INTRODUCTION
When firms initiate internationalization activities, they face tremendous challenges from the complex environments of overseas markets. Their lack of awareness of the operating rules in new overseas markets causes salient liabilities of foreignness and outsidership in outward foreign direct investments (OFDIs) (Johanson & Vahlne, Reference Johanson and Vahlne1977, Reference Johanson and Vahlne2009). Such liabilities may be more pronounced for emerging market firms due to limited resources and international experience (Deng, Reference Deng2012, Reference Deng2013). Firms can directly obtain pertinent knowledge through organizational learning to overcome this challenge, which has been well documented in the literature on the internationalization process (Apaydin, Thornberry, & Sidani, Reference Apaydin, Thornberry and Sidani2020; Johanson & Vahlne, Reference Johanson and Vahlne1977; Zollo & Winter, Reference Zollo and Winter2002). Except for experiential learning, the extant literature has long studied learning from other firms through inter-firm ties and promoting internationalization, among which board interlock is a reliable and effective knowledge conduit.
Board interlock is formed by board members serving different firms simultaneously (Mizruchi, Reference Mizruchi1996). It promotes organizational learning processes and induces strategic similarities among diverse businesses (Gulati & Westphal, Reference Gulati and Westphal1999). However, the influence of organizational learning via interlock ties is not fully accomplished immediately. The exploitation and exploration processes involved in the learning process require time to be realized (March, Reference March1991). While studies have highlighted the knowledge transfer mechanism of board interlock, the literature on how board interlock influences internationalization is rather scant (Xia, Ma, Tong, & Li, Reference Xia, Ma, Tong and Li2018). To the best of our knowledge, no research so far has discussed the time dimension of such a relationship. A critical time dimension is board interlock tenure, that is, the time length of affiliation in a board interlock tie (Basuil & Datta, Reference Basuil and Datta2017). Subsequently, how the appointment tenure of the interlock directorates influences inter-firm foreign knowledge exploration and exploitation is an intriguing topic.
This study bridges the gap by exploring how the tenure of board interlock influences organizational learning from other firms and subsequently influences the internationalization of the focal firm. We argue that time may play a positive role by facilitating trust-building in the knowledge exploration through interlock ties. Moreover, the assimilation of knowledge related to OFDI obtained from interlocked firms, such as destination market evaluation, foreign client identification, cross-cultural management of human resources, and tacit knowledge of managing host market partners, takes considerable time to materialize into effective organizational learning. Considering the law of diminishing marginal utility, we further expect that such a positive role of time may depreciate along with the further increase of tenure. Moreover, firms can learn knowledge from others in their social and spatial networks, and studies have remained relatively silent about the interactions across different knowledge sources. To address this research gap, we also investigate how a learning process over the tenure may be mitigated by the geographical proximity with other firms with OFDI experience as another external source of knowledge.
This study makes two theoretical contributions to the internationalization literature. First, it contributes to the social network literature by emphasizing the dynamic role of time in organizational learning through board interlock during internationalization. It identifies the curvilinear learning effect of social networks on firm internationalization. Second, this study highlights the substitution among different knowledge sources, given that firms possess various channels to gain foreign knowledge (Forsgren, Reference Forsgren2002; Johanson & Vahlne, Reference Johanson and Vahlne1977).
THEORETICAL BACKGROUND
Value of Foreign Knowledge in Firm Internationalization
Conducting international business requires sufficient knowledge about the specific destination market to avoid liabilities of foreignness and outsidership. Firms face risk and uncertainty in the internationalization process (Liesch, Welch, & Buckley, Reference Liesch, Welch and Buckley2011). The internationalization process model suggests that market knowledge is critical to overcoming foreign market uncertainty and that experiential learning is a crucial mechanism that deepens the internationalization process (Apaydin et al., Reference Apaydin, Thornberry and Sidani2020; Johanson & Vahlne, Reference Johanson and Vahlne1977). Such knowledge is particularly significant for OFDI, given that OFDI involves a high degree of strategic commitment and high operational risks in a foreign market (Johanson & Vahlne, Reference Johanson and Vahlne1977, Reference Johanson and Vahlne2009). As an advanced mode of foreign market entry, OFDI necessitates that investors have certain advantages and that firms have market intelligence to generate desirable returns (Wei, Zheng, Liu, & Lu, Reference Wei, Zheng, Liu and Lu2014). Owing to the distinct environment of different countries, the knowledge required for international expansion varies profoundly (Johanson & Vahlne, Reference Johanson and Vahlne1977). Such knowledge may cover the culture, economy, legislation, religion, and language perspective of each country (Cuervo-Cazurra, Maloney, & Manrakhan, Reference Cuervo-Cazurra, Maloney and Manrakhan2007). It may also include information about specific foreign market opportunities and the preferences of customers in foreign markets (Lu & Beamish, Reference Lu and Beamish2001). Moreover, it can help firms identify potential partners and competitors in destination markets and shed light on how to search for and create new knowledge locally (Li, Zhang, & Lyles, Reference Li, Zhang and Lyles2015).
