 2004), which family firms consider as “building blocks of organizational structure” (Pittino & Visintin, 2011, p. 60). Finally, we showed that cultural considerations play a role in internationalization decisions when firms possess specific idiosyncrasies such as a dominant stakeholder, unique FSAs, and an economic and noneconomic logic to balance. State-owned enterprises also possess such idiosyncrasies (Benito et al., 2016; Cuervo-Cazurra et al., 2023). Future research may investigate how non-dominant stakeholders might approve or disapprove of managerial practices implemented by dominant stakeholders in state-owned firms (Estrin et al., 2016).
Conclusion
Using 44 studies with 54 independent family firm samples and 33,493 family firms, we conducted a multilevel meta-analysis on the relationship between family governance heterogeneity and internationalization, including the home country’s level of family collectivism as a moderator. Our study’s contribution to the family firm and IB literature is threefold: (a) Family governance heterogeneity explains family firms’ internationalization levels. The presence of a family CEO and greater family involvement on the board hampers internationalization. In contrast, the non-founding generations act as an economizing mechanism against bifurcation bias, supporting internationalization. (b) Family firm heterogeneity at the country level also explains family firms’ internationalization levels. In home countries with higher levels of family collectivism, nonfamily stakeholders can better manage the detrimental consequences of bifurcation bias. Although family CEOs and family members on the board limit internationalization, they do so less in countries with higher levels of family collectivism. (c) By drawing on the bifurcation bias approach, our study reconciles the theoretical and empirical inconsistencies in the family firm and IB literature on whether family governance hinders or promotes family firm internationalization, considering family firm heterogeneity at both the firm and country levels.
Footnotes
Appendix A
Appendix B
Detailed Description of Variable Measurements.
| Variable | Measure |
|---|---|
| Internationalization | |
| 1. Financial measures | The ratio of foreign sales to total sales. |
| 2. Level of multinationality | The number of a family firm’s foreign subsidiaries. |
| 3. Internationalization scope | The number of foreign markets or geographic regions in which a family firm operates. |
| 4. Entropy measures | All measurements that do not match the previous dimensions, e.g.,: |
| Family CEO | Dummy variable = 1 if a family member holds an executive leadership position, e.g., CEO or President. |
| Family involvement on the board | The percentage or the number of family members on the board. |
| Generational involvement | Dummy variable = 1 if the non-founding generations own and/or manage the firm. |
| Family collectivism | In-group collectivism dimension of the GLOBE project (practices dimension). Since the GLOBE project does not provide the in-group collectivism values for Belgium and Norway, we used the values for the Netherlands and Sweden, respectively. Further, we calculated the in-group collectivism value for Germany as the simple average of the GLOBE values for East and West Germany. Our results remain consistent when using separate values. |
| Home country’s informal institutions | Family Business Legitimacy Index (FBLI; Berrone et al., 2022). |
| Home country’s formal institutions | Anti-self-dealing index (Djankov et al., 2008). |
| Home country’s GDP | Natural logarithm of the values provided by the World Bank (The World Bank, 2024). For Taiwan: natural logarithm of the values provided by the International Monetary Fund (International Monetary Fund, 2024). We matched the GDP value with the year of the sample window in cross-sectional primary studies and the median year for longitudinal studies. If secondary data were unavailable for a specific year, we used the closest available year (Wong & Cheng, 2020). |
| Listed family firm | Dummy variable = 1 if a primary study includes solely listed firms. |
| Median year of the sample window | Year of sample window for cross-sectional studies and median year of sample window for longitudinal studies. If the year the data refers to is not reported, we used the primary study’s publication year instead. |
| Panel design | Dummy variable = 1 if a primary study adopts a panel design. |
| Endogeneity check | Dummy variable = 1 if a primary study controls for endogeneity. |
| Impact factor | The 5-year impact factor of the journal in which a primary study is published (as reported by Web of Science). |
| Published study (included in the Robustness check 2) | Dummy = 1 if a primary study is published. |
| Founding-generation CEO (included in the Robustness check 4) | Dummy variable = 1 if the descriptive statistics of the primary studies report one of the following conditions: |
| Internationalization dummies (included in the Robustness check 6) | Four dummy variables, one for each internationalization dimension. For example, the dummy variable “financial measures” takes the value of 1 if a specific primary study operationalizes internationalization with a financial measure, and 0 otherwise. |
| Firm size (included in the Robustness check 7-MASEM) | The number of a firm’s employees. |
| Firm age (included in the Robustness check 7-MASEM) | The number of years since a firm’s foundation. |
Acknowledgements
The authors would like to thank the following colleagues for their helpful feedback: Alfredo D’Angelo, Frank Lambrechts, Francesco Debellis, Emmanuella Plakoyiannaki, Thomas Lindner, Alexander Mohr, Verena Rieger, Reinhard Prügl, Alexander Kessler, the anonymous reviewers of the IFERA Conference 2023, and the ones of the EIBA Conference 2023, as well as the seminar participants at WU Vienna, University of Vienna, University of Innsbruck, and LUISS Guido Carli.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
