A Dynamic Model of Internationalization and Innovation in Emerging Market Enterprises: Knowledge Exploration, Transformation, and Exploitation

Abstract

Drawing on organizational learning theory and taking an institutional perspective, this study investigates (1) the dynamic relationship between internationalization and innovation in emerging market enterprises (EMEs), and (2) how state ownership moderates the focal relationships. Using a panel dataset of listed Chinese firms from 2007 to 2018, we find that internationalization encourages innovation input in EMEs, which in turn transforms into more innovation output. Higher innovation output leads to further international commitment, creating a dynamic upward spiral of internationalization and innovation. Interestingly, state ownership positively moderates the innovation input-innovation output link but negatively moderates the relationship between innovation output and internationalization. Our paper enriches and refines our understanding of the dynamic relationship between internationalization and innovation in EMEs by integrating the knowledge exploration, transformation, and exploitation perspectives, with the institutional perspective of state ownership.

Introduction

Internationalization and innovation are two crucial strategies for emerging market enterprises (EMEs) to survive and prosper (Piperopoulos et al., 2018; Wu et al., 2016; Xie & Li, 2018). Over the last few decades, firms from emerging economies such as China, Russia, Brazil, and India have been active in international markets and become major players in the global economy (Cuervo-Cazurra et al., 2014; Kumar et al., 2020). Concurrently, to address their latecomer disadvantages, EMEs are trying to catch up with their developed world competitors by engaging intensively in innovation activities (Piperopoulos et al., 2018; Xie & Li, 2018). According to the Global Innovation Index 2020 report (GII, 2020: 24): “This year, the geography of innovation is continuing to shift, as evidenced by the GII rankings. Over the years, China, Vietnam, India, and the Philippines are the economies with the most significant progress in their GII innovation ranking over time. All four are now in the top 50.” As EMEs are increasingly involved in both internationalization and innovation activities, an important question for scholars, practitioners, and public policy makers is: how do the two strategies of internationalization and innovation in EMEs influence each other over time? This question is also important in the light of the United Nations’ Sustainable Development Goals that recognize the significance of internationalization and innovation in securing global economic development, peace, and prosperity for all (Lewis et al., 2021; Montiel et al., 2021).

Previous research on the relationship between internationalization and innovation is mainly based on two hypotheses: the self-selection hypothesis and the post-entry hypothesis. According to the self-selection hypothesis, innovation is a pre-requisite for firm internationalization. For example, early IB studies on multinational enterprises from advanced economies assume that firms expand overseas to exploit their existing ownership advantages such as innovation resources (e.g., Dunning, 1980; Vernon, 1966). In contrast, the post-entry hypothesis regards internationalization as a contributing factor for firm innovation. For example, recent studies suggest that internationalization facilitates organizational learning and knowledge sharing which are important building blocks for innovation (e.g., Chang et al., 2019; Love & Ganotakis, 2013; Tse et al., 2017). In addition, while most studies suggest a positive relationship between internationalization and innovation, some others find a negative or a statistically non-significant relationship, especially in small or resource constrained firms (Bahl et al., 2021; Kumar, 2009). There is therefore a continuing debate on the direction and sign of the relationship between internationalization and innovation, that is, whether innovation leads to internationalization or vice versa, and are they positively or negatively associated.

Notwithstanding the extensive literature on the internationalization-innovation relationship, existing research tends to take a static view to investigate this research question (Chiva et al., 2014). They either focus on the impact of innovation on internationalization (Azar & Ciabuschi, 2017; Oura et al., 2016), or on the effect of internationalization on innovation (Chang et al., 2019; D’Angelo et al., 2020; Piperopoulos et al., 2018). Although a few papers (e.g., Filipescu et al., 2013; Monreal-Pérez et al., 2012) have investigated the relationship between internationalization and innovation in both directions, they examine the post-entry and self-selection hypotheses independently, and the results remain mixed. As a result, how internationalization and innovation (input and output) influence each other dynamically remains unclear. Dynamic relationship here refers to the evolution of the relationship between internationalization and innovation over time. Furthermore, earlier studies, especially on the innovation to internationalization link are mainly based on firms from developed markets (Azar & Ciabuschi, 2017; Henley & Song, 2020), and these theories and empirical findings may not be generalizable to understand the internationalization-innovation relationship in EMEs. In contrast with firms from developed markets, EMEs become more innovative after going international, and often take a different trajectory to internationalize (Kumar et al., 2020; Luo & Tung, 2018). In general, EMEs lag developed country firms and lack experience in both internationalization and innovation activities (Buckley & Hashai, 2014; Nuruzzaman et al., 2019). Hence, there is a critical need for additional research focusing specifically on EMEs to understand the unique and dynamic nature of the internationalization-innovation phenomena in this specific context.