Regardless of the primary source of knowledge through experiential learning (Johanson & Vahlne, Reference Johanson and Vahlne1977), firms can also indirectly obtain knowledge from other firms through inter-firm channels (Forsgren, Reference Forsgren2002). They include board interlocks (Ang, Benischke, & Hooi, Reference Ang, Benischke and Hooi2018; Connelly, Johnson, Tihanyi, & Ellstrand, Reference Connelly, Johnson, Tihanyi and Ellstrand2011; Xia et al., Reference Xia, Ma, Tong and Li2018), foreign ownership linkages (Najafi-Tavani, Robson, Zaefarian, Andersson, & Yu, Reference Najafi-Tavani, Robson, Zaefarian, Andersson and Yu2018), international joint ventures (Howell, Reference Howell2018; Minbaeva, Park, Vertinsky, & Cho, Reference Minbaeva, Park, Vertinsky and Cho2018; Sun, Deng, & Wright, Reference Sun, Deng and Wright2021), strategic alliances (Brass, Galaskiewicz, Greve, & Tsai, Reference Brass, Galaskiewicz, Greve and Tsai2004; Gomes-Casseres, Hagedoorn, & Jaffe, Reference Gomes-Casseres, Hagedoorn and Jaffe2006), and employee mobility (Corredoira & Rosenkopf, Reference Corredoira and Rosenkopf2010). Among these channels, board interlock is a reliable and effective conduit to obtain external knowledge and could significantly influence firm strategy (Krause, Wu, Bruton, & Carter, Reference Krause, Wu, Bruton and Carter2019; Useem, Reference Useem1986) because board members have access to core knowledge of the firm (Kroll, Walters, & Le, Reference Kroll, Walters and Le2007) and also have a significant influence on firm strategic decisions (Ravasi & Zattoni, Reference Ravasi and Zattoni2006). Additionally, Connelly et al. (Reference Connelly, Johnson, Tihanyi and Ellstrand2011) explored the competing effect of various types of board interlocks on the diffusion of knowledge and further international strategies. Ang et al. (Reference Ang, Benischke and Hooi2018) proposed that interlocking with multinational enterprises promotes firms’ application of high control expansion modes in the internationalization process. Moreover, Xia et al. (Reference Xia, Ma, Tong and Li2018) studied how network knowledge influences the global strategy of a firm by comparing board interlocking with joint venturing. However, studies have not yet identified the dynamic nature of the relationship between organizational learning through social network and internationalization of firms and have neglected the interactions between different knowledge sources.
Board Interlock and Organizational Learning
Board interlock is an efficient knowledge conduit and the interlock ties between dissimilar firms may cause managerial similarities (Gulati & Westphal, Reference Gulati and Westphal1999). The knowledge embedded in social networks helps firms alleviate bounded rationality and make appropriate decisions (Borgatti & Foster, Reference Borgatti and Foster2003). Board interlock as an effective connection in social networks can promote innovation inputs and outputs (Srinivasan, Wuyts, & Mallapragada, Reference Srinivasan, Wuyts and Mallapragada2018), diversification strategy (Connelly et al., Reference Connelly, Johnson, Tihanyi and Ellstrand2011), and corporate acquisition (Haunschild, Reference Haunschild1993). Recent studies have highlighted the emergence of the influence of board interlocking on firms’ international business, including cross-border Greenfield investments, mergers, and acquisitions (Ang et al., Reference Ang, Benischke and Hooi2018; Connelly et al., Reference Connelly, Johnson, Tihanyi and Ellstrand2011; Xia et al., Reference Xia, Ma, Tong and Li2018). However, most of them have not distinguished the knowledge difference among various foreign markets (e.g., Ang et al., Reference Ang, Benischke and Hooi2018; Connelly et al., Reference Connelly, Johnson, Tihanyi and Ellstrand2011). Moreover, it takes possible delays for the strategic influence of organizational learning as a process of obtaining knowledge when interacting with the external environment (Cyert & March, Reference Cyert and March1963) to fully realize itself (Deng, Jean, & Sinkovics, Reference Deng, Jean and Sinkovics2018; Simon, Reference Simon1991; Zhang, Li, & Li, Reference Zhang, Li and Li2014). The significant role of time in this relationship remains under-examined in the literature.
Organizational learning has an integral role in knowledge transfers as an important consequence of board interlocking (Krause et al., Reference Krause, Wu, Bruton and Carter2019; Useem, Reference Useem1986). Organizational learning involves exploring and exploiting knowledge (March, Reference March1991). Explorative learning through interlock ties involves certain procedures, including trust-building among board members (Vanneste, Puranam, & Kretschmer, Reference Vanneste, Puranam and Kretschmer2014), search and discovery of appropriate knowledge (March, Reference March1991), and knowledge transfer (Gupta, Smith, & Shalley, Reference Gupta, Smith and Shalley2006). Meanwhile, exploitative learning requires firms to assimilate the knowledge, which comprises the extension, reconfiguration, and optimal integration of explored knowledge with existing knowledge (Zollo & Winter, Reference Zollo and Winter2002). Exploitative and explorative learning processes jointly influence the strategies of firms (Stahl & Tung, Reference Stahl and Tung2015). The learning process only becomes complete when the explored knowledge is integrated into the firms’ knowledge base, which makes organizational learning a time-consuming and progressive process (Deng et al., Reference Deng, Jean and Sinkovics2018; Simon, Reference Simon1991; Winter, Reference Winter2000). Organizational learning through board interlock in social networks also involves the process of knowledge exploration and exploitation. However, most international business studies have neglected the temporal nature of this process.
