Family Firms and R&D Investment: An investigation on the influence of internationalization and national culture

(1)

Family Firms and R&D Investment: An investigation on the

influence of internationalization and national culture

Abstract

This research aims to investigate the differential impact of family firms and non-family firms on the level R&D intensity in the period 2002-2016 in 40 different countries. The moderating effect of internationalization as well as the influence of a country’s national culture on the relationship between family firms and R&D intensity is empirically tested. Indicators for national culture used are based on two cultural dimensions provided by Hofstede’s Framework, collectivism and power distance. Results indicate that family firms engage in less R&D intensity than non-family firms. No moderating role of internalization and culture is found on the difference in R&D investment between family and non-family firms, but internationalization and collectivism are found to have a positive impact on the R&D intensity. The finding are robust when using different proxies in identifying family firms, when excluding firms with zero R&D, when excluding US and Canadian firms and finally when using running a Weighted Least Square regression

University of Groningen Faculty Economics and Business

Supervisor: Dr. H. Gonenc Co-assessor: Prof. dr. C. L. M. Hermes

Date: 9th_{of January 2020}

Student Number: sS251801 Name: Florence Doeksen Study Programme: MSc IFM

Field Key Words: Family firms, Internationalization, National Culture, Power Distance,

(2)

1

Acknowledgements

This final master thesis is the last step in the journey towards the completion of the MSc International Financial Management at the University of Groningen. Conducting this research and writing this thesis has contributed to my professional and personal development. The full thesis trajectory has been a challenging but great experience for me and this would not have been possible without the help and support of my friends, family and fellow students. Hereby, I would like to thank them all for believing in me. Most importantly, I would like to

(3)

2

1. Introduction

The study of the prevalence of family firms across the globe is becoming a recurring topic in the existing literature. Empirical studies indicate a wide variation in the behaviour of family firms and non-family firms, i.e. financial structure, investment structure and corporate governance structure (Caprio, Croci, and del Giudice, 2010; Cesari, Gonenc, and Ozkan, 2016; Chrisman and Patel, 2012; Lee and Park, 2009). The question whether family firms or non-family firms outperforms the other remains an open topic for investigation. Studies by Anderson, Mansi and Reeb, (2003) show that among U.S. corporations, the family firm outperforms the non-family firm. In research by Holderness and Sheehan (1988), the opposite is stated; U.S. non-family firms outperform family firms. A study by Maury (2006) also provides conflicting evidence in the Eurozone; Western European family firms tend to outperform non-family firms.

(6)

5 ratio in these relationships can be explained by dissimilarities in culture between nations. Other scientists propose that investment decision making depends on whether a firm is family owned or not, as family firms are different in terms of disparity between management and ownership. As Benartzie (1995), and Chrisman and Patel (2012), argue that family firms invest in more R&D, Caprio et al. (2010), and Demsetz (1983) argue that family firms are conservative towards R&D investment. As such, the relationship between ownership structure and R&D intensity yields conflicting results.

Following the definition of family firms set by the European Commission, in this research a family firm is defined as a firm in which an individual or family controls a majority of decision-making rights and where at least one family member is formally involved in the firms’ governance (European Commision, 2009). In family firms, in contrast to non-family firms, members can fill multiple positions in the business; creating a complex relational environment with the need for social and formal controls. For the success of a family firm, it is therefore crucial to develop clear governance structures around decision making to avoid conflict and to foster an environment of cohesion, shared visions and innovation (Mustakallio, Autio, and Shaker, 2002) .

(7)

6 extent of these dimensions present in a country has a large influence on societal behaviour and accordingly on corporate behaviour (Chakrabarty, 2009; Hofstede, 1980; Kalm and Gomez-Mejia, 2016). Hence, culture is an important influential factor for investment decision-making in firms, regardless of the distinction between a family firm or a non-family firm, different cultures favour different relations between managers and owners and consequently decision making (Artz et al., 2003). In addition, according to (Dunning, 1981), for successful international expansion, substantial R&D investment is necessary. For family firms, the traditional pathway to internationalization requires three determinants; commitment, financial resource availability and the ability to efficiently manage and use those resources (Graves and Thomas, 2008). As such, family firms and non-family firms differ in terms of expansion strategy and accordingly their investment behaviour.

In this paper, the aim is to deepen our understanding of the R&D investment behaviour in family firms and to provide additional evidence to justify current literature on the differential investment behaviour between family and non-family firms. By investigating the R&D investment behaviour within family firms, I also aim to examine the effect of internationalization and a country’s national culture on the R&D investment behaviour. As such, I intent to answer the following research question: “How is the level of R&D intensity in family firms influenced by the presence of internationalization as well as by different dimensions of a national culture?

(8)

7 family firms have lower levels of R&D intensity. When investigating the impact of internationalization, I do not find a significant difference for the role of internationalization on R&D intensity between family firms and non-family firms, but family firms always have a lower level of R&D than non-family firms. However, I do find a positive and significant direct effect of internationalization on R&D intensity. I also find that a country’s national culture contributes to the investment behaviour of firms. I observe that collectivism has a positive and significant effect on R&D intensity, but when analysing the effect of collectivism on the relationship between family firms and R&D intensity, I do not find a significant difference for the role of collectivism on R&D intensity between family firms and non-family firms. For power distance, I only observe a negative effect on R&D intensity, but I neither find a significant direct effect of power distance on R&D intensity, nor do I find a significant difference for the role of power distance on R&D intensity between family firms and non-family firms.

This paper contributes to the literature in several ways. First of all, I extend the existing literature on the difference in R&D investment behavior between family firms and non-family firms by investigating the difference in R&D intensity. I provide empirical evidence that family firms engage in less R&D intensity than non-family firms. In addition, I confirm the current literature that internationalization increases the level of R&D intensity, but no moderating role. Lastly, I extend the current literature on the effect of culture on the investment behavior of firms. I find no significant difference for the moderating role of culture on R&D investment behavior between family and non-family firms, but a direct effect on the R&D intensity in general.

(9)

(10)

9

2. Literature Review and Hypothesis Development

In this section a close examination of previous research on the distinction between family and non-family firms on investment behavior is presented. Corporate ownership structure plays an important role when investigating the corporate policy decisions regarding investment (Villalonga et al., 2012). Next to that, whether family businesses operate in an international context in general influences the investment strategy of a business (Kontinen and Ojala, 2010). Further, a country’s national culture also affects corporate ownership patterns due to its inherent nature (Chakrabarty, 2009). In which way the level of internationalization and country culture impact the investment decisions of family businesses is still a universal topic for discussion in the field of corporate finance. A detailed review on existing literature is provided below, which is supplemented with hypothesis development.

2.1. Family firms and investment 2.1.1. Defining Family Firms

(11)

10 involved in the firms’ governance (EC, 2009). Some scientists argue that family ownership drives out minority shareholder opportunity, and at the same time family managers tend to lack professional capability (Jiang, Jiang, Kim, and Zhang, 2015). It is also argued that family managers carry more commitment and intrinsic motivation for success and pursue a longer-term vision relative to non-family firms (Jiang et al., 2015). In addition, Andres' (2008) findings suggest that only under particular circumstances family firms achieve superior performance relative to non-family firms.

