Advertisement

Melting pot or tribe? Country-level ethnic diversity and its effect on subsidiaries

  • Jennifer Oetzel
  • Chang Hoon OhEmail author
Article

Abstract

The purpose of our study is to explore the effect of country-level ethnic diversity on subsidiary-level ownership strategy and employee productivity. We examine two characteristics of ethnic diversity in society, the level of diversity and degree of inclusion. Using a multi-source and multi-year (2004–2010) sample of 30,007 subsidiaries from 79 home-countries operating in 63 host-countries, we find that high levels of diversity are not a de facto form of country risk as some have argued. Rather, it is the ability of a dominant ethnic group to exclude others from full economic and political participation. As societies become more fragmented, and one or few groups become more dominant and prevents other groups from fully participating in the political process or economic opportunities, this may increase business risk and lower labor productivity. The broader policy implications are that policymakers should have engagement policies toward disfranchised ethnic groups. Policymakers should develop polices aimed at leveraging the benefits of diversity. MNCs have an interest in promoting such policies and fostering greater economic and political inclusion in countries where they operate.

Keywords

ethnic diversity subsidiary ownership subsidiary performance employee productivity investment risk 

Introduction

Over the last several decades, globalization has been a policy objective of political leaders in the U.S., Europe, and other countries, as well as in important policymaking bodies such as the World Bank, The International Monetary Fund and the World Trade Organization. Globalization is not only about trade but also the free exchange of ideas and the movement of people across borders. The European Union’s (EU) policy of open borders and the free movement of labor across the EU has resulted in growing ethnic diversity across the region (European Commission, 2018; Riggs, 2002).

Ethnicity is defined as a subjectively experienced sense of commonality based on a belief in a common ancestry and shared culture. It is considered subjective because members and nonmembers identify the ethnic groups that they perceive to be present in their society. These groups may be defined by their linguistic, religious, or ethno-somatic similarities1 (Wimmer, Cederman, & Min, 2009: 325). The United States offers an interesting example of the potential advantages of country-level ethnic diversity. Although the U.S. is no stranger to ethnic conflicts and discrimination, and ethnic and racial strife has increased substantially since the 2016 presidential election (Southern Poverty Law Center, 2016), the country has consistently been the number one recipient of foreign direct investment (FDI) and arguably one of the most desirable countries in which to do business (CIA, 2014; UNCTAD, 2014). In fact, the ethnic diversity of the U.S. is often seen as one of the country’s strengths. A significant number of studies conducted by management scholars have found that in the U.S., ethnic diversity – measured at the organizational and national levels of analysis – is often (although not universally) linked to higher levels of creativity and greater productivity (Cox, Lobel, & McLeod, 1991; Elron, 1997; Giambatista & Bhappu, 2010; Richard, 2000; Stahl, Maznevski, Voight, & Jonsen, 2010).

While policies promoting ethnic diversity have always been criticized in some quarters, support for diversity appeared to dominate public debate until the global economic crisis that began in 2007. Since that time, politicians in the United States and the European Union have been openly critical of multiculturalism suggesting that greater diversity is more likely to generate greater socio-political conflict and significant economic costs. Despite the well documented benefits of diversity, some studies have identified potential costs. For example, ethnic diversity is associated with greater intra-group (Pelled, Eisenhardt, & Xin, 1999), organizational (Chua, 2013), and country-level conflicts (Alesina, Baqir, & Easterly, 1999; Alesina & La Ferrara, 2005). Ethnically diverse countries, some suggest, have a lower growth rate and a greater probability of conflict than other countries (Easterly & Levine, 1997). In a widely cited article on cross-country growth rates across Sub-Saharan Africa, researchers found that high levels of ethnic diversity explain a substantial portion of the poor economic growth in the region, largely because diversity tends to decrease the consensus for public goods, such as education, which in turn leads to poor economic outcomes (Easterly & Levine, 1997). As part of their study, the authors tested their findings with a broader sample of countries and found similar results (Easterly & Levine, 1997). Subsequently, other scholars have also argued that ethnic diversity drives social conflict and is generally detrimental to society (Alesina & La Ferrara, 2005; Wimmer et al., 2009).

Research in international business has examine the cultural aspects of ethnic diversity, including linguistic, racial, and religious differences across groups of people and made cross-country comparisons (Hofstede, 1980; Johanson & Vahlne, 1977; Ralston, Gustafson, Cheung, & Terpstra, 1993; Shenkar, 2012). There is also substantial work on the benefits of ethnic ties for international trade and investment (Duanmu & Guney, 2013). A related stream of research has focused on the demographic nature of ethnic fractionalization, which focuses on the number, size, geographic, and socioeconomic distributions of different groups in society (Luiz, 2015).

To contribute to the research on ethnic diversity and its effect on international business policies, we offer another way to conceptualize ethnic diversity in a country that builds on and extends previous research. We suggest that it may be valuable to go beyond demographics and cultural characteristics alone and examine the degree to which different ethnic groups are included in the institutional fabric, or national governance system in a country (Berry et al. 2010; Whitley, 1992, 1997). As Berry et al. (2010) suggest, demographic, cultural and political institutions can give rise to differences in business systems. Thus by examining ethnicity from an institutional perspective we can better understand within country differences in diversity and how they affect firm behavior (Dow, Cuypers, & Ertug, 2016; Tung, 2008). Based on an institutional view of ethnicity, we suggest that regardless of the specific religious, linguistic or ethnic differences in a country, it is the degree of economic and political inclusion experienced by minority groups that may be critical. If ethnic minority groups are excluded from socioeconomic benefits and institutions, for instance, ethnic diversity may be more likely to create socio-political tensions.

Since the research on country-level diversity and its impact on firms is still unsettled, the aim of this paper is to explore the relationship between country-level ethnic diversity, ownership strategy, and subsidiary performance by evaluating how managers view ethnic diversity when formulating subsidiary-level ownership strategy and how ethnic diversity in society may affect employee productivity. To do this we undertake a multi-country study (30,007 subsidiaries of 4,003 MNCs from 79 home-countries operating in 63 host-countries) using a multidisciplinary approach. Our findings show that the effect of ethnic diversity on subsidiary ownership and employee productivity is more complex than one might think. Altough ethnic diversity may pose a risk in a country when there are either low or high-levels of ethnic diversity, ethnic diversity per se is not a risk. Rather, diversity only becomes a risk if some ethnic groups are disenfranchised or unrepresented in the political and economic life of a country. We also find that the structure of ethnic composition and the relative power of different groups in a host country are important in determining the risk from ethnic diversity and subsequently firm ownership structure and employee productivity.

These findings indicate that a country’s policies related to ethnic diversity, cultural pluralism, and immigration should not be considered separately from economic policies. The policy implications are that government officials should support policies that increase political and economic inclusion in a country. Likewise, businesses should lobby government officials to support ethnic inclusion and promote policies that do so since effective management of diversity can positively impact business and ultimately economic growth in the country. Business leaders, in particular those from MNCs, have done so for other social issues such as climate change, gender inequality, and anti-corruption through national and international business forums. If MNCs tackle the issue of promoting ethnic inclusion in the workplace, such policies will likely spread to other countries where the MNC operates. In this way, MNCs can positively contribute to increasing diversity and inclusion and minimizing ethnic conflict through effective corporate and human resource policies.

LITERATURE and HYPOTHESES

Country-level Ethnic Diversity and a Non-linear Relationship with Firm Ownership Strategy

In a number of influential studies on ethnic diversity, researchers found that ethnically diverse societies are more prone to conflict (Easterly & Levine, 1997; Esteban et al., 2012). One explanation for this finding is that some ethnic groups may be able to marginalize, or outright exclude, other minority groups from participating fully in the economic and political life of a country. The greater the degree of exclusion, the more motivated minority groups may be to pursue their interests through violent means. When the excluded group is relatively homogenous, studies show that it may be easier to organize and pursue a collective political agenda (Collier, 2001), at times even becoming a separatist movement. Examples include the Basques and Catalans in Spain.

When ethnic groups compete with one another, managers may worry that the political “winners” may adopt policies and political regimes that negatively affect or exclude other ethnic groups in the country. Indeed, numerous articles have shown that political conflicts over public policies are increasingly determined by ethnic and racial cleavages rather than by class (e.g., Alesina et al., 1999; Hacker, 1995; Huckfeldt & Kohfeld, 1989). Also, in the case of ethnic conflicts, businesses can be a direct target of opposing ethnic groups. For example, ethnic Chinese and their businesses in Indonesia have suffered from ethnic conflict. The conflict has not only existed between Islamic nationalists (Sarekat Islam) and Chinese, but also between several ethnic and religious groups in Indonesia including indigenous peoples, Muslims, Sunni Muslims, Japanese, and Dutch (Purdey, 2006; Shekhar, 2014).

Given such findings, it is not surprising that historically, traders have tended to do business within their own ethnic groups and viewed cross-ethnic business transactions as riskier. By doing so, managers may be able to better monitor trading partners and to minimize opportunistic behavior (Alesina & La Ferrara, 2005; Duanmu & Guney, 2013; Grief, 1993; La Ferrara, 2003). When members of an ethnic group renege on agreements or otherwise act opportunistically, not only can the individual involved be punished, but others in that person’s family or network can be as well (Alesina & La Ferrara, 2005; La Ferrara, 2003). Interestingly enough, more recent work has shown that venture capitalists (VCs) are more likely to invest in start-ups when the VC and the entrepreneur share the same ethnicity (Hegde & Tumlinson, 2014). This suggests that entrepreneurs with a different ethnicity than potential VCs may face more difficulty in obtaining funding for their ventures. Another study found that ethnic networks have a positive impact on imports and exports, further confirming the importance of ethnicity to business (Duanmu & Guney, 2013).

These patterns of doing business within rather than across ethnic groups may become path dependent because weak contract enforcement or information asymmetries about the integrity and reputation of buyers and sellers may lead managers to avoid doing business across ethnic groups (Duanmu & Guney, 2013; Grief, 1993; La Ferrara, 2003) and view ethnic diversity as a risk. It can also be hard to coordinate and organize given linguistic and geographic barriers that may exist between ethnic groups (Dow et al., 2016). Even when the groups are not in conflict and not biased or prejudiced against one another, contractual concerns may make it difficult for firms to develop new business ties across ethnic groups. Also, even in the absence of conflict, managers may prefer to strengthen their own ethnic ties rather than to diversify ownership and control across other ethnic groups. Over time, path dependency may lead to the persistence of these relationships even when the quality of contract enforcement increases.

There is a relatively large literature in international strategy that details how country-level factors may impact corporate governance, specifically firm ownership strategy, employee productivity, and organizational performance. For instance, studies show that the higher the political risk in a country the more likely it is that managers will choose to lower their ownership and investment levels (Agarwal & Ramaswami, 1992; Delios & Henisz, 2000; Gatignon & Anderson, 1988). Thus, all else being equal, if ethnic diversity is a risk than the higher the perceived (ethnic) risk in a country, the more likely it is that managers will choose governance structures that involve lowering their investment levels.

It is well established that risk tends to deter entry and investment (Anderson & Gatignon, 1986; Delios & Henisz, 2003; Feinberg & Gupta, 2009; García-Canal & Guillén, 2008; Gatignon & Anderson, 1988). When firms operate in a country that is experiencing risk, managers will often choose to share ownership so that they can also share the risk (Delios & Beamish, 1999; Gatignon & Anderson, 1988). Thus, we would expect that firms would avoid establishing subsidiaries with high-ownership levels in countries that have a greater potential for ethnic conflict and violence. Managers may be uncertain about the risks in highly diverse societies. When uncertain, they are more likely to avoid markets – or segments of markets – that are seen as riskier and/or reduce their ownership commitment.

For these reasons, instead of looking at market entry decisions or the volume of a firm’s investment, we investigate a firm’s subsidiary ownership strategy. The rationale is that firms often prepare for and manage anticipated risks in a host country through their ownership choice. Depending upon the level of perceived host country risk, headquarters will decide how much autonomy to grant subsidiaries. When the subsidiary faces lower risks, greater legitimacy, and positive employee and community relationships in the host society, ownership levels can be expected to increase (Anderson & Gatignon, 1986; Boytsun, Deloof & Matthyssens, 2011; Feinberg & Gupta, 2009; Gatignon & Anderson, 1988; Whitley 1997). Thus, we would expect that firms would avoid establishing subsidiaries with high-ownership levels in countries that have a greater potential for ethnic conflict and violence.