HYPOTHESES DEVELOPMENT
Board Interlock Tenure and OFDI Decisions
The knowledge about a particular foreign market obtained through interlock ties over time could facilitate foreign market entry of the focal firm, and board interlock is a reliable and effective conduit of knowledge (Krause et al., Reference Krause, Wu, Bruton and Carter2019; Useem, Reference Useem1986). In the experience of conducting foreign direct investment, firms could accumulate knowledge about international business. Such knowledge may be location-specific, including information about market opportunities (Zhou, Wu, & Luo, Reference Zhou, Wu and Luo2007) and the culture, religion, language, and legislation perspectives of the institutional environment (Cuervo-Cazurra et al., Reference Cuervo-Cazurra, Maloney and Manrakhan2007; North, Reference North1990). The knowledge could also be general and include technology-related information (Salman & Saives, Reference Salman and Saives2005; Wadhwa & Kotha, Reference Wadhwa and Kotha2006) and knowledge about how to manage inter-firm relationships (Mayer & Argyres, Reference Mayer and Argyres2004; Salman & Saives, Reference Salman and Saives2005; Wadhwa & Kotha, Reference Wadhwa and Kotha2006) or complex environments (Cuervo-Cazurra, Reference Cuervo-Cazurra2011). Such knowledge helps overcome the liability of foreignness in the internationalization process (Eriksson, Johanson, Majkgard, & Sharma, Reference Eriksson, Johanson, Majkgard and Sharma1997; Johanson & Vahlne, Reference Johanson and Vahlne2009). The board members of a firm have access to insider knowledge, which is not readily accessible to outsiders (Haunschild, Reference Haunschild1993), and firms lacking international experience can learn international knowledge through interlock ties. Given that directors serve as influential players in firms’ strategic decisions (Judge & Talaulicar, Reference Judge and Talaulicar2017; Kroll, Walters, & Wright, Reference Kroll, Walters and Wright2008), knowledge transferred through interlock ties affects the focal firm's internationalization decisions. When firms are equipped with abundant knowledge about a specific foreign market, their motivation to enter it increases (Xia et al., Reference Xia, Ma, Tong and Li2018). Therefore, the board interlock with firms with OFDI projects could encourage the focal firm to conduct OFDI in the same country.
However, the learning process through interlock ties that contribute to internationalization takes time to be accomplished (March, Reference March1991). We suggest the time-consuming nature of the learning process by analyzing two approaches in organizational learning: knowledge exploration and exploitation (March, Reference March1991). Before strategic influence from the board interlock comes true, firms first need to explore knowledge from other firms through interlock ties. Mutual trust is the foundation of knowledge sharing and assimilation (Holste & Fields, Reference Holste and Fields2010; Johanson & Vahlne, Reference Johanson and Vahlne2009), whereas inter-personal and inter-organizational trust are not built overnight (Deng, Liesch, & Wang, Reference Deng, Liesch and Wang2021; Vanneste et al., Reference Vanneste, Puranam and Kretschmer2014). Trust-building serves as the premise of knowledge exploration through interlock ties, and the degree of trust-building of board interlocks has an important influence on firm behavior (Tuschke, Sanders, & Hernandez, Reference Tuschke, Sanders and Hernandez2014; Vanneste et al., Reference Vanneste, Puranam and Kretschmer2014). When the board interlock is appointed, limited trust level at the interlocked firm prevents the focal firm from exploring core knowledge in the firm's operations (Sonnenfeld, Reference Sonnenfeld2002). As the trust is built gradually, board interlock could increase access to more core knowledge of the interlocked firm. When the trust level of the board interlock at the focal firm is low, the potential knowledge exploration could also be limited because of low credibility of this knowledge. Such knowledge will become influential when the board interlock grows into a trustworthy source (Carpenter & Westphal, Reference Carpenter and Westphal2001). Thus, as the tenure of the board interlock increases, the degree of trust is low at the beginning but then increases, and the focal firm may explore more knowledge through interlock ties as tenure increases.