Large shareholders differ in terms of their level of influence on financing decisions, incentive structures and acquisition decisions. In their research, Anderson et al., (2003), distinguish family ownership as a “special class of shareholders”, with “unique incentive structures, strong voices in the firm, and powerful motives to manage one particular firm”. Family firms differentiate themselves from non-family firms on the basis of long-term market presence and on the premise for the family’s reputation (Anderson et al., 2003). Family firms view their firm as the most valuable family asset to bequeath from generation to generation, rather than to be consumed during a lifetime (Casson, 1999; Chami, 1999). A common trait in family ownership is therefore the long-term survival of the firm; experience and knowledge circulates within the family, from executives to heirs. This creates an environment where loyalty, confidence and trust often contribute to successful firm performance and lower costs in terms of both turnover and recruitment (Anderson et al., 2003; Ward, 1988). As such, there is a broad spectrum of characteristics that are specific to family firms.

2.1.2. Separation of Ownership from Control

(12)

11 they also differentiate in capital structures, public disclosures and acquisition strategies (Cesari et al., 2016). As commonly discussed in literature, these differences are linked to the existence of the principal-agent problem. The discussion on the effect of misalignment between shareholders and management on firm value is rather controversial (Villalonga et al., 2012). Drawing upon the documentation of Berle and Means (1932), ownership concentration eliminates potential conflict between shareholders and management, and thus positively influences firm value. Adversely, Demsetsz (1983), suggests that ownership concentration is rooted in and is a a consequence of wealth-maximizing decisions by existing ownership (Demsetz, 1983; Himmelberg, Hubbard, and Palia, 1999). With the presence of large shareholders in a corporate, the potential conflict between management, debtholders and shareholders can be mitigated. Large shareholders have the power and advantage to monitor management at the expense of minority shareholders (Berle and Means, 1932).

(13)

12 Supporting also the study by Anderson et al., (2003), Burkart et al. (2003) contribute to the literature by identifying three broad strings of literature containing the benefits of preserving family ownership and control. The first is ‘amenity potential’ of a firm, this is related to the nonmonetary private benefits of controlling a family firm, i.e. the additional benefits that are not at the expense of profits (Burkart et al., 2003). The second benefit of this preservation is the reputational benefit of the family name and heritage in political and economic market. Close network connections and qualitative relationships enrich the prestige of the family firm in a national country (Burkart et al., 2003). Thirdly, the preservation of family control eliminates opportunities for and threats by outside investors that desire to take control over the family firm. This is beneficial since engaging in external investment will cause a shift of profit towards outsiders (Burkart et al., 2003).

2.1.3. Research & Development Intensity

Family firms focus on long-term profit-maximization in order to be able to pass on wealth to future generations. This is also known as socioemotional wealth theory; in family firms socioemotional wealth plays a large role in management decision making strategies. Accordingly, family firms are common to embed a long-term management approach in their corporate strategies (Kalm and Gomez-Mejia, 2016). This also reflected in the investment behaviour. Stein (1988) classifies family firms’ investment decisions as being more efficient and long-term oriented due to the absence of managerial myopia in family firms. These firms tend to invest for the purpose of boosting net present value as opposed to management in non-family firms that rather focus on boosting current earnings at the expense of long-term firm value (Stein, 1988).

(14)

13 refer to the ‘equity premium puzzle’ of Benartzi and Thaler (1995) as a foundation for myopic loss aversion theory, which investigates why investors would invest in bonds when stocks would always yield higher returns. Chrisman and Patel (2012) refer to this theory by indicating that similar to the ‘equity premium puzzle’, family firms with the presence of myopic loss aversion, tend to invest more in long-term R&D investments despite of the short-term creation of wealth. Even if at risk, long-term oriented family goals tend to outweigh the creation of short term wealth (Chrisman and Patel, 2012). Taking into account the length of the period over which assets are evaluated, with presence of myopic loss aversion, family decision makers accept greater risk with longer periods in time (Loewenstein and Thaler, 1989); reducing risk aversion, attracting riskier investments and investing more in R&D (Benartzi and Thaler, 1995).

Based on the above mentioned existing literature, scientific literature argues that family firms invest in more R&D intensity than non-family firms. In order to empirically test this reasoning, I propose the following hypothesis:

H1A: Family firms engage more in R&D intensity than non-family firms

(15)

14 stewardship theory assumes a stronger focus on non-financial objectives than on financial objectives and closely-held ownership by family members (Nyman and Silbertson, 1978). Therefore, family firms have less propensity and incentive to invest in R&D intense projects (Barth and Gulbrandsen, 2005; Nyman and Silbertson, 1978). Supporting this study, Caprio et al., (2010) suggest the conservative approach of family firms towards large investment; family firms rather closely monitor and minimize the risk of expropriation by engaging in large R&D investment projects (Pindado, Requejo, and Torre, 2011).

In addition, R&D intense investments do not omit substantial financial contribution. Family firms are often reluctant towards seeking outside investors, as this will impact and reduce the level of control and decision making power of the family (Sirmon and Hitt, 2003); in case of failure, this will in turn damage the prestige of the family firms’ reputation (Chrisman and Patel, 2012). These potential risks largely impact the decision making of family firms to invest in R&D intense projects.

Another study by Demsetz (1983) contributes to the above as he underlines the notion of concentrated ownership on financial diversification and higher risk premiums. Higher risk premiums and limited financial diversification refrains family firms from new investment opportunities; they are reluctant regarding raising loans and admitting external investors due to a potential loss in control and ownership and as such, family firms limit their investment behaviour in terms of R&D intense projects (Agrawal and Nagarjan, 1990).

(16)

15 This reduces the level of R&D investment and results in tolerance for suboptimal investment efforts by family members (Pollack, 1985).

Building upon this existing research, the theoretical framework identifies that the level of influence of a family firm could also decrease the consent for engaging a high level of R&D intensity (Chrisman and Patel, 2012; Miller, Le Breton-Miller, and Lester, 2011). Therefore, the following alternative hypothesis has been formulated:

H1B: Family firms engage less in R&D intensity than non-family firms

2.2. Family firms, internationalization and investment

An increasingly recognized topic in the current body of literature on family firms, is the presence of internationalization strategy. It is still a debate if family firms have the capability to successfully internationalize and develop a strong competitive position in the foreign market (Carney, Duran, van Essen, and Shapiro, 2017). Continuous globalization is one of the key drivers for family firms to expand their businesses abroad (Claver, Rienda, and Quer, 2009; Parker, 1998); intensifying global competition, worldwide technological enhancement and continuous growth opportunities pressure firms to revise current operational boundaries and opts for international expansion (Pukall and Calabrò, 2014). The availability of additional foreign competitive advantages, in terms of technology, knowledge, costs and skills could serve as an attractive attempt to stimulate family firm growth (Claver et al., 2009). However, the effect of internationalization and whether at all family firms are successful in internationalizing is inconsistent in the current literature.

(17)

16 investments in order to be able to operate successfully on an international level (Graves and Thomas, 2008).

