"Internationalization is to a large extent a matter of establishing relationships in foreign markets" (Johanson and Vahlne 1990, p. 21). 1 Introduction
Ever since Sullivan's (1994) seminal paper on measuring the firm's degree of internationalization (DOI, also abbreviated as DI), little effort has been expended in the international diversification literature on the development of new and more precise proxies of this multifaceted construct (Hitt et al. 2006). Shortly after the publication of Sullivan's work, a prominent discussion arose on the theoretical and methodological implications of employing Sullivan's multi-item aggregate index over single-item metrics (Ramaswamy et al. 1996; Sullivan 1996). Despite the enduring relevance of this debate, there are still plenty of unresolved issues that were identified in this discussion. In particular, researchers have recognized a lack of progress on two main aspects: (1) the selection and development of DOI metrics that are theoretically sound (Asmussen et al. 2007; Glaum and Oesterle 2007; Hennart 2007, 2011; Hitt et al. 2006; Nguyen 2016; Verbeke and Forootan 2012), and (2) the theoretical and methodological merits of multi-dimensional versus uni-dimensional DOI constructs (Glaum and Oesterle 2007; Hassel et al. 2003; Hitt et al. 2006; Nguyen 2016; Wiersema and Bowen 2011).
In this paper, we address these two issues simultaneously. First, we replicate Sullivan's original research efforts while introducing two social-based DOI metrics based on the transnational board interlock connections of the firm. We argue that the inclusion of social markers of internationalization can strengthen the DOI construct by capturing internationalization activity that is relevant from a theoretical and practical perspective, but that is often overlooked by traditional measures. Second, we use exploratory and confirmatory factor analysis to reveal and test the multi-dimensional structure of the DOI construct and uncover the existence of a stand-alone social dimension of internationalization. Our results show support for a multi-dimensional DOI construct composed of three distinct dimensions: a real (1) one, composed of the firm's foreign sales and assets; an exposure one, represented by the number and cultural zone dispersion of the firm's foreign subsidiaries; and a social one, composed of the top managers international experience and the number and cultural zone variety of the firm's transnational board linkages. These results call for a fundamental reexamination of the way researchers group and conceive of DOI metrics, and underscore the importance of conducting factor reliability tests before employing these metrics in empirical research (see also Oesterle and Richta 2013).
The rationale for measuring a firm's foreign footprint resides in the DOI constructs potential to explain important causes and consequences of international diversification. However, diverse, inconsistent and undertheorized operationalizations of this construct have resulted in mixed results in the field (Glaum and Oesterle 2007; Hennart 2007, 2011; Hitt et al. 2006; Oesterle and Richta 2013; Thomas and Eden 2004; Verbeke and Brugman 2009; Verbeke and Forootan 2012). At the root of this discussion is the question of whether or not different operationalizations of the DOI construct indeed capture the firm's degree of internationalization. Organizational researchers have defined construct validity as "the extent to which an operationalization measures the concept it is supposed to measure" (Bagozzi et al. 1991, p. 421). For a DOI construct to be valid, it should reflect a wide variety of international exchange activities that occur across the firm's entire value chain (Asmussen et al. 2007; Verbeke and Brugman 2009). According to Wiersema and Bowen (2011, p. 155), a robust DOI construct should include "all foreign aspects of a firm's value chain, from the geographic markets where it sells its products/services to the global locations where it produces its products/services and the geographic locations where its capabilities reside". These requirements set a very high bar for any DOI metric to surpass in isolation. For this reason, Sullivan (1994) suggested that the DOI of the firm be measured with a variety of underlying markers of firm internationalization. In particular, he recommended five internationalization metrics to be combined into an aggregate index. Sullivan's approach, however, has been criticized as statistically inappropriate and theoretically unsound (Hassel et al. 2003; Hennart 2007, 2011; Ramaswamy et al. 1996). Despite these short-comings, new research has failed to introduce new metrics and methods for measuring a firm's DOI, particularly social-based markers of internationalization that have the potential to strengthen this construct by providing an untainted reflection of the firm's foreign interdependencies.