Furthermore, the knowledge explored through interlock ties also takes time to be exploited gradually (Simon, Reference Simon1991; Zhang et al., Reference Zhang, Li and Li2014). After accessing such knowledge, time is required to realize its economic value to the firm (March, Reference March1991). Such an exploitative learning process involves the decoding and effective assimilation of knowledge resources (Gupta et al., Reference Gupta, Smith and Shalley2006; Ringberg & Reihlen, Reference Ringberg and Reihlen2008; Simon, Reference Simon1991). Some OFDI knowledge for specific countries obtained through board interlock requires scrutiny and interpretation before its potential value is fully assessed (Ringberg & Reihlen, Reference Ringberg and Reihlen2008). For example, information about local culture and customs requires more effort to decode, while the decoding process of the information about explicit regulation is relatively easier. Besides, applicability of such externally obtained knowledge in the focal firm is subject to various limitations (Gupta et al., Reference Gupta, Smith and Shalley2006; Simon, Reference Simon1991; Zhang et al., Reference Zhang, Li and Li2014). For example, when the focal firm and interlocked firm conduct business in different industries, the product or industry-related knowledge may not be relevant, but the environmental knowledge, such as institutional information, could also be useful for the focal firm. Finally, the focal firm requires a new organizational design, routine, protocol, and team to assimilate the pertinent experience (Perkins, Reference Perkins2014). In summary, fully decoding and assimilating knowledge about OFDI in specific countries that are obtained through board interlocks is a progressive and time-consuming process. Therefore, trust-building in the exploration process and knowledge assimilation in the exploitation process require time. As the tenure of the board interlock increases, firms could iteratively explore and exploit more knowledge about OFDI projects conducted by the interlocked firm, which could enhance the OFDI propensity of the focal firm.
Moreover, we expect that the learning effect through board interlock is subject to the law of diminishing marginal utility. The OFDI propensity of the focal firm will increase at a declining rate because the marginal benefit from trust-building decreases when the tenure of the board interlock increases. When the tenure of the interlock tie is sufficiently high, the mutual trust required for knowledge transfer through this tie can be fully established (Westphal, Reference Westphal1999). Further time commitment exerts a lower marginal influence on trust-building and thus may not effectively contribute to the knowledge exploration. Moreover, as the focal firm becomes familiar with the interlocked firm, knowledge explored from the latter will become fully extracted (Hoang & Rothaermel, Reference Hoang and Rothaermel2005). When a focal firm hosts interlocked directors during its early years, the knowledge from interlocked firms is relatively new and warrants learning (Deng et al., Reference Deng, Liesch and Wang2021). After several years of organizational learning, knowledge from interlocked firms adds limited fresh perspectives for the focal firm (Zhang et al., Reference Zhang, Li and Li2014). Furthermore, some knowledge could be idiosyncratic to interlocked firms and thus difficult to learn and imitate (Nelson & Winter, Reference Nelson and Winter1982). The knowledge that failed to be transferred in the early years of the established interlock tie may possess little value and have low compatibility with the focal firm. Thus, continuous increase of board interlock tenure will have a lower marginal effect on knowledge exploitation. Considering the decreasing marginal benefits from trust-building and the decaying knowledge value as the tenure increases, we hypothesize as follows:
Hypothesis 1 (H1): The tenure of board interlock with firms with OFDI experience in a country has a positive relationship with the OFDI propensity of the focal firm to the same country, and this positive relationship presents a declining trend as the tenure continues to increase.
Moderator: Geographical Proximity as an Alternative Conduit of OFDI Knowledge
In addition to social networks, firms can also explore knowledge in spatial networks. Similar to social networks, spatial networks consist of direct and indirect relations among firms but are confined to a specific geographic area (Molina-Morales & Martinez-Fernandez, Reference Molina-Morales and Martinez-Fernandez2010). Exploring interactions of the learning process from these two different networks helps to enhance the understanding of the learning mechanism from different networks. Geographical proximity refers to the close physical distance between different economic actors (Boschma, Reference Boschma2005), and geographical proximity to firms with OFDI experience in a particular country that could provide the focal firm with OFDI knowledge. Knowledge could be transferred to nearby actors (Lee, Reference Lee2009; Petruzzelli, Reference Petruzzelli2008; Petruzzelli & Murgia, Reference Petruzzelli and Murgia2021) but will decay as distance grows (Knoben & Oerlemans, Reference Knoben and Oerlemans2006). Subsequently, firms can take advantage of geographical proximity to acquire knowledge from other firms (Howells, Reference Howells2002; Petruzzelli, Reference Petruzzelli2008; Petruzzelli, Albino, & Carbonara, Reference Petruzzelli, Albino and Carbonara2007).
When neighbor firms have OFDI experiences in specific host countries, focal firms can learn neighboring OFDI knowledge through various channels. First, the focal firm can observe the best local practices of neighboring enterprises closely, and international knowledge is codified in the practices of firms and can be transferred through the observation of practices (Ringberg & Reihlen, Reference Ringberg and Reihlen2008; Wang & Wu, Reference Wang and Wu2016). Second, firms in the same region develop implicit and explicit connections through the local business community (e.g., business clubs), which provides managers with frequent contacts that promote organizational learning (Howells, Reference Howells2002; Ramasamy, Goh, & Yeung, Reference Ramasamy, Goh and Yeung2006). Third, employee mobility can also create knowledge transfers among firms (Corredoira & Rosenkopf, Reference Corredoira and Rosenkopf2010), with turnover of managers with international experience in the prospective host countries also fostering knowledge flows across firms. All three channels help transfer OFDI-related knowledge from neighboring firms to focal firms.