Perhaps, however, a more nuanced view of ethnic diversity prevails. It may be that the ethnic make-up of society and how well it is managed – both by society and by managers – is more important. For instance, some national level studies have often found that ethnic diversity provides a variety of skills, abilities, perspectives, and cultures that may lead to greater productivity, innovation, and creativity (Alesina & La Ferrara, 2005; Ottaviano & Peri, 2006; Ottaviano & Prarolo, 2009). Other studies suggest that ethnic diversity may be neither beneficial nor detrimental to business. Researchers have found that when ethnic composition is diverse enough in a society, ethnicity is not representative of identity (Bannon, Miguel, & Posner, 2004). One reason is that in some contexts no ethnic group may be large enough to legitimize its political power or ideological and religious supremacy, or to privatize or redistribute public goods for the interests and preferences of its own group members (e.g., Esteban et al., 2012; Hale 2004).

Other researchers (Collier & Hoeffler, 1998, 2002) have questioned the assumption of ethnic diversity as risk. In fact, they empirically demonstrated that ethnolinguistic diversity is not necessarily linked to a greater likelihood of violence as previously thought. Rather, the so-called Collier-Hoeffler (CH) model suggests that for several reasons, including the difficulty in organizing a sustained rebellion, the likelihood of sharing the same primary grievances, and the group’s susceptibility to a ‘divide and conquer’ strategy by opponents, diverse societies are significantly safer than (even) homogeneous societies (Collier & Hoeffler, 1998). Most importantly, the authors argue that the relationship between ethnic diversity and violence is nonlinear. Ethnic salience is lower in ethnically fragmented societies, but it is bound up in the competition for political power (Bannon et al., 2004; Eifert, Miguel, & Posner, 2010).

Ethnicity becomes salient when ethnic groups compete for political power (Bannon et al., 2004; Eifert et al., 2010). To the extent that these theoretical arguments hold, we would expect that subsidiary risk would be at its highest when a country is characterized by a few dominant ethnic groups. Risk is expected to be higher in such cases because the majority groups are more likely to exclude and/or oppress other minority groups, or because a few ethnic groups of similar size are likely to compete over economic and political interests. Since firms are more likely to increase their ownership in foreign ventures as risk decreases (all else being equal) and decrease ownership as risk rises (Anderson & Gatignon, 1986; Feinberg & Gupta, 2009; Gatignon & Anderson, 1988), we would expect that subsidiary ownership and ethnic diversity would follow a U-shaped relationship. In addition, when ethnic risk is high in a society, social capital and trust tend to be very low. In an effort to increase their legitimacy in the society, managers of subsidiaries may respond to such conditions by expanding their ownership structure to include several types of local partners and other stakeholders. For these reasons we hypothesize that:

Hypothesis 1:

When the levels of ethnic diversity in a country are high or low, one ethnic group is less likely to exclude other groups therefore ethnic conflict risk is low for businesses. Thus ethnic diversity will follow a U-shaped relationship with the percentage of subsidiary ownership held the parent firm.

Country-Level Ethnic Diversity and a Non-linear Relationship with Employee Productivity

To the extent that there is ethnic conflict in a society, we would expect that business challenges would likely manifest themselves in terms of labor force productivity. In fact, ethnic tensions (to the extent that they exist) in the wider society may spill over into the workplace resulting in actual realized risk to the firm (Brief, Umphress, Dietz, Butz, & Scholten, 2005; Chua, 2013). For these reasons, businesses have a strong interest in engaging in discussions with governments and other stakeholders and lobbying host country governments to support inclusion and address ethnic tensions.

Poor management can also lead to increased ethnic tensions. In some countries, ethnic diversity may be accompanied by economic, social, and/or political exclusion of certain groups making it difficult to even meet or work with those who have different backgrounds. When this type of pattern occurs over decades, ethnic groups may develop skill sets that are traditional in their ethnic group (Borjas, 1992). The exclusion of certain ethnic groups can increase the propensity to form ethnic enclaves. When this happens, ethnic minorities may only find economic opportunities with firms owned by those of the same ethnicity. While ethnic firms positively contribute to ethnic groups in a segmented society, the pool of potential workers is limited for firms in the segmented enclave economy compared to that in the open economy (Sanders & Nee, 1987). Thus the overall quality and experience of the workforce is lower in a country with high ethnic conflict risk than it is in a country with low ethnic conflict risk.

When some ethnic groups are formally or informally disenfranchised, this can lead to periods of violence and even civil war (Alesina & La Ferrara, 2005). Tensions in society can spillover into the workplace (Brief et al., 2005). As an example, Protina, a local firm in the Cote D’Ivoire, found that workers on the day shift and night shift tended to come from two different ethnic groups (UNGC, 2009)2. When the groups met during shift changes, there were problems with workplace tensions and conflicts. By bringing the groups together for a dinner, held each night at the shift change, the workers got to know one another. Over time, this dramatically reduced workplace tensions and may have generated positive spillovers to society.

Some studies have shown that diverse workforces are more likely to suffer from poor social integration and have difficulty in building strong interpersonal relationships than homogeneous ones (Hambrick, 1994). It has also been argued that such firms experience more conflicts, have lower levels of trust, lower job satisfaction, high turnover, and low organizational performance (Tsui, Egan, & O’Reilly, 1992). A study about the productivity of coethnic group and non-coethnic group workers in Kenya found that inter-ethnic rivalries negatively affect the allocative efficiency of resources for production in the private sector, which can be also intensified during ethnic conflicts (Hjort, 2014). Likewise, Brief et al. (2005) found that when people experienced more interethnic conflict in their communities they responded negatively to workplace diversity.

This is not an unusual case. Researchers have noted that, “sensitivity to ethnic conflict in the societal context is positively related to ethnic homophily perceptions in the workplace” (Lee & Reade, 2015: 1645). In the absence of managerial intervention, companies can experience a “balkanization” of the workforce (Chua, 2013; Lee & Reade, 2015). Since ethnic homophily is not uncommon, and best practices on managing ethnic tensions in the workplace are not yet widespread (Chua, 2013; Lee & Reade, 2015), we expect that if ethnic diversity is more of a risk as some have argued, then diversity in the workplace will lead to greater workplace tension and lower productivity. Not surprisingly, research in Sri Lanka, a country that has face decades of ethnopolitical conflict, has shown that the more sensitive employees are to ethnic conflict, the lower is their organizational commitment, and presumably their productivity (Reade & Lee, 2012).

While diversity can result in workplace division, studies have also often found that ethnic diversity provides a variety of skills, abilities, perspectives, and cultures that may lead to greater productivity, innovation and creativity at country-level (Alesina & La Ferrara, 2005; Ottaviano & Peri, 2006; Ottaviano & Prarolo, 2009) as well as at the firm-level (Gong, 2003; Richard, 2000), and the value generated from diversity is not easily imitable (Tan & Mahoney, 2006). Of course in some cases, ethnic diversity may be neither beneficial nor detrimental to business. One reason is that in some contexts, no ethnic group may be large enough to legitimize its political power or ideological and religious supremacy (e.g., Esteban et al., 2012; Hale, 2004). Another reason that ethnic identity is likely to be weaker in a highly diverse society is that individuals are most affected by social sanctions exercised by their own ethnic group. This type of sanction will not be as effective in an ethnically diverse society as it will in a less diverse one (Lassen, 2007). Research have found that the benefits of ethnic diversity may increase as the levels of per capita output and consumption increase so high levels of economic properity may also be important for realizing the benefits of diversity (Alesina & La Ferrara, 2005).

Public policies in highly ethnically diverse societies may not favor specific ethnic groups since ethnicity is not a salient social identity. Instead, in ethnically diverse societies each ethnic group may develop different skills and occupational niches depending upon the capital and resources available to them (Esteban et al., 2012). Skills of individuals from different ethnic groups can be complementary in the production process for private goods (Alesina & La Ferrara, 2005). In particular, the benefits of complementarities are realized when the production process is sufficiently diversified (Alesina & La Ferrara, 2005). Thus different skills in production processes may increase employee productivity at the subsidiary level.

Studies have also found that ethnic and cultural diversity have positive effects on knowledge creation, production, and consumption, which will eventually increase human, financial and social capital available in the society (Florida, 2002; Ottaviano & Peri, 2006; Ottaviano & Prarolo, 2009). It is not difficult to predict that abundant public goods or a better institutional context will improve the productivity of employees, and business activities more generally (e.g., Baker, 1990). On the other hand, some have argued that ethnically homogeneous societies, particularly in the case of less developed countries, grow faster than other countries (Easterly & Levine, 1997; Tiemann, Das & DiRienzo, 2006). In homogeneous societies, suggest some, trust is higher and it is easier for people to reach consensus on policy objectives (Tiemann et al., 2006). This, in turn, may lead to a more efficient use of human, financial, and social capital in society. Similar dynamics may occur in the workplace when diversity levels are low. In addition, low ethnic-based risks will enable managers to best utilize the talent of workers without the concern about ethnic-based conflicts in their workplace. For these reasons, we argue that employee productivity, as a measure of organizational productivity (Gong, 2006; Richard, 2000), is likely to be higher when ethnic diversity in society is either quite high or quite low; that is when society is quite diverse or quite homogenous. Thus we would expect that there is a nonlinear U-shaped relationship between ethnic diversity and employee productivity such that:

Hypothesis 2:

When the levels of ethnic diversity in a country are either high or low, one ethnic group is less likely to exclude other groups therefore ethnic conflict risk is low for business. Thus ethnic diversity will follow a U-shaped relationship with employee productivity.

Research Methods

Sample

To test our hypotheses we constructed a unique panel data set from several datasets in regard to ethnicity information, subsidiary characteristics, and country-level characteristics. After merging these datasets, the final sample includes 52,280 observations for the time period 2004–2010. We divided the final sample into a domestic subsidiary sample and a foreign subsidiary sample to investigate whether the effects of ethnic diversity on subsidiary ownership and employee productivity differ between domestic and foreign subsidiaries. Of that sample, 26,488 observations are for domestic subsidiaries and 25,792 are for foreign subsidiaries. The final sample started in 2004 due to the availability of subsidiary characteristic information and ended in 2010 due to the availability of ethnicity information. We note that not many subsidiaries reported their financial information and ownership information for multiple years in the sample period, and therefore the panel data is an unbalanced sample of 30,077 subsidiaries in 63 host countries from 4,003 companies that originated from 79 home countries.

Data

Ethnicity Information Data

The main source of ethnicity information is the Ethnic Power Relation (EPR) project data supplemented by the information from the Minority at Risk (MAR) project data. The EPR project is a collaboration between researchers at the ETH Zurich and the University of California at Los Angeles to identify all politically included or excluded groups and their access to state power. The MAR project is managed by the University of Maryland to analyze the status and conflicts of politically active communal groups in all countries with a current population of at least 500,000. These projects provide information about the name, status, and population of ethnic groups as well as ethnic conflicts among them from the 1940s to the 2000s with detailed information on each politically included or excluded ethnic group (The MAR dataset ends in 2006 while EPR ends in 2010). Research on ethnicity in sociology and political science regularly relies on datasets from the EPR and MAR projects (for several examples among many, see Gurr & Moore 1997; Roessler 2011; Wimmer et al. 2009) that are concerned with the characteristics of each ethnic group and its relationship with conflicts. In this paper, however, we are interested in the ethnic structure in a country and the degree of ethnic diversity at the country-level to compare and analyze subsidiaries’ ownership and employee productivity across countries.

Subsidiary Characteristics Data

We used the OSIRIS dataset to identify subsidiaries of public companies in the world. OSIRIS’ historic CDs provide ownership information on subsidiaries from 2001 and employee productivity information from 2004. The subsidiary ownership information is not available on OSIRIS online so we had to purchase and collect such information from OSIRIS’ historic yearly CDs. Although these CDs include information about the subsidiaries of more than 80,000 companies, the majority of the companies and their subsidiaries do not provide information about the parent firm’s percent of ownership in its subsidiaries, employee productivity, or location information. In the end, we were able to find information for 30,077 subsidiaries from 4,003 companies. Therefore, we acknowledge that there might be unobserved biases in our sample, which could be one of the limitations in our study. We were also able to find matching parent company information from the OSIRIS CDs.