Fernández and Nieto (2005), in their research, find evidence that passing on wealth to the second generation plays an important role in decision to internationalize. As such, family firms are tempted to invest more heavily in R&D in order to create long-term value, social recognition and succession for the second generation within the family (Moini, 1995). Investing in international business networks and adopting new way of management practices to minimize risk of overseas failure will help family firms to grow and innovate the business abroad (Graves and Thomas, 2008).

Following from the international business literature, the success of international expansion depends a firm’s capability to transfer its competitive advantage abroad whilst balancing international control and coordination structures (Bartlett and Ghoshal, 1989; Dunning, 1981). According to Dunning's internationalization theory, (2001), successful international expansion requires additional investment. This, consequently, requires the presence of highly sophisticated and professional managers to manage cross-functional processes and the entire shift abroad (Carney et al., 2017). In case family firms have the capacity, the resources and the ability to develop international competitive advantages, R&D investment is attractive and will contribute to the successful international expansion. With the presence of these capabilities, family firms are not hindered to transfer their competitive advantages and develop foreign business operations by investing in R&D (Duran, Kammerlander, van Essen, & Zellweger, 2015). By further enhancing these capabilities with long-term oriented visions and possibilities to innovate, family firms are more willing to invest in and stimulate R&D (Carney et al., 2017).

(18)

17 important role in determining the international investment behavior. In their research, Arregle et al., (2012) found evidence that the inclusion of non-family members as board members in the family firm will foster even more R&D investment and higher levels of foreign activity. Additionally, they find that the scope of internationalization and the consequent effect on R&D investment is positively influenced by external board members. External involvement, i.e. non-family members in governance contributes to additional skills, knowledge and expertise which in turn leads to more valuable insight in strategic decision making (Arregle et al., 2012). So, the authors indicate that in case there are external board members within the family firm, the foreign R&D investment behavior is even more intense.

Based on the above, the literature suggests that with the presence of internationalization, the level of R&D intensity is positively influenced. As such, I propose the following hypothesis: H2A: There is a positive interaction between family firms and internationalization on R&D intensity

On the contrary, some scientists argue that internationalization will not foster R&D investment due to its intensive and complex nature abroad. It is often suggested in literature that family firms face to many constraints to internationalize successfully and as such they are limited in their foreign investment behavior (Carney et al., 2017). First of all, the high levels of coordination necessary for the transfer of competitive advantages abroad is often outside of the scope of the management team of family firms.

(19)

18 which are often considered key value drivers that enable success (Lincoln and Gerlach, 2004). Relational networks are often deeply rooted into the domestic corporate operations and therefore family firms tempt to refrain from investing abroad where there is a lack of strong relational ties. Investing in R&D abroad will demand the establishment of new relational ties and the development of communication networks; families firms are often unable to operationalize this and fail to realize the benefit of investing in R&D (Carney et al., 2017). As such, they will not have the tendency to invest in high R&D intense projects and will limit their investment behavior abroad (Carney et al., 2017).

Based on the above, the literature also argues that internationalization will not foster more R&D for family firms. As such, I propose the following alternative hypothesis:

H2B: There is a negative interaction between family firms and internationalization on R&D

intensity

2.3. Family firms, national culture and investment

As Arregle et al., (2012) have found in their research, foreign activity of family firms is limited to a relatively small set of markets. This is because not all countries have similar benefits in terms of skills, knowledge and expertise in the field of business the firm is in (Shao, Zhang, and Kwok, 2013). Often this is dependent on the national culture and the government structure of the foreign country. The national culture in terms of the norms, values and principles embedded in society and the governmental structure in terms of rules, regulations and institutions that are being imposed on firms, all influence the investment decisions (Arregle et al., 2012; Shao et al., 2013).

(20)

19 Porta, Lopez-de-Silanes, and Shleifer (1999) who indicate that all family firms are risk-averse as they invest their entire personal wealth and capital in the firm, Jiang et al., (2015) conclude that the level of risk aversion varies based on the personal values embedded in the family firm. A country national culture is a key influencer of personal norms and values; societal behavior is a reflection of national culture (Hofstede, 1980). Jiang et al., (2015) contribute to the literature as they provide evidence that family firms founded by religious entrepreneurs invest significantly less in R&D relative to family firms initiated by non-religious entrepreneurs (Jiang et al., 2015). Again, whether a society is highly religious depends on the cultural norms and values that are embedded in the development of a society (Hofstede, 1980; Jiang et al., 2015). These findings support the literature by Hilary and Hui who associate risk-aversion with religion and its subsequent effect on corporate decision making in terms of R&D investment (Hilary and Hui, 2009). On the other hand, Chakrabarty (2009) performs a cross-country level study in which he identifies the expansion patterns of family firms and the influence of a national culture on the investment intensity.

2.3.1. Collectivism

(21)

20 other” (Hofstede, 1980). This theory contributes to the sociological theory which recognizes the prominence of family ownership.

According to a study by Lee (2015), who performs a cross-cultural study on Western countries and East-Asian countries, national culture has a significant effect on corporate investment decisions. R&D intensity is different depending on the country of operation due to both the national culture but also due to the nature of large R&D investments (Lee, 2015). Generally known, R&D investment are risky investment as their outcomes are highly uncertain, but more importantly for this research is the fact that the output of an investment is realized after a longer period of time and thus requires diligence and long-term orientation. In collectivist cultures, the latter are highly valued and thus it is expected that firms with a long-term perspective and a drive for the persistence of long-long-term wealth are more active in R&D investment (Lee, 2015). According to Lee (2015), family firms prefer social wealth and transgenerational wealth over short-term economic profits and therefore engage in more R&D investment. In collectivist cultures these values align and as such encourage family firms to invest more in long-term R&D projects.

Based on this theoretical argumentation, if H1A is validated, indicating that family firms engage more in R&D than non-family firms, I can expect that in the presence of a strong collectivist culture, the relationship between family firms and R&D investment to be more positive and for H1B vice versa, a less negative relationship. As such, I propose the following alternative hypothesis:

H3A: The country level of collectivism and family firms interact positively on R&D intensity

(22)

21 and external stakeholders to enhance the business, family firms strive for minimal effort and efficient performance of its existing business (Williamson, 1994). Williamson and Ouchi (1981), who further investigate the extent of monitoring and control in family firms and non-family firms, find that non-family firms in collectivist societies operate under more efficient and smooth coordination structures as their values are aligned with collectivist values. The family firm in a collectivist society besides having well-organized and efficient flows of information as well as low monitor and control costs, creates convergence and mutual interest within the firm, with minimal formal external contracting (Hofstede, 1980; Williamson, 1994; Williamson and Ouchi, 1981). This often results in minimal effort to expand the business as expertise is kept within the firm itself and does not lend the opportunity to enhance this, hence these firms invest less in R&D. Due to the respective nature of collectivist societies and family firms, retaining internal wealth and internal ownership is a key value for generations of persistence of the family firm.