When interlock tenure is short, OFDI knowledge from board interlocking and neighboring firms is more overlapping than complementary; when interlock tenure is long, OFDI knowledge from the two sources becomes more complementary than overlapping. Knowledge from different sources can be both complementary and overlapping (Han, Jo, & Kang, Reference Han, Jo and Kang2018). The knowledge obtained through board interlock is more informative, trustworthy, and reliable (Krause et al., Reference Krause, Wu, Bruton and Carter2019; Useem, Reference Useem1986) than auxiliary knowledge acquired from neighbors. When firms have multiple overlapping knowledge sources, the benefits of using one of them will be weakened (Kim, Mukhopadhyay, & Kraut, Reference Kim, Mukhopadhyay and Kraut2016). When the tenure increases at the initial stage of board interlock, access to neighboring OFDI knowledge will weaken the uniqueness and value of knowledge from the board interlock, and thus the positive influence of board interlock on OFDI strategies is mitigated.
As the board interlock tenure further increases, knowledge obtained through interlocked directorates becomes less novel (Zhang et al., Reference Zhang, Li and Li2014). Therefore, focal firms will appreciate the value of knowledge acquired through spatial networks as an alternative external source. The knowledge obtained from neighboring firms may help focal firms better extract and assimilate tacit knowledge from interlocked directorates; thus, the complementary effect becomes nontrivial. Therefore, the declining trend in the relationship between board interlock and firm internationalization will be mitigated by the neighboring knowledge. Therefore, we hypothesize:
Hypothesis 2 (H2): Neighboring OFDI knowledge mitigates the effects of board interlock tenure on the focal firm's OFDI in that country in both rising and declining trends.
METHODS
Data
We used a sample of publicly listed firms in China to test our hypotheses for several reasons. First, publicly listed firms are intensely embedded into board networks, and their directors may hold positions on multiple corporate boards (Palmer, Barber, Zhou, & Soysal, Reference Palmer, Barber, Zhou and Soysal1995). Second, China is one of the largest sources of OFDI outflow globally and many firms have invested in OFDI projects (Deng, Yan, & Sun, Reference Deng, Yan and Sun2020). According to the World Investment Report (UNCTAD, 2019: 7), China is the second-largest home country of OFDIs, reaching USD 133 billion in 2020. Third, although emerging market firms have recently become influential in the OFDI process, they encounter significant liabilities when entering foreign markets, which drives them to obtain OFDI knowledge from various channels, such as interlocked and neighboring firms.
Our sample includes listed firms from 2005 to 2014, and we collected data from two primary sources. First, we obtained board interlock information from the China Stock Market & Accounting Research Database (GTA Education Tech Ltd., 2019), which also reports information on firm performance, including balance sheets, income statements, and capital structures of firms, and detailed firm-level information, such as foundation year, location, and governance structure. Second, we collected data from the annual OFDI census of the Chinese Ministry of Commerce (Deng et al., Reference Deng, Yan and Sun2020). To avoid sample selection bias, we included all observations of listed firms with any interlock ties regardless of OFDI experience. Our final dataset includes 21,387 observations of 2,697 listed firms from the Shanghai and Shenzhen Stock Exchanges located across all provinces of China.
Dependent Variable: OFDI Dummy
We adopted an OFDI dummy variable to measure the OFDI propensity of the focal firm. We assigned 1 to the observation if the firm carries out one or more OFDI projects in a specific year and at least one of the destinations of the projects is the same as the OFDI destination of interlocked firms over the past three years; otherwise, 0. When we tracked the OFDIs conducted by the focal and interlocked firms, we observed three possible scenarios. First, the firm had no OFDI that year. In this case, we assigned 0 to the observation. Second, the focal firm had one or more OFDI projects in that year, but the focal firm's interlocked firms had not invested in the destinations over the past three years. For example, a firm has OFDI projects in countries A and B, but its interlocked firms have OFDI projects in countries C and D. In this case, we still assigned 0 to the observations. Third, the focal firm had one or more OFDI projects in that year, and at least one of the destinations had been invested in by the focal firm's interlocked firms over the past three years. For example, a firm has OFDI projects in countries A and B, and its interlocked firms have OFDI projects in countries B and C. In this case, we assigned 1 to the observation.