In addition, state ownership of large international enterprises is common in emerging markets, with governments continuing to play a critical role in national economic activities and firms’ decision-making (Cuervo-Cazurra et al., 2014; Feng et al., 2018). For example, China has 145 firms on Fortune Global 500 list in 2022, more than half of which are central/local state-owned enterprises (SOEs). Likewise, the SOEs in Vietnam are trying to reach out to the world to seize more opportunities (Vietnamnet, 2022). In response to this trend, scholars are increasingly interested in understanding the impact of state ownership on EME activities, mainly from an institution-based perspective (Hong et al., 2015; Liang et al., 2015). In recent years, researchers have also revisited state capitalism, focusing on whether government control may stimulate or hinder firms’ competitive advantage, and how various forms of state ownership affect firm decisions (e.g., Bass & Chakrabarty, 2014; Cui & Jiang, 2012). With growing importance of SOE internationalization in international business (IB) research, it is crucial to understand the effect of state ownership on the determinants and outcomes of EME internationalization (Cuervo-Cazurra et al., 2014; Wang et al., 2012).

Despite growing interest in the role of government in EMEs’ internationalization and innovation, there is limited research that examines the impact of state ownership on the internationalization-innovation relationship, especially in the context of developing countries. From an institution-based view, state ownership enables EMEs to gain political legitimacy, and to enforce business contracts or stop unlawful imitations by foreign rivals (Yi et al., 2017; Zhou et al., 2017). Some other studies, however, have examined the costs of business-government relationship (Meyer et al., 2014), and argued that political connections with the home country government could be counterproductive (Zhou et al., 2017). It is therefore important to examine if, on balance, the state has a positive or a negative moderating effect on the relationship between internationalization and innovation.

In sum, the objective of our study is to address two key research questions: (1) what is the nature of the dynamic relationship between internationalization and innovation in EMEs, and (2) how state ownership moderates the internationalization-innovation relationship. Overall, our study seeks to make three key contributions: First, we develop a dynamic theory of the two-way relationship between internationalization and innovation. Unlike previous research that takes a static view to understand the relationship between internationalization and innovation, our study unpacks the “black box” by illustrating how internationalization and innovation contribute to each other over time in the context of EMEs. Second, we identify three phases in the dynamic relationship, namely, knowledge exploration, knowledge transformation, and knowledge exploitation. By doing so, we introduce knowledge transformation as the bridge between knowledge exploration and exploitation activities. Although knowledge exploration and exploitation have attracted extensive attention in prior research, how firms make the transition between these two activities has been largely neglected. Third, our study investigates how the institution of the state can help or hinder the internationalization-innovation link. By doing so, we contribute to the literature on internationalization and innovation by including firm-level institutional variable (that is, state ownership) as a crucial contextual factor that enhances or weakens the focal relationship between internationalization and innovation. From a practical perspective, our study will not only help managers in designing effective internationalization and innovation strategies for EMEs, but also guide policymakers in crafting appropriate state intervention to advance the innovation and internationalization agenda of EMEs.

The remainder of this paper is organized as follows. The following section presents a brief literature review of the major theories on the relationship between internationalization and innovation. Next, we provide the theoretical background and hypotheses for our study. We then present the methods, including data, variables, and our analytical approach to test our hypotheses. This is followed by our results and their discussion, including how our study contributes to international business theory, practice, and policymaking. We conclude with our study’s limitations and possible directions for future research.

Existing Research on the Relationship Between Internationalization and Innovation

Innovation and internationalization constitute two alternative pathways for firms to grow and utilize their resources (Ansoff, 1965; Prashantham, 2008). As both forms of strategies are built upon firm resources and capabilities, growth along one dimension is likely to be systematically associated with growth along the second one (Chiva et al., 2014; Kumar, 2009). Therefore, scholars in international business (IB) and innovation research have long been interested in the relationship between innovation and firm internationalization (Chetty & Stangl, 2010; Hitt et al., 1997; Kumar, 2009).