Country-Level Characteristics and Other Control Variables

The World Development Indicators dataset managed by The World Bank is used for variables such as per capita GDP, population, population density, private capital flows, R&D expenditure, unemployment rate, trade, and FDI openness at the country-level. The Correlates of War (COW) dataset is the source of per capita oil production. POLCON database, downloaded from the website of Witold Henisz, is the source for our measure of policy certainty. The Index of Economic Freedom produced by the Heritage Foundation is the source for our measure of legal protection for property rights. To control for industry heterogeneity around ethnicity related risks, we followed Kennedy’s broad industry categories for external environmental risk which are based on the type of products produced by the industry: natural resource-based, intermediate goods, and final goods. Kennedy (1984) noted that, due to the relative capital intensity, the length of pay-back for initial investments, and the degree of political sensitivity, natural resource-based industries are more likely to take a longer-term strategic perspective toward external environmental risks than firms in intermediate goods industries, and firms in intermediate goods industries are more likely than firms in final goods industries to do so.

Measures

Dependent Variables

Our dependent variables are subsidiary ownership for Hypothesis 1 and employee productivity at the subsidiary level for Hypothesis 2. Subsidiary ownership is the total ownership share held by the parent company in a subsidiary. The parent company’s ownership share of a subsidiary is a measure of the level of control held by the parent, the level of resource commitment made by the parent to manage risk, institutional environments, and the legitimacy of the parent company in the host country (Chan & Makino, 2007; Ito & Rose, 1994; Pan, 2017).

Employee productivity at the subsidiary-level is measured by the logarithm of the ratio of total subsidiary revenue to number of subsidiary employees (Datta, Guthrie, & Wright, 2005; Huselid, 1995). Employee productivity is known to be a direct and crucial indicator of organizational performance (Datta et al., 2005; Gong, 2006), especially for analyzing the relationship between ethnic diversity and subsidiary performance (Gong, 2003; Hyun, Oh, & Paik, 2015). Because employee productivity and ownership may be determined simultaneously, we used a simultaneous equation modeling technique. We will discuss the simultaneity issue in the section describing our model.

Independent variables

To test our hypotheses, we computed the entropy measure of diversity using an ethnicity index with multi-dimensional characteristics. Ethnicity, in the measures we employ, is defined as “a subjectively experienced sense of commonality based on a belief in a common ancestry and shared culture. This definition includes ethnolinguistic, ethnosomatic, and ethnoreligious groups but not tribes and clans that conceive of ancestry in genealogical terms, nor regions that do not define commonality on the basis of shared ancestry” (Wimmer et al. 2009, p. 325). Perceptions of ethnic group members and non-members vary across societies. In the Supplementary Appendix, we have explained how the MAR and EPR datasets identified ethnic groups (with examples) and how those groups have changed over time and are different across countries.

Typically a Herfindahl-type diversity measure is used in the literature (Alesina et al., 1999; Collier, 2000; Collier & Hoeffler, 1998; Easterly & Levine, 1997; Lassen, 2007; Luiz, 2015; Ottaviano & Peri, 2006; Richard, 2000) and sometimes called it ethnic fractionalization, but we used the entropy measure to take into account the theoretical differences between included and excluded groups, and groups without political and economic influence (the group that does not have any ethnically-based influence in politics). An ethnic group is considered politically included if, “at least one political organization has claimed to represent its interests at the national level,” or excluded, “if its members are subjected to state-led political discrimination” (Cederman, Wimmer, and Min 2010: 99). Included ethnic groups are those that control or have access to the central government, and excluded groups are ethnic groups whose political leaders are excluded from participation in the central government (Cederman et al., 2010). A third ethnic group includes those groups that do not have any political or economic influence in the country and whose interests are not formally represented in government.

Studies (e.g., Cederman et al., 2010; Wimmer et al., 2009) show that ethno-politically included and excluded groups, and those groups without influence, experience and respond to ethnic conflict risk very differently. Included groups fight each other in a struggle over the spoils (infighting), excluded groups fight to shift the boundaries of inclusion (rebellions), and both included and excluded groups may attempt to change territorial boundaries (secession). A group that does not have any ethnically-based influence in politics is unlikely to influence the outcome but it will be directly or indirectly affected by any of the three (i.e., rebellions, infighting, and secession). Depending upon the outcome, ethnic groups without political influence may be either victims or beneficiaries of conflicts.

The entropy measure is widely used in workgroup diversity research to form an aggregate measure of categorical dimensions of diversity within workgroups (e.g., Choi, 2007; Jehn, Northcraft, & Neale, 1999; Østergaard, Timmermans, & Kristinsson, 2011). The entropy measure of diversity not only considers relative population size across ethnic groups, but also differences across included and excluded groups, and ethnic groups without political influence. Thus the entropy measure assumes that changes in relative size within an included- (or excluded-) group category are more relevant (less diversified) than changes across categories. We draw readers’ attention to Teachman (1980) for more information about the entropy measure. In short, ethnic diversity was measured by using the following formula.
$$Ethnic\;diversity = \sum\limits_{j = 1}^{3} {\sum\limits_{i \in j} {P_{i}^{j} \ln (1/P_{i}^{j} )P^{j} + \sum\limits_{j \in 1}^{3} {P^{j} } \ln (1/P^{j} ),} }$$
where i is an ethnic group of an ethnic category j (i.e., j is an included ethnic group, excluded ethnic group, or an ethnic group that lacks political influence). Pij is defined as the share of the ethnic group i of ethnic group category j in the total populations of the ethnic category. In other words, P i j  = Pi//Pj and \(P^{j} = \sum_{j \in 1}^{3} {P_{i}^{j} } .\) We tested our model with the Herfindahl-type diversity measure as a robustness check. In Table 1, we show total ethnic diversity, within included group diversity, within excluded group diversity, and across category diversity of each country in the analysis.
Table 1

Ethnic diversity across countries

Country

Ethnic diversity

Within included group diversity

Within excluded group diversity

Across category diversity

Albania

0.20

0.00

0.56

0.18

Algeria

0.59

0.00

0.00

0.59

Argentina

0.20

0.00

0.00

0.20

Australia

0.35

0.00

0.00

0.35

Belgium

0.72

0.67

0.00

0.06

Bosnia and Herzegovina

1.05

1.00

0.00

0.06

Brazil

0.90

0.00

0.24

0.88

Bulgaria

0.60

0.33

0.00

0.29

Canada

0.75

0.59

0.00

0.26

Chile

0.21

0.00

0.00

0.21

China

1.92

0.00

2.84

0.68

Colombia

0.64

0.00

0.25

0.58

Costa Rica

0.15

0.00

0.00

0.15

Croatia

0.32

0.05

0.93

0.23

Czech Republic

0.26

0.00

0.69

0.22

Ecuador

0.77

0.00

0.49

0.62

Egypt

0.31

0.00

0.00

0.31

El Salvador

0.00

0.00

0.00

0.00

Estonia

0.75

0.00

0.44

0.62

France

0.13

0.00

0.98

0.11

Germany

0.00

0.00

0.00

0.00

Greece

0.14

0.00

0.66

0.13

Hungary

0.24

0.00

0.00

0.24

India

2.26

2.27

1.55

0.07

Indonesia

1.35

0.00

1.48

0.93

Ireland

0.00

0.00

0.00

0.00

Israel

1.57

1.04

0.69

0.69

Italy

0.04

0.00

0.62

0.04

Japan

0.13

0.00

1.10

0.11

Jordan

0.68

0.00

0.00

0.68

Kuwait

0.87

0.62

0.00

0.65

Kyrgyzstan

0.86

0.00

0.82

0.64

Latvia

0.89

0.00

0.60

0.68

Lithuania

0.51

0.00

0.69

0.42

Malaysia

1.06

0.84

0.69

0.26

Mexico

0.41

0.00

0.00

0.41

Morocco

0.37

0.00

0.00

0.37

Netherlands

0.37

0.34

0.00

0.05

New Zealand

0.48

0.48

0.00

0.00

Norway

0.00

0.00

0.00

0.00

Pakistan

1.24

0.71

1.23

0.44

Panama

0.53

0.00

1.06

0.40

Paraguay

0.13

0.00

0.00

0.13

Peru

0.77

0.00

0.11

0.72

Philippines

0.54

0.00

0.94

0.41

Poland

0.08

0.00

0.83

0.07

Portugal

0.00

0.00

0.00

0.00

Romania

0.38

0.25

0.26

0.14

Russia

1.07

0.00

2.91

0.56

Saudi Arabia

1.08

0.68

0.36

0.60

Slovenia

0.38

0.00

1.31

0.31

South Africa

1.64

1.64

0.00

0.00

South Korea

0.00

0.00

0.00

0.00

Spain

0.96

0.00

1.13

0.62

Sri Lanka

0.92

0.64

0.00

0.36

Sweden

0.00

0.00

0.00

0.00

Switzerland

0.80

0.77

0.00

0.12

Trinidad and Tobago

0.73

0.00

0.00

0.73

Tunisia

0.00

0.00

0.00

0.00

Ukraine

0.58

0.47

1.02

0.11

United Kingdom

0.74

0.43

1.30

0.25

United States

0.95

0.00

1.14

0.62

Uruguay

0.28

0.00

0.00

0.28

Note: Average values for 2004–2010. The table is only for the purpose of providing information. Because our empirical models consider simultaneity between employee productivity and subsidiary ownership and control for region-, host country-, industry-, parent company-, and subsidiary-level variables and fixed effects, interpreting or comparing nominal values of ethnic diversity across country is not appropriate.

Control variables

As discussed above we controlled for subsidiary, parent company, and host country characteristics. For subsidiary characteristics, we included subsidiary-level employee productivity when the dependent variable is subsidiary ownership, and subsidiary ownership when the dependent variable is subsidiary-level employee productivity. For parent company characteristics we included company revenue (logged), the number of total subsidiaries (logged), and the number of shareholders (logged). For host country characteristics, we included per capita GDP (logged), population density (population divided by land size), private capital investment intensity (private capital flows divided by GDP;  %), R&D investment intensity (R&D expenditure divided by GDP;  %), the unemployment rate (percentage of total population), per capita oil production, import openness (import divided by GDP), FDI openness (FDI inflows divided by GDP), policy certainty (POLCON index), property rights (index), the portion of ethnically included or excluded population to the total population, number of people killed in terrorism (logged) and on-going wars in the country (dummy). We also included an indicator for domestic (versus foreign) subsidiaries and industry dummy variables for natural resource-based, intermediate goods, and final goods industry sectors. Finally, dummy variables for region (Africa, Asia, Europe, Middle East, North America, and South America) and year fixed effects were included.

Model

To test our models we used three-stage least squares regression modeling to test the simultaneous equation model. Our hypotheses can be tested with the following system of equations:
$$Ownership_{i,j,k,t} = a_{0} + a_{1} \times Productivity_{i,j,k,t} + a_{2} ED_{k,t - 1} + a_{3} ED_{k,t - 1}^{2} + \sum\limits_{k} {a_{k} \times COM_{k,j,t - 1} + \sum\limits_{l} {a_{l} } } \times CTY_{l,j,t - 1} + IND_{i \in L} + REG_{k \in r} + YEAR_{t} + e_{1,i,j,k,t}$$
(1)
$$Productivity_{i,j,k,t} = a_{0} + a_{1} \times Ownership_{i,j,k,t} + a_{2} ED_{k,t - 1} + a_{3} ED_{k,t - 1}^{2} + \sum\limits_{m} {a_{m} \times COM_{m,j,t - 1} + \sum\limits_{n} a } \times CTY_{n,j,t - 1} + IND_{i \in L} + REG_{k \in r} + YEAR_{t} + e_{2,i,j,k,t} ,$$
(2)
where i is subsidiary, j is a parent firm of i, k is host country, t is time, l is industry, and r is region. Ownership is company k’s total percentage in subsidiary i at time t. Employee Productivity is the revenue divided by number of employees of subsidiary i at time t. ED is an entropy measure of ethnic diversity of the host country k at time t. COM and CTY are company- and host country-level control variables, including identification variables, respectively. IND, REG, and YEAR are industry, region, and year fixed dummy variables, e1 and e2 are error terms.