Independent Variable: Tenure of OFDI Ties
Firms interlocked with peers with OFDI experience in specific countries can obtain the OFDI-related knowledge of those countries through board interlock ties. Consistent with the literature (Xia et al., Reference Xia, Ma, Tong and Li2018), we tracked the OFDI experience of interlocked firms over the past three years as a knowledge source. We regard the board interlock with firms with OFDI experience over the past three years as OFDI ties. This stock measure, rather than an annual flow measure, may help us overcome the potential endogeneity concern caused by reverse causality or omitted variables that simultaneously affect short-term investment motives and OFDI ties (Gulati & Westphal, Reference Gulati and Westphal1999). We employed different time spans to measure the independent variable in our robustness tests which will be discussed shortly. Following the board tenure measurement in the literature (Basuil & Datta, Reference Basuil and Datta2017; Huang & Hilary, Reference Huang and Hilary2018), we measured the tenure of OFDI ties by summing up the tenure of each tie and dividing this by the number of interlocked firms with OFDI experience. For instance, company A is interlocked with companies B and C, both of which have OFDI experience over the past three years. This A–B interlock has existed for two years, whereas the A–C interlock has existed for five years. In this example, the average tenure of the OFDI tie is (2 + 5)/2 = 3.5 years.
Moderating Variable: Neighboring OFDI Knowledge
Subnational variance exists extensively in many countries (Chan, Makino, & Isobe, Reference Chan, Makino and Isobe2010; Deng et al., Reference Deng, Yan and Sun2020; Hutzschenreuter, Matt, & Kleindienst, Reference Hutzschenreuter, Matt and Kleindienst2020). Different provinces in China have experienced different levels of firm internationalization. For instance, the OFDI intensities in southeastern provinces are generally higher than those in central and western provinces. We first obtained data on the number of OFDI projects in the province of corporate headquarters over the past three years. Then, to measure the overlap between the OFDI knowledge from neighbors and that from interlocked firms, we calculated the number of province-level OFDIs whose destinations are the same as the country in which the interlocked firms have invested over the past three years. To scale down the measure, we divided this number by 1,000 and obtained the neighboring OFDI knowledge in the models.
Control Variables
We incorporated several firm and region-level control variables into our models. First, we included the dependent variable (OFDI dummy) in the previous year to control the potential serial correlation and reverse causality. At the firm level, we adopted a logarithm of the total asset of the firm to measure firm size (Kumar & Aggarwal, Reference Kumar and Aggarwal2005). Return on assets (ROA), apart from reflecting the financial capability of a firm, also influences its international investment decisions (Tan & Vertinsky, Reference Tan and Vertinsky1996). Firm age was included to reflect the degree of organizational rigidity and experiential resource accumulation of the firm (Dowell & Killaly, Reference Dowell and Killaly2009). We also included financial leverage, measured by total liabilities divided by total assets, and operating leverage, measured by operating profit divided by gross profit, to reflect the risk tolerance of the firm (Gamba & Triantis, Reference Gamba and Triantis2013). We measured state ownership, which fosters relationships between firms and the government, using the share of state-owned equity in non-current capital stock (Sun et al., Reference Sun, Deng and Wright2021).
Following the literature (Lu, Liu, Wright, & Filatotchev, Reference Lu, Liu, Wright and Filatotchev2014), we measured the OFDI experience of the focal firm using the number of its OFDI projects over the past three years. We also controlled for other sources of external international knowledge, including foreign ownership and international joint ventures. Foreign ownership may link Chinese firms to foreign markets through ownership shares in non-current capital stock (Zou & Adams, Reference Zou and Adams2008). We measured the international knowledge that firms may obtain from international joint ventures using the number of international joint ventures the focal firm invested in. These data were collected from the National Bureau of Statistics of China (Zhou & Li, Reference Zhou and Li2008).
At the regional level, we calculated regional competition using the number of listed firms operating in the same business category in the province (Haveman, Reference Haveman1993). We incorporated industry dummy variables into our models to control the influence of industry-specific factors and year dummy variables to control the periodic disturbance. We lagged all independent, moderating, and control variables in the models by one year to accommodate the time delay between the antecedents and consequences of the OFDI decision.
RESULTS
Descriptive Statistics
Table 1 reports the descriptive statistics and correlation coefficients of the variables. All correlation coefficients between variables are lower than 0.40. We further calculated the variance inflation factors (VIFs) of all moderating, control, and independent variables. The maximum VIF is 1.27, and the average VIF is 1.10, both of which are far below the threshold when assessing the presence of multicollinearity. The average tenure of OFDI ties in Table 1 is 0.554, which is relatively low, given that 71.9% of the observations in our sample have no interlock ties with firms with OFDI experience over the past three years. If we exclude these 71.9% of our observations, then the average tenure of the OFDI ties of firms with at least one valid OFDI interlock tie is 2.56.
Table 1. Descriptive statistics and correlation coefficients
Notes: N = 12,977. Correlation coefficient| ≥ 0.02, significant at p < 0.01.