There are two main theoretical rationales for a positive relationship between internationalization and innovation. First, according to early economics studies, firms have the motivation to geographically diversify to achieve higher returns on their investments in innovation activities (e.g., Caves, 1971; Vernon, 1966). This argument is reasonable since internationalization could improve the “appropriability regime” of innovation resources (Teece, 1986). As firms operating only in domestic markets may find it difficult or time consuming to recoup initial investments in producing innovation, they expand overseas to allocate the costs of innovation activities (Vernon, 1966). That is, innovative firms may “self-select” into the international market, and innovation is therefore considered a determinant of firm internationalization. Moreover, according to the resource-based view (RBV), innovation is a major source for firms to gain competitive advantage overseas (Barney, 2001; Roper et al., 2002). Thus, firms with innovation capabilities are better able to internationalize successfully (Bodlaj et al., 2020; Saridakis et al., 2019).

Although the “self-selection hypothesis” has been widely supported in research, scholars have suggested an alternative inducement for firms to engage in international activities. That is, firms may expand overseas to improve their innovation performance, as they can tap into diverse knowledge sources in foreign countries (e.g., Salomon & Shaver, 2005; Tse et al., 2017). Labeled as the “post-entry hypothesis”, this stream of research suggests that firms internationalize to benefit from knowledge spillovers in international markets, learn from foreign operations, and facilitate their innovation development (Castellani et al., 2017; Tse et al., 2017). With the emergence of EMEs in the international market, recent literature has shifted the focus to examine if internationalization can help firms to become more innovative (Wu et al., 2016; Xie & Li, 2018).

A summary of some key papers on the relationship between innovation and internationalization is included in Table 1. As shown in the table, the relationship between internationalization and innovation has been studied extensively. However, with few exceptions (e.g., Filipescu et al., 2013; Monreal-Pérez et al., 2012), existing studies either focus on the impact of innovation on internationalization (the “self-selection” hypothesis) or emphasize the learning effect of internationalization on innovation (the “post-entry” hypothesis). As a result, how internationalization and innovation influence each other dynamically over time is underexplored. Although a few studies such as Monreal-Pérez et al. (2012) and Filipescu et al. (2013) investigate the two-way relationship between internationalization and innovation, their research is based on developed market enterprises, and the results remain mixed. For instance, by using a longitudinal dataset of Spanish firms, Monreal-Pérez et al. (2012) find evidence for the self-selection hypothesis, but the firms do not experience a significant post-entry effect during the observation period. In the study by Filipescu et al. (2013), however, they find support for both the self-selection hypothesis and post-entry hypothesis in Spanish firms.

Table 1.

A summary of key papers about the relationship between innovation and internationalization