Because Ownership and Employee Productivity can be determined simultaneously, e1 and e2 are correlated. When a parent company’s ownership of its subsidiary is high, the parent company can control and manage resources that have been committed to the subsidiary (Anderson & Gatignon 1986; Kim & Hwang, 1992; Woodcock, Beamish, & Makino, 1994). Thus the parent company likely provides more tangible and intangible resources to its subsidiary, which eventually improves employee productivity at the subsidiary level (Gaur & Lu, 2007; Woodcock et al., 1994). In regard to ownership, on the one hand, when a subsidiary’s profitability is high, its parent company is likely able to find additional investors. Thus the parent company may have a low ownership level toward the more productive subsidiary. On the other hand, insiders (the parent company) will have more information about the subsidiary, and thus the parent company may be able to increase its level of ownership when its subsidiary is expected to show better performance (Chang, 2003). Also, the potential for higher profits creates a demand for closer monitoring of management by owners (Demsetz & Lehn, 1985). Thus firm performance and ownership structure endogenously determine each other. In such circumstances, the literature suggests simultaneous modeling as a proper estimation method (Chang, 2003; Cho, 1998; Demsetz & Villalonga, 2001) since OLS is a biased estimator (Greene, 2003). The simultaneous equation modeling constructs the set of instruments for the endogenous variables from exogenous variables. Since there are potential endogeneity and causality concerns for all independent and control variables, these variables (e.g., variables treated as exogenous) are lagged 1 year.

We used three-stage least squares (3SLS) estimation to test the simultaneous equation model described above. 3SLS is a full information maximum likelihood estimator. We also tested the model with two-stage least squares (2SLS). Both results were consistent and improved the robustness of our findings. Since 3SLS is more consistent and asymptotically efficient compared to 2SLS (Greene, 2003; Kumar, 2009), we present the results from 3SLS throughout the paper. To test the simultaneous equations models, we needed to include at least one variable (i.e., identification variable) that does not enter into the other equation. We have explained the identification procedure in the Supplementary Appendix in a greater detail.

Results

The summary statistics and correlation matrix for the total sample appear in Table 2. Those for the sub-samples (i.e., domestic subsidiaries; foreign subsidiaries) are available upon request. We checked variance inflation factors (VIFs) in OLS estimations separately. The average VIFs are: 4.14 for the ownership model and 4.37 for the employee productivity model in the total sample; 4.78 for the ownership model and 4.96 for the employee productivity model in the domestic subsidiary sample; 4.04 for the ownership model and 4.42 for the employee productivity model in the foreign subsidiary sample. Regarding VIFs for individual variables, except for the linear and quadratic terms of ethnic diversity variables, all VIFs for individual variables are less than 8. When we use only the linear term for the ethnic diversity variables, the average VIFs are less than 3 and VIFs for the individual variables are less than 7. Thus both the correlation matrix and VIFs do not show any symptoms of multicollinearity, so we conclude that multicollinearity is not a concern in our analysis.
Table 2

Summary statistics and correlation matrix

Variable

Mean

S.D.

1

2

3

4

5

6

7

8

9

10

1. Subsidiary ownership

0.559

0.407

          

2. Employee productivity (log)

5.378

1.991

−0.072

         

3. Ethnic diversity

0.512

0.418

−0.238

−0.099

        

4. Portion of ethically included and excluded pop.

0.830

0.354

−0.143

−0.031

0.494

       

5. Number of killed in terrorism (log)

0.142

0.643

0.072

−0.043

0.171

0.097

      

6. On-going war (dummy)

0.030

0.169

0.034

−0.134

0.222

0.067

0.082

     

7. Domestic subsidiary (dummy)

0.507

0.500

0.171

−0.107

−0.191

−0.056

0.025

0.115

    

8. Per capita GDP (log)

10.383

0.541

−0.152

0.128

−0.050

−0.035

−0.088

−0.409

−0.084

   

9. Population density

4.467

1.130

0.108

0.160

−0.415

−0.045

0.034

−0.301

0.036

0.071

  

10. R&D expenditure intensity

2.031

0.774

−0.210

0.074

−0.083

−0.242

−0.175

−0.213

−0.057

0.574

0.005

 

11. Unemployment rate

6.964

2.613

0.277

−0.033

−0.146

−0.080

0.093

0.015

0.190

−0.334

0.071

−0.405

12. FDI openness

3.646

4.415

0.037

0.002

0.139

0.100

0.064

0.019

−0.102

−0.032

0.091

−0.225

13. Trade openness

30.311

13.632

0.255

0.041

−0.268

−0.279

−0.012

−0.100

−0.017

−0.125

0.338

−0.269

14. Per capita oil production

1.037

3.514

−0.026

−0.028

0.013

−0.200

−0.019

0.095

−0.032

0.160

−0.357

−0.071

15. Policy certainty

0.432

0.102

0.134

0.071

−0.444

−0.077

−0.026

−0.136

0.052

0.305

0.308

0.307

16. Private capital intensity

1.205

6.431

−0.137

−0.035

0.416

0.256

0.114

−0.033

−0.188

−0.114

−0.034

−0.175

17. Property rights

76.456

16.873

−0.175

0.075

0.136

−0.222

−0.056

−0.489

−0.227

0.751

0.008

0.599

18. Parent company revenue (log)

15.507

2.330

−0.244

0.162

0.203

0.153

−0.007

0.001

−0.279

0.062

−0.095

0.065

19. Number of subsidiaries (log)

4.611

2.098

−0.509

0.115

0.283

0.180

−0.048

−0.061

−0.365

0.245

−0.095

0.150

20. Number of shareholders (log)

3.044

1.115

−0.089

0.061

0.194

0.121

0.087

−0.127

−0.104

0.106

−0.029

−0.057

Variable

Mean

S.D.

11

12

13

14

15

16

17

18

19

1. Subsidiary ownership

0.559

0.407

         

2. Employee productivity (log)

5.378

1.991

         

3. Ethnic diversity

0.512

0.418

         

4. Portion of ethically included and excluded pop.

0.830

0.354

         

5. Number of killed in terrorism (log)

0.142

0.643

         

6. On-going war (dummy)

0.030

0.169

         

7. Domestic subsidiary (dummy)

0.507

0.500

         

8. Per capita GDP (log)

10.383

0.541

         

9. Population density

4.467

1.130

         

10. R&D expenditure intensity

2.031

0.774

         

11. Unemployment rate

6.964

2.613

         

12. FDI openness

3.646

4.415

−0.018

        

13. Trade openness

30.311

13.632

0.169

0.554

       

14. Per capita oil production

1.037

3.514

−0.235

0.004

−0.090

      

15. Policy certainty

0.432

0.102

0.001

0.169

0.386

−0.001

     

16. Private capital intensity

1.205

6.431

0.003

0.108

−0.132

−0.240

−0.387

    

17. Property rights

76.456

16.873

−0.311

0.051

−0.002

0.120

0.200

0.035

   

18. Parent company revenue (log)

15.507

2.330

−0.115

−0.024

−0.153

−0.044

−0.101

0.093

0.061

  

19. Number of subsidiaries (log)

4.611

2.098

−0.191

0.059

−0.179

−0.015

−0.095

0.191

0.212

0.607

 

20. Number of shareholders (log)

3.044

1.115

−0.011

0.030

−0.032

−0.010

−0.102

0.094

0.104

0.415

0.272

Note: N = 52,280. Correlations above |0.008| are significant at p < 0.5; correlations above |0.011| are significant at p < 0.01.

Table 3 shows our main results. In Columns 1 and 2, we test our model with only control variables for the total sample. The results show that employee productivity at the subsidiary level negatively affects the parent company’s ownership level in a subsidiary. Thus when a subsidiary is productive, the parent company is able to find additional investors. Subsidiary ownership positively affects subsidiary employee productivity. As a result, the parent company may commit more resources to the subsidiary when it has higher levels of ownership toward the subsidiary. Most of the control variables are statistically significant. There are some important findings that we want to note. First, the portion of the ethnically included or excluded population is negative and significant for both ownership and employee productivity models when we do not add ethnic diversity. Terrorist activities do not affect ownership levels, but they do reduce employee productivity in a subsidiary. On-going war reduces both ownership levels and employee productivity in a subsidiary. Thus ethnic and other conflicts increase the business risk in society.
Table 3

Ethnic diversity, subsidiary ownership, and employee productivity

Dependent variable

Control only

Ethnic diversity

Total sample

Linear model

Quadratic model

Total sample

Total sample

Foreign subsidiary sample

Domestic subsidiary sample

Ownership

Productivity

Ownership

Productivity

Ownership

Productivity

Ownership

Productivity

Ownership

Productivity

Ethnic diversity

  

0.0100

−0.0439

−0.2428***

−0.8409***

−0.1168*

−1.0024***

−0.2058***

−1.0315***

  

(0.0106)

(0.0447)

(0.0381)

(0.1182)

(0.0559)

(0.1394)

(0.0519)

(0.2034)

Ethnic diversity square

    

0.2011***

0.6424***

0.0989*

0.7734***

0.1886***

0.7616***

    

(0.0287)

(0.0909)

(0.0420)

(0.0975)

(0.0385)

(0.2137)

Employee productivity (log)

−0.2023***

 

−0.2021***

 

−0.2219***

 

−0.1893***

 

−0.1636***

 

(0.0199)

 

(0.0199)

 

(0.0229)

 

(0.0416)

 

(0.0283)

 

Subsidiary ownership

 

1.9188**

 

1.9766***

 

1.4583*

 

0.6042

 

−0.9129

 

(0.5883)

 

(0.5970)

 

(0.6068)

 

(0.6063)

 

(2.1010)

Portion of ethnically included and excluded population

−0.0526***

−0.2411***

−0.0588***

−0.2109***

−0.0117

−0.0542

−0.0808***

−0.0653

0.0021

0.1062

(0.0103)

(0.0388)

(0.0123)

(0.0508)

(0.0136)

(0.0518)

(0.0163)

(0.0825)

(0.0179)

(0.0940)

Number of killed in terrorism (log)

0.0036

−0.0958***

0.0028

−0.0929***

0.0038

−0.0774***

−0.0071

−0.0392*

0.0169**

−0.0458

(0.0039)

(0.0171)

(0.0040)

(0.0172)

(0.0042)

(0.0172)

(0.0050)

(0.0187)

(0.0055)

(0.0636)

On-going war

−0.2587***

−0.4697***

−0.2666***

−0.4280***

−0.2364***

−0.3607**

−0.1463***

0.0777

−0.1305***

−0.4469*

(0.0255)

(0.1123)

(0.0266)

(0.1239)

(0.0268)

(0.1189)

(0.0313)

(0.1577)

(0.0313)

(0.2118)

Domestic subsidiary

−0.1213***

−0.2150***

−0.1213***

−0.2117***

−0.1281***

−0.2423***

    

(0.0084)

(0.0368)

(0.0084)

(0.0373)

(0.0095)

(0.0376)

    

Per capita GDP (log)

0.1545***

0.9045***

0.1565***

0.8935***

0.1850***

0.9175***

0.2016***

0.7572***

0.1265***

0.9417***

(0.0200)

(0.0429)

(0.0201)

(0.0441)

(0.0237)

(0.0429)

(0.0352)

(0.0576)

(0.0325)

(0.1047)

Population density (log)

 

0.2493***

 

0.2509***

 

0.2212***

 

0.1188***

 

0.1520*

 

(0.0280)

 

(0.0282)

 

(0.0288)

 

(0.0223)

 

(0.0690)

R&D expenditure intensity

0.0064

−0.1284***

0.0057

−0.1277***

−0.0092

−0.1613***

−0.0392***

−0.1513***

0.0138

−0.1004

(0.0077)

(0.0312)

(0.0077)

(0.0313)

(0.0089)

(0.0304)

(0.0111)

(0.0362)

(0.0126)

(0.0780)

Unemployment rate

0.0099***

 

0.0097***

 

0.0091***

 

0.0081***

 

0.0031†

 

(0.0011)

 

(0.0011)

 

(0.0012)

 

(0.0013)

 

(0.0017)

 

FDI openness

−0.0027***

 

−0.0028***

 

−0.0027***

 

−0.0036***

 

−0.0002

 

(0.0006)

 

(0.0006)

 

(0.0006)

 

(0.0007)

 

(0.0007)

 

Trade openness

0.0020***

0.0043***

0.0020***

0.0045***

0.0025***

0.0060***

0.0040***

0.0008

0.0002

0.0117*

(0.0003)

(0.0011)

(0.0003)

(0.0011)

(0.0004)

(0.0011)

(0.0004)

(0.0019)

(0.0006)

(0.0051)

Per capita oil production

−0.0043***

0.0062

−0.0045***

0.0069

−0.0052***

0.0018

−0.0033*

−0.0136**

−0.0061***

−0.0047

(0.0009)

(0.0048)

(0.0009)

(0.0050)

(0.0010)

(0.0050)

(0.0014)

(0.0043)

(0.0013)

(0.0167)

Policy certainty

0.2245***

−0.4463*

0.2396***

−0.5284*

0.2521***

−0.3602

0.1764**

0.6960***

0.2300***

−0.3552

(0.0324)

(0.2072)