Baseline Results
Given that the dependent variable is a dummy variable, we used a panel data logit model (xtlogit in Stata 15.0) to run our analyses (Agresti & Kateri, Reference Agresti and Kateri2011), and Table 2 reports the empirical results. Following the literature (Ma, Zhang, Zhong, & Zhou, Reference Ma, Zhang, Zhong and Zhou2020; Shi, Sadowski, Li, & Nomaler, Reference Shi, Sadowski, Li and Nomaler2020), we report both the coefficients and odds ratio to illustrate the effect size of explaining variables on the dependent variable. We set firm serial number as the identifier variable and year as the time variable and, therefore, used a firm-year combination as our unit of analysis. As our baseline model, Model 1 includes all control and moderating variables. Meanwhile, we include the tenure of OFDI ties in Model 2. As illustrated in the table, the tenure of OFDI ties has a statistically positive relationship with the OFDI decision of the firm at the 1% level. The odds ratio reveals that for each one-year increase in board interlock tenure with firms with OFDI experience, the probability ratio of the firm OFDI increases by 24.0% when the board interlock tenure with firms with OFDI experience increases by a year.
Table 2. Logit regression of panel data
Notes: N = 12,977. F-value or log-likelihood χ 2 is significant at p < 0.01 in each model. p-values in parentheses.
In Model 3, we added the squared term of tenure of OFDI ties. In this model, the tenure of OFDI ties has a consistently positive effect on firms’ OFDI at the 1% level, whereas the squared term exerts a negative effect at the 1% level. However, the turning point of the curvilinear relationship is at 3.62 years, which is even at the right-hand side of the 95th percentile of the observations. Therefore, instead of an inverted U-shape relationship, the tenure of OFDI ties positively affects the OFDI decision in all value ranges of the independent variable. In other words, the positive marginal effect of H1 demonstrates a diminishing trend, but it will never become negative. Following the literature (Zelner, Reference Zelner2009; Zhang et al., Reference Zhang, Li and Li2014), we plotted this relationship in Figure 1 to facilitate interpretation. The sign in the vertical axis is negative because the positive effect has been treated with a logarithm transformation. The tenure of OFDI ties in Figure 1 has an overall positive effect on the OFDI decision of the firm, whereas the slope of this positive relationship diminishes as the tenure continues to increase. Therefore, Hypothesis 1 is supported.
Figure 1. Tenure of OFDI ties and predicted OFDI likelihood (in logarithm)
We then added the interaction term of neighboring OFDI knowledge with the tenure of OFDI ties and its squared terms into Model 4, both of which are significant at the 1% level. The odds ratios for these terms are 0.312 and 1.164, respectively. These results suggest that the influence of the board interlock tenure and its squared term on the probability ratio of the firm OFDI decreases and increases, respectively. Therefore, Hypothesis 2 is supported. To facilitate interpretation, we plotted the interaction effect in Figure 2. Compared with the curve with a low value of neighboring OFDI knowledge, that with a high value of neighboring OFDI knowledge becomes flatter.
Figure 2. Moderating effects
Robustness Tests
To examine the robustness of our findings, we ran several sets of tests. For the brevity of the article, we outlined the regression results of robustness tests without providing the corresponding tables. First, we redefined board interlock with firms that have conducted OFDI over the past five, ten, or all historical years, instead of over the past three years, as an OFDI tie. Consistent with this measurement, we also re-measured the moderator using the number of OFDI projects in the province over the past five, ten, or all historical years when the destinations were the same as the countries that the interlocked firms have invested in over the same period. The results are consistent with those in Table 2, firmly supporting both hypotheses.
Second, we incorporated another time dimension into our measurement of the independent variable. Board members with a longer tenure in the focal firm have more power over its strategic decisions (Golden & Zajac, Reference Golden and Zajac2001). We multiplied the tenure of OFDI ties by the serving tenure of board members, took the sum of the product term for each individual tie, and divided it by the number of interlocked firms with OFDI experience. The empirical results consistently support all hypotheses.
Third, we adopted an instrumental variable approach to address the potential endogeneity. In line with the extant literature (Patro, Zhang, & Zhao, Reference Patro, Zhang and Zhao2018), we used the industrial average value of the tenure of OFDI ties as the instrumental variable. The industry- and firm-level director tenures are likely to be highly correlated, but the industrial director tenure could be less related to the corporate OFDI decisions. Thus, it is an appropriate instrumental variable. Since the main models in this study are nonlinear models for the panel data, we adopted an instrumental variable probit model (ivprobit in Stata) and clustered it around the firm identifier. To obtain algorithm convergence in these ivprobit models, we omitted industrial and year dummies. The Wald χ 2 tests confirm the endogeneity of the tenure of OFDI ties. The results are consistent with those in Table 2, further supporting all hypotheses.