The impasse in the development of social-based metrics of internationalization is partly explained by the asocial nature of mainstream theories of internationalization. The overarching argument behind some of these major theories, such as transaction cost/internalization theory (Buckley and Casson 1976; Hennart 1977, 1982; Rug-man 1981), is that firms are in possession of firm specific advantages (FSAs), usually in the form of intangible assets, that are exploitable abroad. However, because of market failures or high transaction costs in foreign countries, firms might prefer to exploit their FSAs internally, rather than externally, though either horizontal or vertical integration. By internalizing foreign operations, these theories contend, firms can take advantage of their intangibles without fear of misappropriation by foreign firms. Traditional models of internationalization are thus atomistic in nature. It is not surprising, then, that existing DOI metrics are atomistic as well. Firms, however, do engage in a variety of formal and informal relationships in their host countries. Firms need to develop these relationships to secure the knowledge and resources that are required to successfully compete in foreign markets. By engaging in exchange activities with foreign firms, firms effectively externalize some of its activities to foreign partners without risking control of critical FSAs (Johanson and Mattsson 1987, 1988). Firms that fail to embed themselves in foreign networks experience a liability of outsidership (Johanson and Vahlne 2009) that, together with the liability of foreignness (Zaheer 1995), can jeopardize the success of their international operations. Firm and network internationalization are thus closely intertwined, and delving deeper into the degree of embeddedness of the firm in foreign business networks can offer us a direct window into the extent and relevance of the firm's foreign activities.
Our argument develops in stages. First, we review the literature on measuring the DOI of the firm. Second, we offer theoretical and practical support for the consideration of social-based measures of DOI, as well as for our selection of transnational board interlocks as underlying metrics. Third, we perform an exploratory and confirmatory factor analysis to test and reveal the multi-dimensional structure of the DOI construct. Finally, we provide some thoughts regarding the implications of our results for international diversification research and offer some ideas for future research on measuring the DOI of the firm.
2 Measuring the Degree of Internationalization of the Firm
In their review of the field, Hitt et al. (2006) documented 25 different operationalizations of the DOI construct. The authors noted that most papers employed either the ratio of foreign sales over total sales (FSTS) or some form of linear combination of this and other DOI metrics based on Sullivan's uni-dimensional approach. In the section below, we briefly review how researchers in the field have employed different uni- and multi-dimensional metrics to estimate the DOI of the firm. We identify their collective strengths and weaknesses to best determine how a social dimension of internationalization can contribute to the development of stronger conceptualizations of the DOI construct (see Tables 1 and 2 for a summary of this literature).
2.1 Uni-dimensional Measures
The concept of international diversification is also known in the literature as geographic diversification, geographic scope, global diversification, or simply internationalization (Wiersema and Bowen 2011). Early research in international diversification relied on uni-dimensional measures to estimate the DOI of the firm. In their study of the world's largest multinationals, Stopford and Dunning (1983) introduced the FSTS ratio as a means for measuring firm internationalization. This ratio remains popular among IB researchers, probably because of the widespread availability of this information in popular databases (Wiersema and Bowen 2011). Some years later, Daniels and Bracker (1989) introduced the foreign assets over total assets (FATA) ratio to the analysis of the relationship between firm multinationality and performance. Uni-dimensional ratios of this kind capture the scale or depth of the international activities of the firm and reflect the degree of commitment of the firm to foreign markets. Similar uni-dimensional scale ratios include foreign employees over total employees (FETE) and foreign income over total income (FITI) (Annavarjula and Beldona 2000; Hitt et al. 2006; Thomas and Eden 2004).
A limitation of scale ratios is that they are unable to provide information about the geographical, cultural or functional spread of the firm's international activities. As a result, these ratios ignore the complexity of the firm's foreign footprint. Also, scale ratios do not consider the size of the firm's home market, i.e. potential structural differences in the denominator of the ratio that can make cross-country comparisons...