(0.0362)

(0.2309)

(0.0386)

(0.2314)

(0.0571)

(0.1873)

(0.0512)

(0.7749)

Private capital intensity

−0.0011**

 

−0.0012**

 

−0.0008*

 

−0.0005

 

0.0003

 

(0.0004)

 

(0.0004)

 

(0.0004)

 

(0.0004)

 

(0.0004)

 

Property rights

−0.0026***

−0.0093***

−0.0027***

−0.0086***

−0.0018***

−0.0052***

−0.0041***

−0.0013

0.0011*

−0.0024

(0.0003)

(0.0014)

(0.0004)

(0.0016)

(0.0004)

(0.0016)

(0.0004)

(0.0029)

(0.0005)

(0.0045)

Parent company revenue (log)

0.0495***

0.1072***

0.0495***

0.1057***

0.0522***

0.1154***

0.0599***

0.1436***

0.0434***

0.1669***

(0.0032)

(0.0131)

(0.0032)

(0.0134)

(0.0036)

(0.0135)

(0.0070)

(0.0196)

(0.0045)

(0.0408)

Number of subsidiaries (log)

−0.0950***

0.1841**

−0.0952***

0.1904***

−0.0952***

0.1407*

−0.1171***

0.0738

−0.0725***

−0.1117

(0.0017)

(0.0566)

(0.0017)

(0.0576)

(0.0018)

(0.0585)

(0.0023)

(0.0720)

(0.0031)

(0.1346)

Number of shareholders (log)

0.0202***

−0.1160***

0.0199***

−0.1164***

0.0190***

−0.0996***

0.0002

−0.1386***

0.0191***

0.0394

(0.0028)

(0.0211)

(0.0028)

(0.0211)

(0.0030)

(0.0213)

(0.0063)

(0.0196)

(0.0034)

(0.0346)

Natural resource industry

0.1554***

−0.2133*

0.1547***

−0.2184*

0.1573***

−0.1398

0.1817***

−0.1281

0.1050***

0.1523

(0.0083)

(0.0924)

(0.0083)

(0.0929)

(0.0088)

(0.0941)

(0.0145)

(0.1278)

(0.0094)

(0.2040)

Intermediate goods industry

0.1610***

−0.3565***

0.1606***

−0.3646***

0.1617***

−0.2722**

0.2086***

−0.1151

0.1060***

0.0807

(0.0051)

(0.1018)

(0.0051)

(0.1029)

(0.0055)

(0.1047)

(0.0070)

(0.1287)

(0.0064)

(0.2314)

Region fixed effects

YES

YES

YES

YES

YES

YES

YES

YES

YES

YES

Year fixed effects

YES

YES

YES

YES

YES

YES

YES

YES

YES

YES

Number of observations

52,280

 

52,280

 

52,280

 

25,792

 

26,488

 

Log likelihood

−137,867

 

−138,287

 

−135,326

 

−56,037

 

−60,491

 

Akaike Criterion

275,855

 

276,699

 

270,781

 

112,197

 

121,107

 

Hansen–Sargan test of over-identification: χ2 (p value)

1.761

(0.415)

 

1.342

(0.511)

 

0.793

(0.673)

 

2.015

(0.365)

 

0.170

(0.428)

 

Note: p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001. Two-tailed test. Two dependent variables (i.e., ownership and productivity) are estimated simultaneously by using 3SLS. Final goods industry dummy was included but dropped in estimation due to multicollinearity.

In Columns 3 and 4 we included the linear term of ethnic diversity. The linear term of ethnic diversity is insignificant in both the ownership model (β = 0.0100, n.s.) and the employee productivity model (β = −0.0439, n.s.). The results for all other control variables only show minor differences from those in Columns 1 and 2. In Columns 5 and 6 we tested the quadratic model of ethnic diversity. The results show that ethnic diversity has a U-shaped relationship with both ownership and employee productivity as expected from our hypotheses. The second order terms of ethnic diversity are positive and significant in the ownership model (β = 0.2011, p < 0.001) and in the employee productivity model (β = 0.6424, p < 0.001). It confirms that Hypotheses 1 and 2 are strongly supported. The U-shaped relationship is economically very important for managers. When we change one standard deviation (SD) of ethnic diversity from the vertex point (i.e., 0.605, which is close to the mean value, 0.512) of the quadratic relationship with subsidiary ownership, subsidiary ownership increases by 8.7% of its SD. This increase is more than changes from other important variables such as unemployment rate, FDI openness, trade openness, and policy certainty. Likewise, one SD change of ethnic diversity from the vertex point (i.e., 0.654) of the quadratic relationship with employee productivity, results in an employee productivity increase of 5.6% of its SD, more than changes from other important variables such as R&D expenditure intensity, trade openness and policy certainty. Thus the substantial positive effects of ethnic diversity on both subsidiary ownership and employee productivity beyond the vertex point are in favor of the positive effect of ethnic diversity.

In Columns 7–10, we divided the total sample into foreign subsidiaries and domestic subsidiaries. The results show that ethnic diversity has a U-shaped relationship with both ownership and employee productivity for foreign and domestic subsidiaries. The second order terms of ethnic diversity are positive and significant for the foreign subsidiary sample (i.e., β = 0.0989, p < 0.05 in the ownership model; β = 0.7734, p < 0.001 in the employee productivity model) and for the domestic subsidiary sample (i.e., β = 0.1886, p < 0.001 in the ownership model; β = 0.7616, p < 0.001 in the employee productivity model). When we change one SD of ethnic diversity from the vertex point of the quadratic relationship with subsidiary ownership, subsidiary ownership increases by 4.2% of its SD for the foreign subsidiary sample and by 8.3% for the domestic subsidiary sample. Additionally, a one SD change of ethnic diversity from the vertex point of the quadratic relationship with employee productivity results in an employee productivity increase of 6.8% of its SD for the foreign subsidiary sample and 6.6% for the domestic subsidiary sample. Thus Hypotheses 1 and 2 are consistently supported in foreign and domestic subsidiary sub-samples.

As discussed in the Measures section above, the datasets enable us to divide ethnic diversity into a within included-group category, a within excluded-group category, and an across-category diversity group because diversity within and between group categories may have different relationships to and effects on subsidiary ownership and employee productivity. We have tested these variables as an ad-hoc analysis and the results are shown in Table 4. The results show that ethnic diversity within the included-group category has a U-shaped relationship with subsidiary ownership for both foreign and domestic subsidiaries and with employee productivity for foreign subsidiaries, while ethnic diversity within the excluded-group category does not have a strong, consistent effect on ownership and employee productivity at the subsidiary-level across samples. Across-categories ethnic diversity has a U-shaped relationship with employee productivity for both domestic and foreign subsidiaries but not for subsidiary ownership. We will discuss these findings in more detail in the Discussion section.
Table 4

Within included-group category, within excluded-group category and across-category diversity

Dependent variable

Total sample

Foreign subsidiary sample

Domestic subsidiary sample

Ownership

Productivity

Ownership

Productivity

Ownership

Productivity

Within included-group diversity

−0.3250***

−0.8940***

−0.2099***

−1.0525***

−0.2036***

−0.4975

(0.0357)

(0.2249)

(0.0470)

(0.1875)

(0.0495)

(0.5530)

Within included-group diversity square

0.2195***

0.6364***

0.1549***

0.6867***

0.1494**

0.3545

(0.0338)

(0.1816)

(0.0386)

(0.1833)

(0.0484)

(0.4440)

Within excluded-group diversity

−0.1122***

−0.1700

−0.0665**

−0.1365

−0.0508

−0.4072*

(0.0206)

(0.1247)

(0.0210)

(0.1208)

(0.0328)

(0.1612)

Within excluded-group diversity square

0.0213*

−0.0041

−0.0023

0.0009

0.0284*

0.1224†

(0.0083)

(0.0391)

(0.0085)

(0.0398)

(0.0119)

(0.0683)

Across-category diversity

−0.1296

−1.9809***

−0.0946

−1.2886**

−0.2041

−2.2211**

(0.1220)

(0.4367)

(0.1282)

(0.4303)

(0.1719)

(0.8122)

Across-category diversity square

0.2562†

2.6260***

0.2697†

1.8752***

0.2705

2.6730**

(0.1494)

(0.4829)

(0.1566)

(0.4960)

(0.2038)

(0.8948)

Employee productivity (log)

−0.1916***

 

−0.1402***

 

−0.1539***

 

(0.0271)

 

(0.0413)

 

(0.0330)

 

Subsidiary ownership

 

−0.5868

 

−1.1677

 

−1.3312

 

(0.8908)

 

(1.0135)

 

(2.8148)

Portion of ethnically included or excluded groups relative to the

population

0.0956***

0.3261***

0.0162

0.1140

0.0482

0.4900**

(0.0204)

(0.0818)

(0.0205)

(0.0909)

(0.0331)

(0.1629)

Number of killed in terrorism (log)

0.0027

−0.0489**

−0.0072

−0.0476*

0.0182***

−0.0273

(0.0039)

(0.0180)

(0.0045)

(0.0188)

(0.0054)

(0.0805)

On-going war

−0.1798***

−0.3133†

−0.0840**

−0.0491

−0.1893***

−0.5696

(0.0287)

(0.1667)

(0.0318)

(0.1725)

(0.0420)

(0.4080)

Domestic subsidiary

−0.1197***

−0.3597***

    

(0.0103)

(0.0528)

    

Per capita GDP (log)

0.1147***

0.6853***

0.1296***

0.7113***

0.1008***

0.7240***

(0.0208)

(0.0488)

(0.0288)

(0.0612)

(0.0290)

(0.0950)

Population density (log)

 

0.1286***

 

0.0901***

 

0.1332†

 

(0.0311)

 

(0.0257)

 

(0.0798)

R&D expenditure intensity

0.0109

−0.0694†

−0.0156

−0.1160**

0.0172

−0.0612

(0.0081)

(0.0377)

(0.0096)

(0.0364)

(0.0122)

(0.0997)

Unemployment rate

0.0043***

 

0.0038**

 

0.0024

 

(0.0012)

 

(0.0013)

 

(0.0018)

 

FDI openness

−0.0009*

 

−0.0017**

 

−0.0004

 

(0.0004)

 

(0.0006)

 

(0.0007)

 

Trade openness

0.0012**

0.0050***

0.0030***

0.0063*

0.0005

0.0043

(0.0004)

(0.0015)

(0.0005)

(0.0025)

(0.0006)

(0.0032)

Per capita oil production

−0.0031***

−0.0003

−0.0008

−0.0064

−0.0052***

−0.0045

(0.0009)

(0.0050)

(0.0010)

(0.0041)

(0.0013)

(0.0188)

Policy certainty

0.1247**

−0.3849

0.0135

0.1068

0.1488**

−0.4259

(0.0389)

(0.2614)

(0.0405)

(0.1874)

(0.0574)

(0.8239)

Private capital intensity

−0.0000

0.0003

−0.0027***

−0.0022

0.0020**

0.0036

(0.0004)

(0.0017)

(0.0004)

(0.0036)

(0.0007)

(0.0062)

Property rights

−0.0006*

 

−0.0003

 

0.0000

 

(0.0003)

 

(0.0003)

 

(0.0004)

 

Parent company revenue (log)

0.0479***

0.1570***

0.0519***

0.1959***

0.0414***

0.1715**

(0.0042)

(0.0188)

(0.0069)

(0.0307)

(0.0051)

(0.0543)

Number of subsidiaries (log)

−0.0960***

−0.0562

−0.1179***

−0.1384

−0.0709***

−0.1322

(0.0017)

(0.0860)

(0.0019)

(0.1198)

(0.0030)

(0.1797)

Number of shareholders (log)

0.0186***

−0.0435

0.0074

−0.0912**

0.0175***

0.0463

(0.0030)

(0.0285)

(0.0058)

(0.0278)

(0.0034)

(0.0390)

Natural resource based industry

0.1463***

0.1243

0.1821***

0.1962

0.1006***

0.1815

(0.0080)

(0.1276)

(0.0126)

(0.1974)

(0.0092)

(0.2610)

Intermediate goods industry

0.1565***

0.0540

0.2067***

0.2505

0.1022***

0.1135

(0.0050)

(0.1483)

(0.0060)

(0.2098)

(0.0063)

(0.3009)

Region fixed effects

YES

YES

YES

YES

YES

YES

Year fixed effects

YES

YES

YES

YES

YES

YES

Number of observations

52,280

 

25,792

 

26,488

 

Log likelihood

−114,341

 

−48,566

 

−58,673

 

Akaike Information Criterion

228,825

 

97,272

 

117,487

 

Hansen–Sargan test of over-identification: χ2 (P-value)

2.595

(0.273)

 

2.040

(0.361)

 

1.608

(0.448)

 

Note: p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001. Two-tailed test. Two dependent variables (i.e., ownership and productivity) are estimated simultaneously by using 3SLS. Final goods industry dummy was included but dropped in estimation due to multicollinearity.