DISCUSSION
Theoretical Contributions
This study offers two contributions to the internationalization literature. First, it emphasizes the time dimension and curvilinear nature of organizational learning in the social network, which has been neglected in the social network and internationalization literature. Firms can learn foreign market knowledge from social networks to overcome the liabilities of foreignness and outsidership in the process of internationalization (Deng, Reference Deng2013; Johanson & Vahlne, Reference Johanson and Vahlne1977, Reference Johanson and Vahlne2009). As an effective knowledge conduit, board interlock may provide firms with important enabling knowledge (Krause et al., Reference Krause, Wu, Bruton and Carter2019; Useem, Reference Useem1986). Based on our analysis of the time-consuming yet curvilinear nature of the learning process, our findings suggest that organizational learning through interlock ties promotes the internationalization of firms when interlock tenure increases, but such a relationship exerts a diminishing marginal effect as the tenure increases continually. Studies on the influence of social networks on firm internationalization tend to be static (e.g., Ang et al., Reference Ang, Benischke and Hooi2018; Xia et al., Reference Xia, Ma, Tong and Li2018) and ignore the temporal nature of such effect. However, the exploitation and exploration of knowledge take time to be gradually accomplished. Through the analysis of the learning curve and diminishing marginal learning effect through the board interlock, we underscore that time is a critical boundary condition of the learning process in social networks as well as a crucial dimension in internationalization process studies (Zhang, Li, Li, & Zhou, Reference Zhang, Li, Li and Zhou2010).
Second, this research extends the internationalization process literature by revealing the interactions among different sources of foreign knowledge. Firms obtain knowledge from diverse sources to reduce information asymmetry (Boeh, Reference Boeh2011). Social and spatial networks are both informative sources of knowledge. Our study confirms the presence of substitutions between different types of knowledge sources that have been relatively ignored in the internationalization literature. The simultaneous existence of multiple knowledge sources weakens their independent roles in firm internationalization. Businesses need to choose among diverse knowledge sources when they expend resources to seek knowledge. This study, therefore, provides a framework of multi-channel foreign knowledge and extends our understanding of the association between knowledge acquisition and firm internationalization.
Limitations and Future Research Directions
First, potential knowledge complementarity may be considered when nuanced board information is available. Various knowledge sources have both overlapped components and different values, and in this study, we mainly discussed the trade-off between different knowledge sources. However, potential complementarity deserves further exploration. More detailed data on each external directors’ OFDI experience could allow researchers to pin down their exact contributions and the potential complementary effects between board interlock and alternative sources of OFDI knowledge.
Second, apart from board interlock, other sources of external knowledge in social networks, such as foreign ownership linkages, joint ventures, and strategic alliances, also serve as knowledge conduits. In our study, we controlled for the influence of foreign ownership and international joint ventures, apart from neighboring knowledge sources as a moderator. When data on other inter-firm ties are accessible, future studies could explore if and how other ties in social networks play the role of knowledge conduits and affect the corporate internationalization process over time. Moreover, firms could also learn from neighbor firms of their subsidiaries. The knowledge transfer from neighbors to subsidiaries and then to the headquarters is quite an interesting topic that deserves comprehensive research in the future.
Third, each board member may have a different level of influence on the strategic decisions of firms. However, the empirical studies based on archive data analyses cannot capture these differences. In the robustness test of this study, we added tenure of the board member in the focal firm to the measurement of the IV, which is an attempt to capture the different influences on the strategic decisions by board members. Future studies can further analyze these differences through case studies, which could incorporate detailed information about the role of board members in the focal firm.
Managerial Relevance
This study has rich managerial implications for firms that intend to conduct OFDI. To overcome the liability of information asymmetry, firms could search for foreign knowledge directly or indirectly. According to our results, instead of treating board members as a knowledge repository, firms should optimize them as a conduit to acquire new knowledge externally, thus assisting firms in making effective strategic decisions. Moreover, the firm owner should consider the tenure of the interlock tie because the marginal effect of organizational learning decreases along with increasing interlock tenure. This finding highlights the value of retaining the acquired recent and fresh knowledge by regularly rotating board members who hold external positions and ‘exotically’ valuable knowledge. The moderating results in this study also demonstrate that scanning the external market for useful OFDI knowledge is an effective method that may reduce overdependence on the knowledge learned through board interlocks. Geographic agglomeration is a possible substitutive knowledge source for firms. However, it is worth noting that the overlapping between multiple knowledge sources may weaken the possible benefits of each. In other words, less is more for multi-channel knowledge sources. Therefore, managers should deploy their limited resources wisely to obtain knowledge from certain sources and avoid knowledge redundancy.
CONCLUSION
This study enhances our understanding of the temporal role of organizational learning through social networks and its influence on firm strategies by examining how board interlock of firms influences their internationalization decisions over time and how multiple knowledge sources interactively affect these decisions. Through empirical tests, we find that board interlocks with firms with OFDI experience can provide OFDI-related knowledge that helps them overcome the liabilities of foreignness and outsidership and apply effective internationalization strategies. Such influence changes over time because of the temporal and dynamic nature of organizational learning. The organizational learning effect unfolds as the tenure of the interlock tie increases, while its marginal effect demonstrates a diminishing trend. This study identifies an important alternative knowledge source – spatial network – and verifies its interactions with organizational learning through interlock ties.
DATA AVAILABILITY STATEMENT
We may not publish the data due to third-party property rights. GTA Education Tech Ltd. possesses the copyright of China Stock Market & Accounting Research Database containing the board interlock information. The Ministry of Commerce in China possesses the copyright of the information on the OFDI projects of the sample firms.