Based on the above, if H1A is validated, indicating that family firms engage more in R&D than non-family firms, I can expect that in the presence of a strong collectivist culture, the relationship between family firms and R&D investment to be less positive and for H1B a more negative relationship. As such, the following hypothesis is formulated:

H3B: The country level of collectivism and family firms interact negatively on R&D intensity

2.3.2. Power Distance

(23)

22 as a dimension of culture reduces the value of the perception of a society on institutional settlements and so influences corporate strategic decision making whether or not to invest in R&D. In countries where power distance is high, businesses are considered an elite group, and the ‘upper strata’ in society and as such they have the power and wealth to pursue their business activities as desired; resulting in more favorable R&D investment (Kirkman, Lowe, and Gibson, 2006). The higher the level of inequality, the more incentivized family firms are to take advantage of their elite position in society. Wealthy businesses that have large power in society are often faced with less political constraints and softened governmental policies (Krueger, 1974), which allows family firms privilege in accessing country level resources (Chacar and Vissa, 2005). By having access to these superior market circumstances, the risks associated with investments will be much lower and become more attractive towards the existing ‘elite, upper strata’ family firms. This will result in more intense investment behavior of the family firm (Chacar and Vissa, 2005; Krueger, 1974).

(24)

23 literature, in highly power distant countries, governments often impose regulations in favor of the elite part of the society. This results in entry barriers for ‘competitors’ of these groups; creating an environment in which the inequality between the high strata of society and the low strata of society is even more accentuated. Family firms with an elite position in such a country are equipped with market circumstances that are first of all grants them access to unique resources and secondly, eliminates competition (Ghemawat and Khanna, 1998; Khanna and Palepu, 2000). These unique market circumstances encourage long-term investment and allows for long-term orientation, and at the same time reduces the pressure for short-term economic gains (Strychalska-rudzewicz, 2016).

Following from the literature on family firms, the opportunity for long-term orientation and the persistence of wealth over generations is crucial for the success of a family firm (Ghemawat and Khanna, 1998; Khanna and Palepu, 2000). This indicates that in power distant cultures, where family businesses are powerful, family firms are able to invest in long-term R&D projects which will result in long-term value.

Following from the above, if H1A is validated, indicating that family firms engage more in R&D than non-family firms, I can expect that in the presence of a strong power distant culture, the relationship between family firms and R&D investment to be more positive and for

H1B, a less negative relationship. Based on the theoretical research framework provided above,

I formulate the following hypothesis:

H4A: The country level of power distance and family firms interact positively on R&D intensity

(25)

24 that with high power distance, centralized power rules and as such creates hierarchy, centralized knowledge and less sharing of expertise. This creates less incentive for long-term investment in R&D as only a few, (Ghemawats’ “Upper-Strata”), are faced with these opportunities. The remainder of society lags behind and rather does not put their business at risk as opposed to the centralized large powers (Humphries & Whelan, 2017; Shao et al., 2013). This is in line with the norms and values of the family firms, who wish to preserve the reputation and wealth of the family firm, rather than putting this at risk. As such, it could be argued that in countries with high power distance, family firms will invest less in R&D.

Based on the theoretical framework provided above, if HIA is validated, indicating that family firms engage more in R&D than non-family firms, I can expect that in the presence of a strong power distant culture, the relationship between family firms and R&D investment is less positive and for H1B , a more negative relationship. Consequently, I propose the following two hypotheses:

(26)

25

3. Research Method

3.1. Data Collection

Using a large international sample of industrial firms, this study aims to investigate the R&D investment behavior of family-owned firms during the period 2002-2016. I examine the investment behavior of family-owned firms relative to non-family-owned firms and also, I examine both the impact of firm-level internationalization and the national culture on the investment decisions of family-owned firms. The main data source used in this research paper is Thomson Reuter’s DataStream for firm level financial variables. The sample excludes financial firms (SIC 6000-6999) and regulated utility firms (SIC 4900-4999) because of the inevitable difference in nature of capital and investment compared to non-financial firms, and the heavy regulated industry (Gugler, Mueller, and Yurtoglu, 2008). Also, US and Canadian firms are, despite their different definition of family firms, included in the sample. However, as a robustness check, firstly empirical analysis will be performed with US and Canadian firms, and a second regression will be performed without US and Canadian firms.

National culture data is retrieved from the cultural dimension data matrix provided by Hofstede, originally collected in the 1970s and updated ever since (1980s; 1990s; 2007), latest in 2015. Empirical research has shown the validity of the dimensions over time due to its consistent and extensive use in empirical research at multiple analysis levels (Kirkman et al., 2006).

(27)

26

3.2. Variables and Methodology

This study will perform an Ordinary Least Squares (OLS) regression to perform an empirical analysis on the linear relationship between family control and R&D investment behavior. To support the relevance of the main relationship, additional regressions will be performed to test whether, internationalization and national culture (in terms of Power Distance and Collectivism) have an influence on the investment behavior of family firms. In addition, as a robustness check a Weighted Least Square (WLS) regression based on the number of observations per country will be performed.

3.2.1. Dependent Variable

For the analysis of the main relationship, specified in hypothesis one, I use the construct provided by Thomson Reuter’s DataStream for determining the value of the dependent variable R&D intensity. Following the research paper by Seifert & Gonenc (2012), who use also use R&D intensity as a dependent variable, I construct the dependent variable R&D intensity calculated as the ratio of the total research and development expenditures over total assets.

𝑅&𝐷 𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 = 𝑇𝑜𝑡𝑎𝑙 𝑅𝑒𝑠𝑒𝑎𝑟𝑐ℎ 𝑎𝑛𝑑 𝐷𝑒𝑣𝑒𝑙𝑜𝑝𝑚𝑒𝑛𝑡 𝐸𝑥𝑝𝑒𝑛𝑑𝑖𝑡𝑢𝑟𝑒𝑠

𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠 (1)

3.2.2. Independent Variable

(28)

27 strategic or institutional investor, nor a different firm, a government or other legal instance (Jiang et al., 2015). I will follow the criteria presented above meaning that if a firm that does not meet one of the two presented criteria will not be considered a family firm.

3.2.3. Moderating Variable

In our research, I need to classify firms based on whether they are international or not. To classify a firm as an international firm, I will use two different determinations of a family firm for the robustness of the empirical test. Firstly, I will use a dummy variable of foreign sales, with a value of 1 if the family firm has foreign sales >25%, and zero otherwise. A firm without a percentage <25% foreign sales will be considered a domestic firm. Secondly, I will use the foreign sales ratio to determine whether a firm is considered an international firm.

These two variables will be used in H2 where internationalization is used as the interaction variable to identify its effect on the relationship between family firms and R&D intensity.

(29)

28 3.2.4. Control Variables

Due to the existence of potential other variables besides our main variables of interest that have an effect on the level of R&D intensity, control variables are added in the empirical analysis. The control variables account for any influence on the main relationship between family firms and R&D intensity that might bias the results of the analysis. As this research is based on national and firm level variables, the control variables included are firm- and country specific. As firm-level control variables firm performance (ROA), Tobin’s Q, firm size and capital structure are included in this research. For country-level specific control variables country GDP, country political stability and education are used. Lastly, industry, and year fixed effects will be included in the empirical analysis to control for potential biased effects on the main variables of interest.