Robustness Checks

We tested the robustness of our findings in multiple ways. First, we tested our model with two-stage least squares (2SLS) estimation; a limited information maximum likelihood estimation. Three-stage least square (3SLS) has a potential misspecification problem because misspecification of one equation can contaminate parameter estimates in other equations (Judge, Hill, Lutkepohl, & Lee, 1985). However, the misspecification of one equation in 2SLS estimation does not affect the results of the other equation in the estimation (Baltagi, 2002). Therefore, if the results of 3SLS and 2SLS are consistent, equations in simultaneous models are well specified. We found that the results from 3SLS and 2SLS provide consistent results for our model across samples.

Second, because our data has a multi-level structure, one may argue that ignoring that structure (where subsidiaries are nested within a country) in the empirical analysis may generate biased estimates of the standard errors. Because no statistical method has been provided for simultaneous modeling in 3SLS, we explicitly generate the process of 2SLS in a multi-level random effects model. The results are consistent, and coefficients show about the same statistical significance and signs. We presume that because we controlled for industry-, region-, and year-fixed effects, the within group characteristics are not very different than across group characteristics.

Third, we used the Herfindahl-based measure of ethnic diversity (ethnic fractionalization; Luiz, 2015) instead of the entropy-based measure. We found that the statistical results are roughly identical although the information captured by each measure is quite different. In our research context the entropy measure has both methodological and theoretical benefits over the Herfindahl-based measure because it reveals the degree of political and economic exclusion or marginalization of minority ethnic groups. We have included the results of three robustness checks in the Supplementary Appendix.

Discussion

The purpose of this paper was to explore the relationship between country-level ethnic diversity, ownership strategy, and employee productivity to reveal how managers view ethnic diversity when formulating subsidiary-level ownership strategy and how ethnic diversity in society may affect employee productivity. Using the detailed data on country ethnicity, as well as advanced econometric analysis that considers the simultaneity between ownership and employee productivity at the subsidiary-level, our multi-country study provides new insights into the literature on firm strategy, structure, and country-level conditions. Our findings suggest that high levels of diversity are not a de facto form of country risk as some have argued. Rather, managers – at least in our study – increase subsidiary ownership at low and high levels of ethnic diversity and reduce ownership at moderate levels. The relationship between country-level ethnic diversity and employee productivity follows a similar pattern. Thus it appears to be the ability of a dominant ethnic group to exclude other groups from full economic and political participation that leads to greater risk. At high and low levels of ethnic diversity in a society, the ability of one group to exclude others is less likely.

To elaborate, in our main findings we found that managers do not appear to respond to ethnic diversity as a form of country risk, at least in terms of their ownership choices. Managers do not lower their ownership levels in subsidiaries simply because ethnic diversity is high. The relationship is actually more complex as it follows a U-shaped pattern. At low levels of ethnic diversity, subsidiary ownership levels are relatively high. This suggests that when making their investment decisions, managers may view relatively ethnically homogenous countries as more stable and less risky. As diversity begins to rise, managers, particularly of MNE subsidiaries, may not be adequately prepared nor understand how to manage potential ethnic tensions in the country. It is also possible that at this stage ethnic tensions in society begin to spillover into the workplace. Finally, as diversity increases to higher levels, managers may no longer perceive diversity to be as a salient since it is widely accepted and taken for granted. By understanding the complexity in ethnic diversity and tensions in society, managers may be able to effectively manage diversity in the workplace and play a role in reducing potential sources of conflict; in the company and in the wider society. In this way, businesses play a clear and direct role in increasing socioeconomic stability and peace (Oetzel & Miklian, 2017).

Ethnic diversity and employee productivity followed a similar relationship. At low levels of ethnic diversity, employee productivity is high. We found that this relationship is equally important for domestic and foreign subsidiaries. These findings are consistent with the idea that relatively homogenous populations may be easier to manage (Easterly & Levine, 1997). As ethnic diversity begins to increase, however, challenges with managing diversity begin to arise. Firms with small or moderate levels of diversity need to establish effective practices for managing diversity to turn it into an advantage. At this stage, however, these practices may not be institutionalized and the diversity poses challenges to the organization. Once diversity becomes more pervasive in society and in the organization, employees’ exposure to ethnic differences becomes normalized. Diversity is no longer as salient, at least as a point of conflict, because it is taken for granted. It may be that at that point, organizations are best positioned to leverage the positive benefits of a diverse workforce. This is consistent with a significant portion of the organizational behavior research that has shown that when diversity is well managed it is a valuable asset that can enhance organizational outcomes.

Building on that argument, our study also contributes to the theoretical and empirical research on ethnic diversity and its impact on organizations. A large body of research has substantially increased our knowledge of intra-organizational diversity and its impact on individuals, teams, and the business as a whole (Cox et al., 1991; Giambatista & Bhappu, 2010; Richard, 2000; Stahl et al., 2010). Relatively few studies, however, have focused on how ethnic diversity in society, and the wider country context in which the subsidiary is embedded, affect managers’ perceived risk of doing business and employee productivity (see Chua (2013) and Delis et al. (2016) for notable exceptions). We also bring new theoretical and empirical insights to the study of diversity and its impact on organizations, insights that may inform future research on the effect of diversity on subsidiary strategy and employee productivity.

Our findings appear to be consistent with the theories of Collier and Hoeffler suggesting that the relationship between country-level ethnic diversity, subsidiary risk, and employee productivity is more nuanced. Judging from their investment decisions, although managers may view ethnically homogenous populations to be relatively low risk, they do not appear to consider more ethnically diverse environments as riskier. In fact, based on their ownership structure, it appears that managers see moderate levels of diversity as the greater risk and high levels of diversity as relatively low risk. Thus, it appears that it is not ethnic diversity per se that poses a risk. Rather, prior research would suggest that it is the degree to which the various ethnic groups politically (and by extension economically) compete in society. Strong institutions that set and fairly enforce the ‘rules of the game’ are needed to reduce inter-ethnic tensions and foster a fair distribution of resources and opportunities.

We suggest that in studies where the relative power of different groups is theoretically important, using an entropy measure to capture diversity may be beneficial since the measure not only considers the relative population size across ethnic groups, but also differences in power across these groups by identifying politically included-, politically excluded-, and ethnic groups without political influence. The entropy measure assumes that changes in relative size within an included- (or excluded-) group category are more relevant than across categories, and diversity across categories is more important in deciding socio-political risk.

Another example of the value of this approach is that we are also able to capture the degree of diversity within the included and excluded groups, and ethnic groups without political influence. This appears to be important since, in our ad-hoc analysis, ethnic diversity within the included-group category has a U-shaped relationship with subsidiary ownership for both domestic and foreign subsidiaries, while ethnic diversity within the excluded-group category overall does not have strong, consistent effects across samples. The implication is that the degree of ethnic diversity within politically included ethnic groups, that are able to influence politics, matters. Thus infighting and secession among politically vibrant groups increased perceived political risks in the country. For those excluded-groups, they are usually economically disenfranchised groups and may increase ethnic enclaves in a country. Foreign firms may consider such ethnic enclaves as a fragmentation of society, but domestic firms may see them differently finding valuable skills and opportunities from ethnic firms.

Regarding employee productivity, however, the included-group diversity has a U-shaped relationship with productivity only for the foreign subsidiary sample. This finding implies that managers of foreign firms may not understand ethnic relations in the host country. Due to this lack of understanding, they may be ill-equipped to manage ethnic tensions in the workplace. Domestic firms likely have developed certain types of political strategies that lower the realization of ethnic risks on their businesses. For instance, one study shows that foreign firms have a competitive disadvantage in utilizing political ties compared to their domestic counter parts, and better political ties lead higher firm performance when uncertainty increases (Li, Poppo, & Zhou, 2008).

Our ad-hoc analysis also shows that ethnic diversity across ethnic group categories has a U-shaped relationship with employee productivity for both domestic and foreign subsidiaries, but not with subsidiary ownership. This implies that discrimination, injustice, and/or political exclusion may result in greater social tensions which in turn affect the overall socioeconomic climate for business and society. Consistent with the theories of Collier and Hoeffler (Collier, 2001; Collier & Hoeffler, 1998, 2002), we interpret these findings to suggest that business risks are lower when ethnic diversity is either low or high. When diversity is moderate, the potential risks are higher. Furthermore, the findings on employee productivity lend weight to the notion that the differences are not solely perceptual. In practice, employees are less productive when there is a moderate level of ethnic diversity. One reason why we did not find a similar effect for subsidiary ownership is managers, particularly of MNCs, may not be as aware of the nuances of diversity in a host country when making ownership decisions. Thus the relationship may have a more direct effect for ownership structure.

The broader policy implications from these ad-hoc results is that nations should have engagement policies toward excluded ethnic groups that will increase the diversity within the included-group category. Such engagement policies will lower political risks and conflicts such as rebellions, infightings, and secessions, increase domestic and foreign firms’s ownership positions toward local subsidiaries, improve employee productivity, and, in turn, increase the economic growth of nations. Thus our study provides important policy implications. We argue that ethnic diversity is highly desirable but needs to be well managed. It is counterproductive to ignore potential ethnic tensions in a country and their potential to spill over to the workplace (Brief et al., 2005). Although some political groups have argued that given current events – such as conflicts and security issues around the world – multiculturalism or engagement policy has failed, we suggest that the policy of multiculturalism is not the issue. Ethnic diversity does not increase socioeconomic risk nor pose a threat to business. Instead, it is poor political and economic policies that disenfranchised minority groups that may generate ethnic tensions.

Policymakers have the ability to ensure that every ethnic group in a country is enfranchised. All ethnic groups must have equal access to the rights and privileges available in a society. Despite the association between lower ethnic diversity and lower ethnic tension, there are potential downsides to high levels of ethnic homogeneity. Each ethnic group has its own valuable cultures, talents, skills and thoughts; supporting and enfranchising these groups has the potential to increase the overall knowledge in society. Furthermore, business research shows that a diverse workforce is often associated with higher levels of innovation and creativity (Cox, Lobel, & McLeod, 1991; Elron, 1997; Giambatista & Bhappu, 2010; Richard, 2000; Stahl et al., 2010). In addition, there are only a few nations on earth that consist of one ethnic group. Since globalization increases the movement of people across borders and interactions across nations and ethnic groups, we consider that preparing for and achieving high levels of ethnic diversity is a better policy approach. In addition, each ethnic group has its own valuable culture, talents, skills, and thoughts. Securing and developing those talents and assets will only increase the amount of knowledge and capabilities in society.

In addition, policies tend to ignore the long-term economic benefits of ethnic diversity and focus on the tensions and conflicts between different ethnic groups. Our results suggest that firms, micro-actors in economic systems, gain benefits from ethnic diversity. While some tensions and conflicts are inevitable in the process of economics and societal growth (Hirschman, 1958), nations may need to recognize and utilize the economic benefits of diversity like firms do.

Policy makers might also need to develop tools to minimize potential conflicts or political or economic disenfranchisement that might occur in moderately diverse societies. One tool could be that policy makers emphasize national identity ahead of ethnic identity as that has been shown to mitigate ethnic conflict (Michaloppoulos & Papaioannou, 2015). One study at the organization-level of analysis shows that when ethnic groups feel insecure they tend to protect themselves against different ethnic groups. An inclusive organizational culture, however, enables all ethnic groups to come together and to identify as one group (Parboteeah, Seriki & Hoegl, 2014). In addition, with a better understanding of the causes of conflict at different levels of ethnic diversity, policy makers may be better prepared to address potential problems at their source (e.g., through increased support of public goods and improved access) before they develop into widespread social problems.

Limitations and Future Research

Like all empirical research, there are limitations with our study that should be noted. First, as we illustrate with the example of Protina, some firms are more proactive about, or have institutionalized management systems for, resolving ethnic tension and conflict than others. Thus it may be valuable to look at which firm characteristics determine behaviors and management strategies around this issue. Managers may not be aware of the positive influence that business has on ethnic diversity and ethnic tensions in a country. For example, it is expected that urbanization, immigration, and an inflow of MNCs will likely increase the ethnic diversity in less developed countries where their governments or political elites either do not recognize the importance of ethnic diversity or keep channeling political interests to their own ethnic groups. Thus simply through their day-to-day operations, businesses may influence ethnic diversity and social stability in a country.