Study	Journala	Country	Research topic	Theoretical perspective(s)b	Major findingsc	Moderating factor(s)
Hitt et al. (1997)	AMJ	/	Post-entry	OL, RBV	(+) Internationalization is positively related to firm innovation, and product diversification negatively moderates this relationship	Product diversification
Wakelin (1998)	RP	UK	Self-selection	Firm behaviour theory	(±) For large firms, there is a positive relationship between innovation and internationalization. For small firms, the relationship is negative	Firm size
Basile (2001)	RP	Italy	Self-selection	Economics theory	(+) Innovative firms has a higher level of export intensity than non-innovative firms	/
Roper and Love (2002)	RP	UK, German	Self-selection	RBV	(+) Being innovative has a positive effect on export probability	/
Salomon and Shaver (2005)	JEMS	Spain	Post-entry	OL	(+) Exporting has a positive relationship with innovation	/
Pla-Barber and Alegre (2007)	IBR	France	Self-selection	Technology-gap theory	(+) There is a positive relationship between innovation and export intensity	/
Salomon and Jin (2008)	JIBS	Spain	Post-entry	OL	(+) The positive effect of export on innovation is stronger in technologically lagging industries	Leading/lagging industries
Salomon and Jin (2010)	SMJ	Spain	Post-entry	OL	(+) Exporting leads to ex post increase in innovative productivity. However, technologically leading firms apply for more patents than technologically lagging firms	Technological leading/lagging
Monreal-Pérez et al. (2012)	IBR	Spain	Both	/	(+ /n.s.) Innovation leads to more export activities, but the impact of export on innovation is not significant	/
Love and Ganotakis (2013)	IBR	UK	Post-entry	OL	(+) Exporting encourages high-tech small and medium-sized enterprises (SMEs) innovate subsequently	/
Filipescu et al. (2013)	JIM	Spain	Both	RBV	(+) There is a two-way causal relationship between internationalization and innovation	/
Yi et al. (2013)	IBR	China	Self-selection	RBV, TCT, IBV	(+) The relationship between innovation and international performance is influenced by the institutional setting in which the firm is embedded	Institutional setting
Golovko and Valentini (2014)	GSJ	Spain	Post-entry	OL	(+) Both large firms and SMEs engage more in innovation after they internationalize	Firm size
Ren et al. (2015)	IBR	China	Post-entry	OL	(±) Internationalization has a positive effect on innovation performance when SMEs’ R&D capability or marketing capability is high. However, the effect of internationalization on innovation performance is negative when R&D capability or marketing capability is low	R&D capacity, marketing capacity
Oura et al. (2016)	IBR	Brazil	Self-selection	RBV, DCP	(+) There is a positive relationship between innovation and international performance in SMEs	/
Wu et al. (2016)	JWB	China	Post-entry	IBV	(+) Host country institutional development is positively related to EMEs’ innovation performance	Institutional Development
Castellani et al. (2017)	IBR	Multiple countries	Post-entry	OL	(+) Multinationality has a positive effect on R&D intensity
Azar and Ciabuschi (2017)	IBR	Sweden	Self-selection	Social-technical theory	(+) Organizational innovation leads to better international performance both directly and indirectly through enhancing technological innovation
Tse et al. (2017)	JOM	China	Post-entry	OL	(+) Firm internationalization is positively associated with innovativeness
Xie and Li (2018)	JIBS	China	Post-entry	Innovation theory, IBV	(+) There is a positive impact of exporting on innovation, but the effect is influenced by institutional factors
Piperopoulos et al. (2018)	RP	China	Post-entry	OL	(+) Internationalization has a positive effect on innovation performance. This effect is stronger when the host country is developed rather than emerging
Galbreath (2019)	JBE	Australia	Post-entry	IBV, RBV	(+) Export intensity is positively related to green innovations, and women in leadership roles enhances the relationship
Thakur-Wernz and Samant (2019)	GSJ	India	Post-entry	OL	(+) There is a positive relationship between international experience and innovation performance
Saridakis et al. (2019)	JBR	UK	Self-selection	RBV	(+) Innovative SMEs are more likely to export than non-innovative SMEs
D'Angelo et al. (2020)	IBR	Italy	Post-entry	OL	(−) A rapid increase in export breadth reduces the firm’s probability of developing new innovative outputs
Bodlaj et al. (2020)	JBR	Italy	Self-selection	RBV, DCP	(+) Marketing innovations are positively related to export growth, and this relationship is moderated by geographic diversification
Bahl et al. (2021)	JWB	Multiple countries	Post-entry	OL	(−) In transaction economies, internationalization is negatively associated with firm innovation

^aAMJ Academy of Management Journal, RP Research Policy, JEMS Journal of Economics & Management Strategy, IBR International Business Review, JIBS Journal of International Business Studies, SMJ Strategic Management Journal, JIM Journal of International Marketing, GSJ Global Strategy Journal, JWB Journal of World Business, JOM Journal of Management, JBE Journal of Business Ethics, JBR Journal of Business Research

^bOL Organizational Learning, RBV Resource-Based View, TCT Transaction Cost Theory, IBV Institution-Based View, DCV Dynamic Capability Perspective

^c(+) means the relationship is positive; (−) means the relationship is negative; (±) means both positive and negative relationships are found; (n.s.) means the relationship is not statistically significant

Moreover, although a positive relationship between internationalization and innovation is suggested by most of previous literature, some of them found a negative or not statistically significant relationship (e.g., Bahl et al., 2021; D'Angelo et al., 2020; Ren et al., 2015; Wakelin, 1998). A possible reason for this is that the relationship between innovation and internationalization is not only complex and dynamic, but also context dependent. As both internationalization and innovation activities are highly risky and uncertain, their relationship is likely contingent on internal and external factors. For example, internal factors such as the absorptive capacity (Ren et al., 2015), marketing capacity (Ren et al., 2015) and firm size (Wakelin, 1998) are found to influence the learning effect of internationalization on innovation. Likewise, external factors such as country of origin (Bahl et al., 2021), industrial environment (Salomon & Jin, 2008) and institutional quality (Xie & Li, 2018) may also strengthen or weaken the relationship between internationalization and innovation. Therefore, critical contingent/moderating factors need to be included in our theoretical models to improve our understanding of the relationship between internationalization and innovation. One such factor in EMEs is state ownership, which has a crucial impact on firm decisions, but its role on the relationship between internationalization and innovation is largely neglected (Cui & Jiang, 2012; Zhou et al., 2017).