3.2.4.1 Firm-level control variables

Firm Performance

(30)

29 Tobin’s Q

Many scientists use Tobin’s Q as a control variable when determining the investment behavior of a firm. Tobin’s Q originates from a study by Brainard and Tobin (1968) in which they explain a key determinant for investment; the market value of the assets relative to the replacement costs of those assets. When the market value of capital outweighs the costs to produce it, investment is stimulated, or else, if the valuation is lower than its cost to replace, investment is discouraged (Brainard and Tobin, 1968). Therefore, firms with a high Tobin’s Q are more incentivized to intensify investment relative to those firms with a lower Tobin’s Q. As supported by research, Tobins’ Q will be calculated as the total book value of debt added to the market capitalization divided by the total book value of assets (Chung and Pruitt, 1994). Firm Size

In order to control for the influence of the size on the R&D intensity of a firm, the natural log of total assets in USD is used (Ln(Total Assets)). This results from the study by Andres (2008), indicating that on average family firms are smaller (with respect to the total number of employees, sales and total assets) relative to non-family firms, although found only partially significant. Jiang et al., (2015) use firm size as a control variable, since they indicate that smaller firms are often associated with higher risk-taking. The level of R&D intensity in a firm is also positively associated with the level of risk-taking, which could mean that small firms engage in more R&D investment (Hilary and Hui, 2009).

Capital Structure

(31)

30 investment as they tend to take more risk in order to enhance wealth. This is supported by Andres (2008), who finds that family firms significantly issue more debt in their financing structures, due to stability in family ownership over the generations (Croci et al., 2011). As such, the leverage structure of a firm potentially has a significant impact on the investment behavior.

3.2.4.2. Country-level control variables

GDP

To account for differences in economic opportunity between emerging markets and mature markets a common variable used is Gross Domestic Product (GDP). In countries with high level of GDP, the opportunity to invest and grow is larger than in countries where GDP is significantly lower (La Porta et al., 1999). Besides, Chakrabarty (2009) argues that the depending on the wealth of a country, the ownership structure within the firm might be different between wealthier and poorer countries, which influences investment decisions. Lastly, John, Litov, and Yeung (2008), argue that GDP also has an influence on the level of R&D intensity, in terms of taking. A higher GDP in a country is often associated with higher levels of risk-taking in terms of investing in large R&D intense projects (John et al., 2008). The data is sourced from the World Bank Data.

Political Stability

(32)

31 III provided by the World Bank Data to control for the strength of institutional settlements that impact investment decision making (Chakrabarty, 2009).

Education level

(33)

32

4. Methodology

In order to test the hypotheses presented in the literature review, I perform empirical analysis by running Ordinary Least Squares (OLS) regression equations based on the main relationship between family firms and R&D intensity. In the first model, I want to determine whether family firms and non-family firms differ in their investment behavior. To test this hypothesis, I create a dummy variable for the key explanatory variable, family firms, that takes a value of one for a family firm and zero otherwise. I also include year and industry fixed effects to control unobserved time-invariant characteristics that could potentially influence the independent and dependent variables in the main relationship. If the coefficient of (𝛽1) is positive, the R&D

intensity is assumed to be higher in family firms relative to nonfamily, and if negative, vice versa. In order to test the impact of family firms on R&D intensity, I use the following regression equation:

𝑅&𝐷 𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦𝑖,𝑡= 𝛽0+ 𝛽1𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1+ 𝛽2 𝑅𝑂𝐴𝑖,𝑡−1+ 𝛽3𝐹𝑆𝑖𝑧𝑒𝑖,𝑡−1+ 𝛽4𝑇𝑜𝑏𝑖𝑛′𝑠𝑄𝑖,𝑡−1+

𝛽5𝐿𝐸𝑉𝑖,𝑡−1+ 𝛽6𝐺𝐷𝑃𝑘,𝑡−1+ 𝛽7𝐸𝐷𝑈𝑘,𝑡−1+ 𝛽8𝐺𝑂𝑉𝑆𝑇𝐴𝐵𝑘,𝑡−1 + 𝛽9𝑃𝐷𝐼𝑘,𝑡−1+ 𝛽10𝐶𝑂𝐿𝐿𝑘,𝑡−1+ 𝑌𝑒𝑎𝑟 𝑓. 𝑒. +

𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑓. 𝑒. + 𝜀𝑖,𝑡,𝑘

(2)

where R&D intensity is the total R&D expenditures over total assets. I control for lagged values of the following firm-level specific variables: ROA, FSize, TobinsQ and LEV. The country specific variables, I use the lagged values of GDP, EDU and GOVSTAB.

Again, for the equation presented above, I interpret the coefficient 𝛽1 of Dummy Family Firm

(34)

33 In order to test the moderating effect of internationalization on the relationship between family firms and R&D intensity in the second hypothesis, I perform two different OLS regressions with each a different definition of internationalization for the robustness of the empirical analysis. In the first regression, I use the dummy variable for internationalization and name it Dummy INT. Secondly, I use the ratio of foreign sales as the moderator variable, where a firm with a foreign sales ratio of more than 25% will be considered an international firm and name it INT. I use the following regression equation:

𝑅&𝐷 𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦𝑖,𝑡= 𝛽0+ 𝛽1𝐷𝑢𝑚𝑚𝑦 𝐼𝑁𝑇 + 𝛽2𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1+ 𝛽3 𝐷𝑢𝑚𝑚𝑦 𝐼𝑁𝑇 ∗

𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1+ 𝛽4 𝑅𝑂𝐴𝑖,𝑡−1+ 𝛽5𝐹𝑆𝑖𝑧𝑒𝑖,𝑡−1+ 𝛽6𝑇𝑜𝑏𝑖𝑛′𝑠𝑄𝑖,𝑡−1+ 𝛽7𝐿𝐸𝑉𝑖,𝑡−1+ 𝛽8𝐺𝐷𝑃𝑘,𝑡−1+

𝛽9𝐸𝐷𝑈𝑘,𝑡−1+ 𝛽10𝐺𝑂𝑉𝑆𝑇𝐴𝐵𝑘,𝑡−1 + 𝛽11𝑃𝐷𝐼𝑘,𝑡−1+ 𝛽12𝐶𝑂𝐿𝐿𝑘,𝑡−1+ 𝑌𝑒𝑎𝑟 𝑓. 𝑒. + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑓. 𝑒. + 𝜀𝑖,𝑡,𝑘

(3)

The second regression equation I use to test the robustness of the moderating effect of internationalization on the relationship between family firms and R&D intensity is the following: 𝑅&𝐷 𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦𝑖,𝑡= 𝛽0+ 𝛽1𝐼𝑁𝑇 + 𝛽2𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1+ 𝛽3 𝐼𝑁𝑇 ∗ 𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1+ 𝛽4 𝑅𝑂𝐴𝑖,𝑡−1+ 𝛽5𝐹𝑆𝑖𝑧𝑒𝑖,𝑡−1+ 𝛽6𝑇𝑜𝑏𝑖𝑛′_{𝑠𝑄𝑖,𝑡−1}_{+ 𝛽7𝐿𝐸𝑉} 𝑖,𝑡−1+ 𝛽8𝐺𝐷𝑃𝑘,𝑡−1+ 𝛽9𝐸𝐷𝑈𝑘,𝑡−1+ 𝛽10𝐺𝑂𝑉𝑆𝑇𝐴𝐵𝑘,𝑡−1 + 𝛽11𝑃𝐷𝐼𝑘,𝑡−1+ 𝛽12𝐶𝑂𝐿𝐿𝑘,𝑡−1+ 𝑌𝑒𝑎𝑟 𝑓. 𝑒. + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑓. 𝑒. + 𝜀𝑖,𝑡,𝑘 (4)