For research in international strategy, some scholars have found that prior experience managing policy uncertainty can be leveraged into other markets (Delios & Henisz, 2003; Holburn & Zelner, 2010). For other types of risk, however, country-specific knowledge may be needed (Li & Meyer, 2009; Oh & Oetzel, 2017). Based on prior work, we would expect that country-specific knowledge may be needed to understand the dynamics of ethnic relations in a particular country, how the country government manages (or not) ethnic diversity, and to what extent people have equal access to economic and political resources.

Another area for future research concerns how managers perceive risk across country and cultural contexts and how home and host country distance affects those perceptions. Earlier studies have measured and incorporated ethnicity-based distance as an important dimension of cultural distance (Dow et al., 2016; Luiz, 2015). It is known that familiarity affects how people rate actual risks outside their home countries or cultural contexts (Schneier, 2003). As the cultural, religious, linguistic and geographical distance between a subsidiary’s home and host country increases, the greater the perceived risk is likely to be between the two (Brouthers, 1995; Dow et al., 2016). Our paper provides some evidence about the difference between domestic and foreign subsidiaries in regard to the effect of included group diversity on subsidiary productivity. An intriguing avenue for future research, however, may be investigating how do cross-country perceptions affect managers’ evaluations and, in turn, how closely aligned are these perceptions with actual risk or country conditions?

In addition, country context and the political dynamics among ethnic groups may matter more than the level of diversity. High levels of diversity might actually make differences less relevant. Future research may benefit from a more complex understanding of country-level ethnic diversity. Research shows that the relative importance of ethnic identity is instrumental and constructed (Laitin, 1998; Scott, 1990). Therefore, future research needs to look at which contextual factors actually trigger ethnicity issues in society.

Second, in this large-scale cross-country study we are not able to gain insight into the internal decision-making process within organizations around subsidiary ownership in response to ethnic diversity and ethnic composition in society. Similarly, we do not know the specific dynamics that occur in ethnic conflicts that might lead to spillover effects across society or to organizations. What is it exactly about this ethnic mix that creates lower levels of employee productivity? Future qualitative research, particularly in-depth cases studies within subsidiaries, may shed light on this question.

Finally, our sample firms are publicly listed firms. It is important to note that the propensity to be listed on stock exchanges differ across countries. Since our analysis includes industry-, region- and year-fixed effects, the propensity issue might not be a serious concern for our empirical analysis. However, it would be interesting to look at private firms and their ownership strategies and employee productivity in regard to ethnic diversity. Another empirical limitation is that although we collected ethnic information from the most updated and widely used sources in academic research, the identification of ethnic groups might be subjective. Detailed ethnic information is available for very few countries and therefore such information is not suitable for a cross-country study like ours. Perhaps future research can extend this study by looking at ethnic diversity at the sub-national level.

Conclusions

Findings from this cross-country study support research suggesting that it is not ethnic diversity per se that poses a risk for subsidiaries or negatively impacts employee productivity. Rather, it appears to be a function of two factors: 1) whether the level of diversity is low, moderate, or high and, 2) the degree of political and economic inclusion of the various ethnic groups. Our findings also show that while foreign and domestic subsidiaries similarly perceive ethnic risks when making ownership decisions, domestic subsidiaries are better in managing ethnic risks and are able to reach higher levels of labor productivity than foreign subsidiaries. These results suggest that it is not just internal firm management that matters for realizing the benefits of diversity but also the wider management of diversity in society. Policy makers play a critical role in ensuring that minority groups are fully integrated into the social and economic life of a country. Businesses working together can be powerful advocates for policy changes.

Notes

  1. 1

    The definition adopted here does not include tribes and clans that conceive of ancestry in genealogical terms, nor regions that do not define commonality on the basis of shared ancestry (Wimmer et al., 2009).

     
  2. 2

    This example comes from a case study produced by the United Nations Global Compact in 2009.

     

Notes

Acknowledgements

An earlier version of this paper received the OB/HRM/OT Best Paper Award from the International Management Division of the Academy of Management, 2015 and a Best Paper Award from European International Business Academy, 2016. We would like to acknowledge the helpful comments we received from Ruth Aguilera, Mary Yoko Brannen, Roy Chua, Rajiv Kozhikode, Seung-Hyun Lee, and Martha Maznevski in an earlier version of this paper. Financial support for this research project was partially provided by the Social Sciences and Humanities Research Council of Canada [435-2017-0897].

Supplementary material

42214_2018_16_MOESM1_ESM.docx (53 kb)
Supplementary material 1 (DOCX 278 kb)