In sum, understanding how EMEs’ internationalization and innovation influence each other dynamically over time is important for the growth and survival of EMEs in international markets. Furthermore, as governments in emerging markets continue to play a significant role in economic activities, it is important to understand how SOEs and non-SOEs differ in their resource endowments and firm strategies (Cui & Jiang, 2012). Accordingly, this study aims to understand the dynamic relationship between internationalization and innovation in EMEs, and how state ownership moderates this relationship.

Theoretical Background and Hypotheses Development

Our study adopts two theoretical lenses to address the research questions. For the first research question, we adopt the organizational learning theory to examine the dynamic relationship between internationalization and innovation, as organizational learning underpins both internationalization and innovation. Organizational learning theory provides a rich perspective on the processes that organizations use to generate and utilize knowledge, and therefore has been widely applied to explain the development of innovation (Lavie et al., 2010; March, 1991). Organizational learning theory also underpins international business (IB) research, as internationalization is a learning process in which firms acquire, integrate, and utilize new knowledge (Johanson & Vahlne, 1977). For the second research question, we use the institution-based view to explore the moderating effect of state ownership on internationalization-innovation relationship. The institution-based view focuses on the interaction between institutions and organizations and how firm behavior is shaped by the surrounding institutions such as governments, clients, and investors (North, 1991; Peng, 2003; Scott, 2013). As government is one of the most influential institutions in emerging markets, we focus on state ownership and how it might affect a firm’s competitive advantage and its strategic decisions in relation to internationalization and innovation (Bass & Chakrabarty, 2014; Wang et al., 2020).

Organizational Learning Theory: The Dynamic Relationship Between Internationalization and Innovation in EMEs

In his landmark work on exploration and exploitation in organizational learning, March (1991) explained the basic distinction between these two kinds of organizational behaviours. Exploration involves organizations engaging in activities such as search, experimentation, and variation, whereas exploitation involves learning through refining and extending existing competencies (March, 1991). In subsequent work, Levinthal and March (1993: 105) limit the scope of these activities to the knowledge domain, stating that exploration involves “pursuit of new knowledge”, while exploitation involves “the use and development of things already known”. More recently, research further elaborates absorptive capacity as an organizational learning process in three phases: knowledge exploration, knowledge transformation, and knowledge exploitation (Lane et al., 2006; Sun & Anderson, 2010; Waddell & Pio, 2015). Knowledge transformation is the bridge between knowledge exploration and exploitation, wherein the newly acquired knowledge is turned into something ready to be exploited. Knowledge transformation therefore involves integrating or modifying the knowledge which could potentially change the nature of knowledge itself (Lavie et al., 2010). Our introduction of the concept of knowledge transformation linking the exploration and exploitation activities shows how these two activities influence each other over time. Overall, based on organizational learning theory, we regard the dynamic relationship between internationalization and innovation in EMEs as a learning process, in which EMEs explore, transform, and exploit knowledge.

Phase 1: Knowledge Exploration

The literature on international EMEs examines the phenomenon of explorative internationalization to explain why internationalization by EMEs may be different from internationalization by traditional multinational enterprises (MNEs) from developed markets (Luo & Tung, 2007, 2018; Mathews, 2017). As Luo and Tung (2007, 2018) indicate, the motivations and challenges to invest abroad faced by firms from emerging economies are different. Compared with conventional MNEs, international EMEs often lack strong firm-specific advantages. To overcome this disadvantage, EMEs are motivated to internationalize to explore strategic resources overseas, and external sources of knowledge are a major consideration (Kumar et al., 2020; Luo & Tung, 2007, 2018).

Moreover, the spread of knowledge centres at various locations throughout the world has strengthened the incentives for EMEs to explore the “global sourcing” of innovation (Govindarajan & Ramamurti, 2011; Lam, 2003). Each host country has unique resource endowments and specific knowledge which may not be available in EMEs’ home countries (Xie & Li, 2018). Because of this, EMEs explore new knowledge and information to capture existing as well as future opportunities in a changing international business environment. Foreign firms’ willingness to sell or share their know-how, technology, or brands further makes it possible for EMEs to fulfil their need for innovation capability development (Child & Rodrigues, 2005; Luo & Tung, 2007).