(35)

34 equation (4). A significant coefficient indicates that internationalization indeed has an impact on the relationship between family firms and R&D intensity. Depending on the outcome of the first posited hypothesis, if H1A is true then a positive significant 𝛽3 coefficient indicates that

internationalization makes the relationship the relation between family firms and R&D intensity more positive. Where, if H1B is true, a positive significant coefficient 𝛽₃ indicates that internationalization makes the relationship less negative. On the other hand, if H1A is true and 𝛽₃ is negative and significant, then internationalization will make the relationship between family firms and R&D intensity less positive. Lastly, if H1B is true and 𝛽₃ is negative and significant, then internationalization will make this relationship more negative.

In order to investigate the moderating effect of a country’s national culture on the main relationship between family firms and R&D intensity, I will perform two different regression equations. I will use two different dimensions provided by Hofstede’s framework of a national culture as proxies for national culture; Collectivism (COLL) and Power Distance (PDI). As these are both indicators for a country specific effect, I remove country fixed effects from the regression equation. The first regression equation that is used is the following:

𝑅&𝐷 𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦𝑖,𝑡= 𝛽0+ 𝛽1𝐶𝑂𝐿𝐿 + 𝛽2𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1+ 𝛽3 𝐶𝑂𝐿𝐿 ∗ 𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1+

𝛽4 𝑅𝑂𝐴𝑖,𝑡−1+ 𝛽5𝐹𝑆𝑖𝑧𝑒𝑖,𝑡−1+ 𝛽6𝑇𝑜𝑏𝑖𝑛′𝑠𝑄𝑖,𝑡−1+ 𝛽7𝐿𝐸𝑉𝑖,𝑡−1+ 𝛽8𝐺𝐷𝑃𝑘,𝑡−1+ 𝛽9𝐸𝐷𝑈𝑘,𝑡−1+

𝛽10𝐺𝑂𝑉𝑆𝑇𝐴𝐵𝑘,𝑡−1 + 𝛽11𝑃𝐷𝐼𝑘,𝑡−1+ 𝛽12𝐶𝑂𝐿𝐿𝑘,𝑡−1+ 𝑌𝑒𝑎𝑟 𝑓. 𝑒. . + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑓. 𝑒. + 𝜀𝑖,𝑡,𝑘

(5)

(36)

35

𝑅&𝐷 𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦𝑖,𝑡= 𝛽0+ 𝛽1𝑃𝐷𝐼 + 𝛽2𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1+ 𝛽3 𝑃𝐷𝐼 ∗ 𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1+

𝛽4 𝑅𝑂𝐴𝑖,𝑡−1+ 𝛽5𝐹𝑆𝑖𝑧𝑒𝑖,𝑡−1+ 𝛽6𝑇𝑜𝑏𝑖𝑛′𝑠𝑄𝑖,𝑡−1+ 𝛽7𝐿𝐸𝑉𝑖,𝑡−1+ 𝛽8𝐺𝐷𝑃𝑘,𝑡−1+ 𝛽9𝐸𝐷𝑈𝑘,𝑡−1+

𝛽10𝐺𝑂𝑉𝑆𝑇𝐴𝐵𝑘,𝑡−1 + 𝛽11𝑃𝐷𝐼𝑘,𝑡−1+ 𝛽12𝐶𝑂𝐿𝐿𝑘,𝑡−1+ 𝑌𝑒𝑎𝑟 𝑓. 𝑒. . + 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦 𝑓. 𝑒. + 𝜀𝑖,𝑡,𝑘

(6)

In order to determine whether a country’s national culture, in terms of Collectivism, has an impact on the relationship between family firms and R&D intensity, I interpret the coefficient

𝛽3 of the variable 𝐶𝑂𝐿𝐿 ∗ 𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1, where for Power Distance I interpret 𝛽3 P𝐷𝐼 ∗ 𝐷𝑢𝑚𝑚𝑦 𝐹𝑎𝑚𝑖𝑙𝑦𝐹𝑖𝑟𝑚𝑖,𝑡−1. A significant coefficient 𝛽3indicates that the country’s level of

Collectivism/Power Distance has an influence on the main relationship between family firms and R&D intensity. Depending on the outcome of the first posited hypothesis, if H1A is true then a positive significant 𝛽₃ coefficient indicates that Collectivism/Power Distance makes the relationship the relation between family firms and R&D intensity more positive. Where, if H1B is true, a positive significant coefficient 𝛽₃ indicates that Collectivism/Power Distance makes the relationship less negative. On the other hand, if H1A is true and 𝛽₃ is negative and significant, then Collectivism/Power Distance will make the relationship between family firms and R&D intensity less positive. Lastly, if H1B is true and β3 is negative and significant, then

(37)

36

5. Results

5.1. Descriptive Statistics

Table 2 presents the descriptive statistics of the sample of the empirical analysis. Within Table 2, both the sample of non-family firms and family firms are described to identify differences in statistics in both groups. The table shows an overview of the number of observations, mean, median, standard deviation, minimum and maximum of the sample of family firms and non-family firms. In addition, I test whether there is a statistical difference in mean and median of all the firm-level variables between family and non-family firms reported by performing a T-test for means and Mann-Whitney Test for medians.

(38)

37 for collectivism, in family firms (non-family firms), Table 2, Panel A, reports a value a mean of .261 (.264), with a median of .4 (.2). Where for power distance, the table reports a mean value of .463 (.456), with a both a median value of .4.

(39)

38 Table 2. Summary Statistics

Panel A: Descriptive Statistics for sample Non-Family Firms and Family Firms

This table presents the full sample summary statistics for the data used in the empirical analysis. The full sample in the OLS regression equation consists of 39,181 observations in 40 different countries. The dependent variable used is R&D Intensity and Family Firms as the independent variable, where two different definitions of internationalization

(dummy INT and INT)) and Collectivism and Power Distance serve as moderating variables in the analysis. Firm-level

financial variables are collected from Thomson Reuter’s Datastream, including ROA, Tobins’Q, Firm Size and Capital

Structure. Country-level variables are collected from the WorldBank, including education, political stability and GDP. The

time period for data collection is 2002-2016. All variable definitions can be found in the Appendix. The variables ROA, Tobin’s Q and Capital Structure are winsorized at the 1st_{and 99}th_percentile.