References

  1. Agarwal, S., & Ramaswani, S. N. 1992. Choice of foreign market entry mode: Impact of ownership, location and internationalization factors. Journal of International Business Studies, 23(1): 1–27.Google Scholar
  2. Alesina, A., Baqir, R., & Easterly, W. 1999. Public goods and ethnic divisions. Quarterly Journal of Economics, 114(4): 1243–1284.Google Scholar
  3. Alesina, A., & La Ferrara, E. 2005. Ethnic diversity and economic performance. Journal of Economic Literature, 43(3): 762–800.Google Scholar
  4. Anderson, E., & Gatignon, H. 1986. Modes of foreign entry: A transaction cost analysis and propositions. Journal of International Business Studies, 17(3): 1–26.Google Scholar
  5. Baker, W. 1990. Market networks and corporate behavior. American Journal of Sociology, 96(3): 589–625.Google Scholar
  6. Baltagi, B. H. 2002. Econometrics. New York, NY: Springer.Google Scholar
  7. Bannon, A., Miguel, E., & Posner, D. N. 2004. Sources of ethnic identification in Africa. Afrobarometer Working Paper No. 44. Afrobarometer.Google Scholar
  8. Berry, H., Guillen, M., & Zhou, N. 2010. An institutional approach to cross-national distance. Journal of International Business Studies, 41(9): 1460–1480.Google Scholar
  9. Borjas, G. J. 1992. Ethnic capital and intergenerational mobility. The Quarterly Journal of Economics, 107(1): 123–150.Google Scholar
  10. Boytsun, A., Deloof, M., & Matthyssens, P. 2011. Social norms, social cohesion, and corporate governance. Corporate Governance: An International Review, 19(1): 41–60.Google Scholar
  11. Brief, A. P., Umphress, E. E., Dietz, J., Burrows, J. W., Butz, R. M., & Scholten, L. 2005. Community matters: Realistic group conflict theory and the impact of diversity. Academy of Management Journal, 48(5): 830–844.Google Scholar
  12. Brouthers, K. D. 1995. The influence of international risk on entry mode strategy in the computer software industry. Management International Review, 35(1): 7–28.Google Scholar
  13. Cederman, L.-E., Wimmer, A., & Min, B. 2010. Why do ethnic groups rebel? New data and analysis. World Politics, 62(1): 87–119.Google Scholar
  14. Central Intelligence Agency (CIA). 2014. The World Factbook. https://www.cia.gov/library/publications/the-world-factbook/. Accessed on 20 March 2018.
  15. Chan, C. M., & Makino, S. 2007. Legitimacy and multi-level institutional environments: Implications for foreign subsidiary ownership structure. Journal of International Business Studies, 38(4): 621–638.Google Scholar
  16. Chang, S. J. 2003. Ownership structure, expropriation, and performance of group-affiliated companies in Korea. Academy of Management Journal, 46(2): 238–253.Google Scholar
  17. Cho, M. H. 1998. Ownership structure, investment, and the corporate value: An empirical analysis. Journal of Financial Economics, 47(1): 103–121.Google Scholar
  18. Choi, J. N. 2007. Group composition and employee creative behavior in a Korean electronics company: Distinct effects of relational demography and group diversity. Journal of Occupational and Organizational Psychology, 80(2): 213–234.Google Scholar
  19. Chua, R. Y. J. 2013. The costs of ambient cultural disharmony: Indirect intercultural conflicts in social environment undermine creativity. Academy of Management Journal, 56(6): 1545–1577.Google Scholar
  20. Collier, P. 2000. Ethnicity, politics and economic performance. Economy and Politics, 12(3): 225–245.Google Scholar
  21. Collier, P. 2001. Implications of ethnic diversity. Economic Policy, 16(32): 127–166.Google Scholar
  22. Collier, P., & Hoeffler, A. 1998. On the economic causes of civil war. Oxford Economic Papers, 50(4): 563–573.Google Scholar
  23. Collier, P., & Hoeffler, A. 2002. On the incidence of civil war in Africa. Journal of Conflict Resolution, 46(1): 13–28.Google Scholar
  24. Cox, T. H., Jr., Lobel, S. A., & McLeod, P. L. 1991. Effects of ethnic group cultural differences on cooperative and competitive behavior on a group task. Academy of Management Journal, 34(4): 827–847.Google Scholar
  25. Datta, D. K., Guthrie, J. P., & Wright, P. M. 2005. Human resource management and labor productivity: Does industry matter? Academy of Management Journal, 48(1): 135–145.Google Scholar
  26. Delios, A. P., & Beamish, P. 1999. Ownership strategy of Japanese firms: Transactional, institutional, and experience influences. Strategic Management Journal, 20(10): 915–933.Google Scholar
  27. Delios, A. P., & Henisz, W. 2000. Japanese firms’ investment strategies in emerging economies. Academy of Manangement Journal, 43(3): 305–323.Google Scholar
  28. Delios, A. P., & Henisz, W. 2003. Policy uncertainty and the sequence of entry by Japanese firms, 1980–1998. Journal of International Business Studies, 24(11): 1153–1164.Google Scholar
  29. Delis, M. D., Gaganis, C., Hasan, I., & Pasiouras, F. 2016. The effect of board of directors from countries with different genetic diversity levels on corporate performance. Management Science, 63(1): 231–249.Google Scholar
  30. Demsetz, H., & Lehn, K. 1985. The structure of corporate ownership: Causes and consequences. Journal of Political Economy, 93(6): 1155–1177.Google Scholar
  31. Demsetz, H., & Villalonga, B. 2001. Ownership structure and corporate performance. Journal of Corporate Finance, 7(3): 209–233.Google Scholar
  32. Dow, D., Cuypers, I. R. P., & Ertug, G. 2016. The effects of within-country linguistic and religious diversity on foreign acquisitions. Journal of International Business Studies, 47(3): 319–346.Google Scholar
  33. Duanmu, J., & Guney, Y. 2013. Heterogeneous effect of ethnic networks on international trade of Thailand: The role of family ties and ethnic diversity. International Business Review, 22(1): 126–139.Google Scholar
  34. Easterly, W., & Levine, R. 1997. Africa’s growth tragedy: Policies and ethnic divisions. Quarterly Journal of Economics, 111(4): 1203–1250.Google Scholar
  35. Eifert, B., Miguel, E., & Posner, D. N. 2010. Political competition and ethnic identification in Africa. American Journal of Political Science, 54(2): 494–510.Google Scholar
  36. Elron, E. 1997. Top management teams within multinational corporations: Effects of cultural heterogeneity. The Leadership Quarterly, 8(4): 393–412.Google Scholar
  37. Esteban, J., Mayoral, L., & Ray, D. 2012. Ethnicity and conflict: An empirical study. American Economic Review, 102(4): 1310–1342.Google Scholar
  38. European Commission. 2018. Free movement – EU nationals. http://ec.europa.eu/social/main.jsp?catId=457&langId=en.
  39. Feinberg, S. E., & Gupta, A. K. 2009. MNC subsidiaries and country risk: Internalization as a safeguard against weak external institutions. Academy of Management Journal, 52(2): 381–399.Google Scholar
  40. Florida, R. 2002. The economic geography of talent. Annals of the Association of American Geographers, 92(4): 743–755.Google Scholar
  41. García-Canal, E., & Guillén, M. F. 2008. Risk and the strategy of foreign location choice in regulated industries. Strategic Management Journal, 29(10): 1097–1115.Google Scholar
  42. Gatignon, H., & Anderson, E. 1988. The multinational corporation’s degree of control over foreign subsidiaries: An empirical test of transaction cost explanation. Journal of Law Economics and Organization, 4(1): 305–336.Google Scholar
  43. Gaur, A. S., & Lu, J. W. 2007. Ownership strategies and survival of foreign subsidiaries: Impacts of institutional distance and experience. Journal of Management, 33(1): 84–110.Google Scholar
  44. Giambatista, R. C., & Bhappu, A. D. 2010. Diversity’s harvest: Interactions of diversity sources and communication technology on creative group performance. Organizational Behavior and Human Decision Processes, 111(2): 116–126.Google Scholar
  45. Gong, Y. 2003. Subsidiary staffing in multinational enterprises: Agency, resources, and performance. Academy of Management Journal, 46(4): 728–739.Google Scholar
  46. Gong, Y. 2006. The impact of subsidiary top management team national diversity on subsidiary performance: Knowledge and legitimacy perspective. Management International Review, 46(6): 771–789.Google Scholar
  47. Greene, W. H. 2003. Econometric analysis (5th ed.). Upper Saddle River, NJ: Prentice Hall.Google Scholar
  48. Grief, A. 1993. Contract enforceability and economic institutions in early trade: The Maghribi traders’ coalition. American Economic Review, 83(3): 525–548.Google Scholar
  49. Gurr, T. R., & Moore, W. H. 1997. Ethnopolitical rebellion: A cross-sectional analysis of the 1980s with Risk Assessments for the 1990s. American Journal of Political Science, 41(4): 1079–1103.Google Scholar
  50. Hacker, A. 1995. Two nations: Black and white, separate, hostile and unequal. New York, NY: Scribner’s.Google Scholar
  51. Hale, H. E. 2004. Explaining ethnicity. Comparative Political Studies, 37(4): 458–485.Google Scholar
  52. Hambrick, D. C. 1994. Top management groups: A conceptual integration and reconsideration of the “team” label. In B. M. Shaw & L. L. Cummings (Eds.), Research in organizational behavior, 16: 171–214. Greenwich, CT: JAI Press.Google Scholar
  53. Hegde, D., & Tumlinson, J. 2014. Does social proximity enhance business partnerships? Theory and evidence from ethnicity’s role in U.S. venture capital. Management Science, 60(9): 2355–2380.Google Scholar
  54. Hirschman, A. O. 1958. The strategy of economic development. New Haven, CT: Yale University Press.Google Scholar
  55. Hjort, J. 2014. Ethnic divisions and production in firms. Quarterly Journal of Economics, 129(4): 1899–1946.Google Scholar
  56. Hofstede, G. 1980. Culture’s consequences: International differences in work-related values. Beverly Hills, CA: Sage.Google Scholar
  57. Holburn, G., & Zelner, B. A. 2010. Political capabilities, policy risk, and international investment strategy: Evidence from the global electric power industry. Strategic Management Journal, 31(12): 1290–1315.Google Scholar
  58. Huckfeldt, R., & Kohfeld, C. W. 1989. Race and decline of class in American Politics. Urbana, IL: University of Illinois Press.Google Scholar
  59. Huselid, M. A. 1995. The impact of human resource management practices on turnover, productivity, and corporate financial performance. Academy of Management Journal, 38(3): 635–672.Google Scholar
  60. Hyun, H.-J., Oh, C. H., & Paik, Y. 2015. Impact of nationality composition in foreign subsidiary on its performance: A case of Korean companies. International Journal of Human Resource Management, 26(6): 806–830.Google Scholar
  61. Ito, K., & Rose, E. L. 1994. The genealogical structure of Japanese firms: Parent-subsidiary relationships. Strategic Management Journal, 15(S2): 35–51.Google Scholar
  62. Jehn, K. A., Northcraft, G. B., & Neale, M. A. 1999. Why differences make a difference: A field study of diversity, conflict, and performance in workgroups. Administrative Science Quarterly, 44(4): 741–763.Google Scholar
  63. Johanson, J., & Vahlne, J. E. 1977. The internationalization process of the firm: A model of knowledge development and increasing foreign market commitments. Journal of International Business Studies, 8(1): 23–32.Google Scholar
  64. Judge, G. G., Hill, R. C., Lutkepohl, H., & Lee, T. 1985. The theory and practice of econometrics (2nd ed.). New York, NY: Wiley.Google Scholar
  65. Kennedy, C. R. 1984. The external environment-strategic planning interface: U.S. multinational corporate practices in the 1980s. Journal of International Business Studies, 15(2): 99–108.Google Scholar
  66. Kim, W. C., & Hwang, P. 1992. Global strategy and multinationals’ entry mode choice. Journal of International Business Studies, 23(1): 29–54.Google Scholar
  67. Kumar, M. V. S. 2009. The relationship between product and international diversification: The effects of short-run constraints and endogeneity. Strategic Management Journal, 30(1): 99–116.Google Scholar
  68. La Ferrara, E. 2003. Kin groups and reciprocity: A model of credit transactions in Ghana. American Economic Review, 93(5): 1730–1751.Google Scholar
  69. Laitin, D. D. 1998. Identity in formation: The Russian-speaking populations in the near abroad. Ithaca, NY: Cornell University Press.Google Scholar
  70. Lassen, D. D. 2007. Ethnic divisions, trust, and the size of the informal sector. Journal of Economic Behavior & Organization, 63(3): 423–438.Google Scholar
  71. Lee, H.-J., & Reade, C. 2015. Ethnic homophily perceptions as an emergent IHRM challenge: Evidence from firms operating in Sri Lanka during the ethnic conflict. International Journal of Human Resource Management, 26(13): 1645–1664.Google Scholar
  72. Li, P. Y., & Meyer, K. E. 2009. Conceptualizing experience effects in international business: A study of ownership strategies. Journal of World Business, 44(4): 370–382.Google Scholar
  73. Li, J. J., Poppo, L., & Zhou, K. Z. 2008. Do managerial ties in China always produce value? Competition, uncertainty, and domestic vs. foreign firms. Strategic Management Journal, 29(4): 383–400.Google Scholar
  74. Luiz, L. M. 2015. The impact of ethno-linguistic fractionalization on cultural measure: Dynamics, endogeneity and modernization. Journal of International Business Studies, 46(9): 1080–1098.Google Scholar
  75. Michaloppoulos, S., & Papaioannou, E. 2015. On the ethnic origins of African development: Traditional chiefs and pre-colonial political centralization. Academy of Management Perspectives, 29(1): 32–71.Google Scholar
  76. Oetzel, J., & Miklian, J. 2017. Multinational enterprises, risk management, and the business and economics of peace. Multinational Business Review, 25(4): 270–286.Google Scholar
  77. Oh, C. H., & Oetzel, J. 2017. Once bitten twice shy? Experience managing violent conflict risk and MNC subsidiary-level investment and expansion. Strategic Management Journal, 38(3): 714–731.Google Scholar
  78. Østergaard, C. R., Timmermans, B., & Kristinsson, K. 2011. Does a different view create something new? The effect of employee diversity on innovation. Research Policy, 40(3): 500–509.Google Scholar
  79. Ottaviano, G. I. P., & Peri, G. 2006. The economic value of cultural diversity: Evidence from US cities. Journal of Economic Geography, 6(1): 9–44.Google Scholar
  80. Ottaviano, G. I. P., & Prarolo, G. 2009. Cultural identity and knowledge creation in cosmopolitan cities. Journal of Regional Studies, 49(4): 647–662.Google Scholar
  81. Pan, Y. 2017. Strategic motives, institutional environments, and firm’s FDI ownership. Multinational Business Review, 25(4): 307–327.Google Scholar
  82. Parboteeah, K. P., Seriki, H. T., & Hoegl, M. 2014. Ethnic diversity, corruption, and ethical climates in sub-Saharan Africa: Recognizing the significance of human resource management. International Journal of Human Resource Management, 25(7): 979–1001.Google Scholar
  83. Pelled, L. H., Eisenhardt, K. M., & Xin, K. R. 1999. Exploring the black box: An analysis of work group diversity, conflict, and performance. Administrative Science Quarterly, 44(1): 1–28.Google Scholar
  84. Purdey, J. 2006. Anti-Chinese violence in Indonesia 1996–1999. Honolulu, HI: University of Hawaii Press.Google Scholar
  85. Ralston, D. A., Gustafson, D. J., Cheung, F. M., & Terpstra, R. H. 1993. Differences in managerial values: A study of US, Hong Kong, and PRC managers. Journal of International Business Studies, 24(1): 177–207.Google Scholar
  86. Reade, C., & Lee, H.-J. 2012. Organizational commitment in time of war: Assessing the impact of attenuation of employee sensitivity to ethnopolitical conflict. Journal of International Management, 18(1): 85–101.Google Scholar
  87. Richard, O. C. 2000. Racial diversity, business strategy, and firm performance: A resource-based view. Academy of Management Journal, 43(2): 164–177.Google Scholar
  88. Riggs, F. W. 2002. Globalization, ethnic diversity, and nationalism: The challenge for democracies. The Annals of the American Academy of Political and Social Science, 581(1): 35–47.Google Scholar
  89. Roessler, G. 2011. The enemy from within. Personal rule, coups, and civil wars in Africa. World Politics, 63(2): 300–346.Google Scholar
  90. Sanders, J. M., & Nee, V. 1987. Limits of ethnic solidarity in the enclave economy. American Sociological Review, 52(6): 745–773.Google Scholar
  91. Schneier, B. 2003. Beyond fear: Thinking sensibly about security in an uncertain world. New Yok, NY: Copernicus Books.Google Scholar
  92. Scott, G. M., Jr. 1990. A resynthesis of the primordial and circumstantial approaches to ethnic group solidarity. Ethnic and Racial Studies, 13(2): 147–171.Google Scholar
  93. Shekhar, V. 2014. The good news from Indonesia’s election stalemate. http://transitions.foreignpolicy.com/posts/2014/07/16/the_good_news_from_indonesia_s_election_stalemate. Accessed on 20 March 2018.
  94. Shenkar, O. 2012. Cultural distance revisited: Towards a more rigorous conceptualization and measurement of cultural differences. Journal of International Business Studies, 43(1): 1–11.Google Scholar
  95. Stahl, G. K., Maznevski, M. L., Voight, A., & Jonsen, K. 2010. Unraveling the effects of cultural diversity in teams: A meta-analysis of research on multicultural work groups. Journal of International Business Studies, 41(4): 690–709.Google Scholar
  96. Southern Poverty Law Center. 2016. Ten days after: Harassment and intimidation in the aftermath of the election. November 29. www.splcenter.org/20161129/ten-days-after-harassment-and-intimidation-aftermat-election. Accessed 22 March 2018.
  97. Tan, D., & Mahoney, J. F. 2006. Why a multinational firm chooses expartiates: Integrating resource-based, agency and transaction costs perspectives. Journal of Management Studies, 43(3): 457–484.Google Scholar
  98. Teachman, J. D. 1980. Analysis of population diversity: Measures of qualitative variation. Sociological Methods and Research, 8(3): 341–361.Google Scholar
  99. Tiemann, T., Das, J., & DiRienzo, C. 2006. A note on an ethnic homogeneity Kuznets Curve. Challenge, 49(2): 112–120.Google Scholar
  100. Tsui, A. S., Egan, T. D., & O’Reilly, C. A. 1992. Being different: Relational demography and organizational attachment. Administrative Science Quarterly, 37(4): 549–579.Google Scholar
  101. Tung, R. L. 2008. The cross-cultural research imperative: The need to balance cross-national and intra-national diversity. Journal of International Business Studies, 39(1): 41–46.Google Scholar
  102. UNCTAD. 2014. World Investment Report 2014. http://unctad.org. Accessed 15 March 2018.
  103. United Nations Global Compact (UNGC). 2010. Doing business will advancing peace and development. https://www.unglobalcompact.org/docs/issues_doc/Peace_and_Business/DBWAPD_2010.pdf. Accessed 23 March 2018.
  104. Whitley, R. 1992. Business systems in East Asia: Firms, markets, and societies. London: Sage Publications.Google Scholar
  105. Whitley, R. 1997. Governance at work: The social regulation of economic relations. Oxford: Oxford University Press.Google Scholar
  106. Wimmer, A., Cederman, L. E., & Min, B. 2009. Ethnic politics and armed conflict: A configurational analysis of a new global data set. American Sociological Review, 74(2): 316–337.Google Scholar
  107. Woodcock, C. P., Beamish, P. W., & Makino, S. 1994. Ownership-based entry mode strategies and international performance. Journal of International Business Studies, 25(2): 253–273.Google Scholar

Copyright information

© Academy of International Business 2019

Authors and Affiliations

  1. 1.Kogod School of BusinessAmerican UniversityWashingtonUSA
  2. 2.Beedie School of BusinessSimon Fraser UniversityVancouverCanada

Personalised recommendations