(0) Non-Family Firms

n mean median sd Min max

R&D Intensityt 35672 .022 0 .055 0 1 ROAt-1 35095 .121 .123 .128 -.501 .483 Tobins Qt-1 35372 1.677 1.205 1.498 .306 9.644 FSizet-1 35672 14.818 14.879 1.732 0 20.497 LEVt-1 35665 .234 .221 .18 0 .834 Dummy INTt-1 35666 .5 1 .5 0 1 INTt-1 35672 .332 .25 .331 0 1 COLLt-1 35672 .264 .11 .224 .09 .87 PDIt-1 35672 .456 .4 .147 .11 1 GDPt-1 35672 28.851 28.955 1.364 23.885 30.534 EDUt-1 35672 .838 .87 .096 .39 .941 GOVSTABt-1 35672 .432 .409 .11 0 .718 (1) Family Firms

n mean median sd Min max

RnD intensityt 4502 .016 0 .047 0 1 ROAt-1 4409 .136 .135 .141 -.501 .483 Tobins Qt-1 4458 1.994 1.404 1.779 .306 9.644 FSizet-1 4502 14.072 14.233 1.982 2.773 19.054 LEVt-1 4501 .234 .219 .195 0 .834 Dummy INTt-1 4502 .488 0 .5 0 1 INTt-1 4502 .337 .225 .347 0 1 COLLt-1 4502 .261 .2 .204 .09 .87 PDIt-1 4502 .463 .4 .152 .11 1 GDPt-1 4502 28.394 28.363 1.321 24.922 30.534 EDUt-1 4502 .84 .867 .086 .409 .941 GOVSTABt-1 4502 .445 .412 .093 0 .718 Panel B: T-test for Equality of Means & Mann-Whitney test for Equality of Medians

This table presents the tests that have been run to test whether there is a statistical significant difference in means and medians for the firm level variables included in the regression sample. The full sample consists of 39,181 observations in 40 different countries, for the time period of 2002-2016.

T-Test equality of means Mann-Whitney test in medians T-value p-value Z-value p-value R&D Intensityt 7.05 <.001 17.37 <.001 ROAt-1 -7.40 <.001 -8.77 <.001 Tobins Qt-1 -13.00 <.001 -12.48 <.001 FSizet-1 26.8 <.001 24.62 <.001 LEVt-1 0.35 .74 1.431 .15 Dummy INTt-1 1.6 .11 1.587 .11a INTt-1 -1.00 .32 0.45 .65

(40)

39

5.2. Sample Distribution

Table 3 reports the summary statistics of the sample distribution based on the mean values by country. As reported in Table 3, Japan, the United Kingdom and the United States have a relatively higher number of firm-year observations to be found within the sample, 5157, 4113 and 15727, respectively. Countries found in the sample with the lowest number of

observations are Colombia, Czech Republic and Hungary, with 49, 30, 44 observations, respectively. Another country indicator that is reported in Table 3 is the mean value of GDP over the sampling years (2002-2016). To give an indication between the countries in the sample with the highest and lowest mean GDP, the three largest countries and three smallest countries are listed. The three smallest countries are Luxembourg, Hungary and Peru, with a GDP value lower than 132 billion USD, and the three largest countries are Japan, China and the United States with a mean value of GDP larger than 4.98 trillion USD. Looking at the minimum and maximum values of GDP for European countries (min = 213,409 USD for Portugal and max = 2,595,594 USD for the UK) and comparing this to the minimum and maximum of Asian countries (min = 186,782 USD for the Philippines and 4,988,621 USD for Japan), we observe a much higher difference in Asia between the richest and the poorest countries.

(41)

40 as being Israel (0.577), The Netherlands (0.657) an Belgium (0.705) over the sampling years (2002-2016).

(42)

41 Table 3

Country Summary Statistics by mean values.

This table represents the full country sample summary statistics of the total sample consisting of 39,181 observations in 40 different countries over the period 2002-2016. The table lists the n observations per country to illustrate the sample distribution. The table includes country specific variables including both the control variables (GDP, Political Stability, and Education) and moderating variables (Power Distance and

Collectivism). Power Distance and Collectivism are used to examine their impact on the relationship between Family Firms and R&D Intensity. The

definitions of all variables can be found in Table 1 in the Appendix A.

CountryName Observations GDP (in $mil) Political Stability Education

Index Power Distance Collectivism

(43)

42

5.3. Pearson Correlation Matrix

(44)

- 43 - Table 4

Pearson Correlation Matrix (Pairwise correlations)

This table represents the Pearson Correlation coefficients for the regression variables present in the empirical study. The coefficients are based on the total sample consisting of 39,181 observations in 41 different countries. The table includes the correlations with the dependent variable, R&D intensity. The table presents both dummy variables for the different definitions of internationalization that are used in the regression; where Internationalization (1) is takes a value of one when there are foreign sales, and zero otherwise and where in INT represents the ratio foreign sales. The next four variables are the firm specific control variables within the analysis (ROA, Tobins’ Q, Firm Size and Capital Structure). Ln (GDP),

Political Stability and Education are country-specific control variables. The table also includes Power Distance and Collectivism that serve as moderating variables in the analysis. All variable

definitions can be found in the Appendix.

Variables (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (1) R&D Intensity 1.00 (2) Dummy INTt-1 0.13* 1.00 (3) ROAt-1 -0.21* 0.03* 1.00 (4) TobinsQt-1 0.24* -0.07* 0.19* 1.00 (5) FSizet-1 -0.14* 0.24* 0.13* -0.39* 1.00 (6) LEVt-1 -0.14* -0.02* -0.06* -0.19* 0.27* 1.00 (7) INTt-1 0.12* 0.86* 0.03* -0.06* 0.21* -0.04* 1.00 (8) COLLt-1 -0.11* -0.07* 0.05* -0.11* 0.20* 0.04* -0.07* 1.00 (9) PDIt-1 -0.10* -0.12* 0.06* -0.04* 0.17* 0.05* -0.14* 0.77* 1.00 (10) GDPt-1 0.16* -0.03* -0.00 0.03* 0.18* 0.01* -0.11* -0.44* -0.18* 1.00 (11) EDUt-1 0.11* 0.13* -0.10* 0.00 -0.10* -0.04* 0.13* -0.77* -0.80* 0.31* 1.00 (12) GOVSTABt-1 0.01 0.05* -0.03* 0.04* -0.07* -0.02* 0.07* -0.13* -0.18* -0.16* 0.10* 1.00

(45)

- 44 -

5.4. Empirical Results

In order to answer the hypotheses formulated in Section 2, multiple models were created and empirically tested by conducting Ordinary Least Squares (OLS) regressions. All models use firm-level clustered SE (Huber-White Sandwich errors) in order to adjust the significance testing of the coefficients for heteroscedasticity. All independent variables are lagged by one year (t-1)relative to the dependent variable, except for time-invariant variables. This is due to

the fact that it is expected that changes in year t-1 will impact the results in year t, rather than

vice-versa; this is done in order to diminish the impact of reverse causality on the inferences/conclusions drawn from the models.

In Table 5 Model (1), family firms are regressed on R&D intensity. From the results presented in model 1, I observe a negative and significant regression coefficient for family firms of B = -0.00468, p < .01. In model 1, the predicted R&D intensity of Family firms is 0.468% lower compared to non-family firms, ceteris paribus. The effect is significant controlling for foreign sales and firm controls (in models 2 and 3 respectively). When only controlled for country characteristics, the effect is not significant (B = -0.00277, p > .05). This underlines the importance of controlling for firm characteristics.

(46)

- 45 - more strongly focus on non-financial objectives and have less propensity to invest in financially heavy R&D projects (Barth and Gulbrandsen, 2005).

(47)

- 46 - Table 5

Family Firms, Internationalization and R&D Intensity.

This table reports the estimates of the OLS regressions for R&D intensity for the full sample of family firms and non-family firms. All regressions include industry and year fixed effects and exclude country fixed effects. All independent variables, firm and country control variables are lagged with respect to the main dependent variable, R&D intensity. Robust standard errors clustered at firm-level are reported in parentheses. Variable definitions can be found in Table 1 in Appendix A.

*** Denotes statistical significance at 1% level ** Denotes statistical significance at 5% level ** Denotes statistical significance at 10% level

(1) (2) (3) (4)

Variable Names R&D Intensity

(48)

- 47 - In Table 6 Model (1), I regress family firms on R&D intensity and the interaction between family firms and internationalization is included” (as well as the direct effect of the dummy-coded internationalization). Looking at the coefficient of internationalization in Table 5 Model (1), I find a statistically significant and positive association between Internationalization and R&D intensity B = 0.00526, p < .01. This indicates that R&D intensity increases with the increasing percentage of Internationalization for non-family firms. In Model (2), I include the interaction variable, foreign sales. Table 5, Model (2) the model reports B = 0.00882, p < .01 and I conclude that R&D intensity increases with the increasing percentage of foreign sales.

In Model (1), Family dummy now represents the difference between family and non-family firms with low level of foreign sales. So, the effect is still statistically significantly negative. The interaction variable captures if the effect of foreign sales on R&D is different for family and non-family firms. The reported estimated coefficient in Model (1) of the Family Dummy with Internationalization, B = 0.00273, is insignificant, as such I cannot conclude that foreign sales creates a difference in the difference of R&D intensity for family firms and non-family firms. The coefficient reported in Model (2) for the interaction effect of ratio foreign sales is consistent with dummy internationalization, B = 0.00314, p > 0.1; insignificant. This indicates that family firms with high level of foreign sales have still lower R&D intensity relative to non-family firms.

(49)

- 48 - Table 6

Family firms, Internationalization and R&D Intensity

This table reports the estimates of the OLS regressions for R&D intensity for the full sample of family firms and non-family firms. All regressions include industry and year fixed effects and exclude country fixed effects. All independent variables, firm and country control variables are lagged with respect to the main dependent variable, R&D intensity. Robust standard errors clustered at firm-level are reported in parentheses. Variable definitions can be found in Table 1 in Appendix A,

*** Denotes statistical signifance at 1% level ** Denotes statistical significance at 5% level * Denotes statistical significance at 10% level

(1) (2)

Dummy Familyt-1 -0.00604*** -0.00478***

(0.00168) (0.00133)

Dummy INTt-1 0.00496***

(0.00124)

Dummy Familyt-1 * Dummy INTt-1 0.00273

(0.00251)

INTt-1 0.00843***

(0.00221)

Dummy Familyt-1 * INTt-1 0.00314

(50)

- 49 - In Table 7, I regress family firms on R&D intensity and include the interaction variable collectivism and power distance. Looking at the coefficient of Collectivism in Table 5 Model (1), I find a statistically significant and positive association between Collectivism and R&D intensity B = 0.01212, p < 0.01. This indicates that R&D intensity increases in countries with a higher degree of collectivism; which is in line with literature, stating that in collectivist countries retaining wealth and ownership within the family firm is a key to success as well as with a focus on diligence and long-term orientation, a culture of long-term R&D investment is stimulated (Hofstede, 1980; Lee, 2015; Williamson, 1994; Williamson & Ouchi, 1981). Looking at the coefficient of Power Distance in Table 5 Model (1), I find an insignificant and negative association between Power Distance and R&D intensity, B = -0.00760, p > 0.1.

The interaction variable that is added to both Model (1) and (2) captures how more collectivistic (power distant)countries where family firms are present is different relative to family firms present in countries having low level of Collectivism (Power Distance). The reported estimated coefficient in Model (1) of the Family Dummy with Collectivism is negative and insignificant, B = -0.00347, p > 0.1. Indicating that increasing Collectivism does not create a difference for the difference in R&D Intensity between family and non-family firms. That is, family firms still have lower levels of R&D Intensity relative to non-family firms. For Power Distance in Model (2), the reported estimated coefficient is also negative and insignificant, B = 0.00342, p > 0.1. Concluding the same as for Collectivism, increasing Power Distance does not create a difference for the difference in R&D intensity between family firms and non-family firms. Table 7, reports similar estimated regression coefficients for the control variables to those reported in Table 5.

(51)

(52)

- 51 - Table 7

Family firms, culture and R&D intensity

This table reports the estimates of the OLS regressions for R&D intensity for the full sample of family firms and non-family firms. All regressions include industry and year fixed effects and exclude country fixed effects. All independent variables, firm and country control variables are lagged with respect to the main dependent variable, R&D intensity. Robust standard errors clustered at firm-level are reported in parentheses. Variable definitions can be found in Table 1 in Appendix A

***Denotes statistical significance at 1% level ** Denotes statistical significance at 5% level * Denotes statistical significance at 10% level

(1) (2)

Dummy Familyt-1 -0.00468*** -0.00466*** (0.00134) (0.00135) Dummy INTt-1 0.00528*** 0.00527*** (0.00121) (0.00121) COLLt-1 0.01239*** 0.01201*** (0.00408) (0.00407) PDIt-1 -0.00760 -0.00702 (0.00521) (0.00560) Dummy Familyt-1 * COLLt-1 -0.00347

(0.00589)

Dummy Familyt-1 * PDIt-1 -0.00324

(53)

- 52 - 5.5. Multi-collinearity, Robust Standard Errors, Heteroscedasticity

(54)

- 53 - 5.6. Additional Analyses

In addition to the output presented in the section above, I perform several additional regressions to test the robustness in the empirical research. As I have already performed regressions using multiple variables for internationalization, I also alternate the definition of family firms in the first robustness check. Thereafter, I perform a regression with using subsamples for both dimensions of culture, then I perform a regression excluding all firms that have a value of zero for R&D expenditures and lastly, I perform another regression excluding US and Canadian firms from the sample.

5.6.1. Family Firm Definition

In this regression, I alternate the definition of family firms. Primarily, I considered a family firm as a family firm as a firm having more than 20% family ownership. Now, I use a different proxy for identifying a family firm to see whether the regression output changes if threshold is adjusted; I use a threshold of 10% of family ownership for a firm to be considered a family firm.

(55)

- 54 - of Family firms, Table 8 reports a negative and only marginal significant regression coefficient, B = -0.00932, p < 0.1. Whereas for the interaction effect of Power Distance, the regression coefficient remains insignificant, B = -0.00984, p > 0.1. Also, the control variables remain consistent with the main regression results in Table 5.

Family Firms and R&D Investment: An investigation on the influence of internationalization and national culture