As EMEs internationalize, they need additional R&D inputs in the form of engineers and scientists who can search, select, and source new ideas from around the world. As the knowledge and information in international markets is highly diverse and scattered, a larger amount of investment in innovation input is necessary for EMEs to search and explore knowledge overseas. In addition, there is a second-order need to increase innovation inputs to absorb, assimilate, and recombine the knowledge obtained from international sources to create new sources of competitive advantage. Due to learning barriers caused by different technology standards, cultural distance, and the liability of foreignness and emergingness, it is challenging for EMEs to identify and absorb new knowledge and information from foreign markets. The growing need to make sense of the large pool of knowledge and information obtained from diverse international sources requires large investments by EMEs into various forms of innovation input infrastructure, including R&D labs, buildings, equipment, and manpower (Tang et al., 2019). In addition, the higher learning potential stemming from international markets is likely to increase the pay-offs associated with innovation investments (Castellani et al., 2017), and create stronger incentives for EMEs to increase innovation input. Therefore, international EMEs need to make considerable investments in innovation inputs to search, source, transfer, absorb, assimilate, and internalize the knowledge and information that is accessed during the internationalization process. Hence, we propose the following hypothesis:

Hypothesis 1 (H1) Internationalization has a positive effect on innovation input in EMEs.

Phase 2: Knowledge Transformation

As discussed above, in the knowledge exploration phase, internationalization leads to more innovation input in EMEs. However, innovation input itself does not generate value until it is transformed into something valuable, in the form of innovation output. During the knowledge transformation phase, innovation input is transformed into innovation output, whereby the newly acquired resources and information are combined with the existing knowledge of the firm to create innovation output such as new products and patents (Lane et al., 2006). At its most fundamental level, innovation output is a function of the efforts put into the innovation process, and innovation input is usually found to be positively associated with innovation output (Duran et al., 2016). As Hall et al. (1986: 265) observe: “the annual research and development expenditures of a firm are considered to be investments which add to the firm’s stock of knowledge.” Therefore, firms need to transform their knowledge input stock into knowledge output flow by creating new patents, products, and services.

Innovation inputs are essential investments in R&D intensive resources such as people, plant, and equipment that enhance the ability of EMEs to successfully produce innovation outputs in the form of new products and patent applications. Higher innovation inputs can encourage firms to invest in basic R&D activities that are costlier and have longer lead times for success, but which can result in novel discoveries with large pay-offs in the long-term. Higher innovation inputs can also support large-scale R&D, involving a large and diverse pool of engineers and scientists who can find novel recombination of knowledge spanning diverse fields such as medicine, IT and engineering, and spur the creation of novel, hitherto unimaginable, products and patents, as seen recently in the rapid invention of novel Covid-19 vaccines. Over the long-term, innovation investment enhances EMEs’ capabilities and gives them the experience to turn the knowledge resources they acquired overseas into tangible innovation outcomes (Klomp & Van Leeuwen, 1999; Ren et al., 2015). For example, Klomp and Van Leeuwen (1999) find that firms that perform R&D regularly have a significant higher level of innovation output than firms that do not perform R&D on a continuous basis. Likewise, Ren et al. (2015) suggest that R&D input helps generate new information that is beneficial to the innovation performance of firms. Hence, we propose the following hypothesis:

Hypothesis 2 (H2) Innovation input has a positive effect on innovation output in EMEs.

Phase 3: Knowledge Exploitation

In the third phase, EMEs exploit their innovation outputs to further extend and expand their international activities. Knowledge exploitation involves matching firm knowledge with market opportunities and applying existing knowledge for commercial ends (Lane et al., 2006). In the context of internationalization, knowledge exploitation involves the use and application of firm’s existing knowledge beyond the firm’s domestic market through exports, licensing, and foreign direct investment (Haakonsson, 2012; Makino et al., 2002). Since the marginal cost of using existing knowledge in new markets is low, scaling up through international market entry and expansion can help in recovering the high investments incurred in creating the innovation outputs, and enhance the returns from EMEs’ innovation outputs. Moreover, the exploitative approach of internationalization is low in risk as the firms essentially replicate what they already have in other markets (Makino et al., 2002). Therefore, firms find it easier to use the innovation outputs that they already have for international markets rather than develop new innovations (Lavie et al., 2010).

Existing research has mostly focused on the impact of innovation output on internationalization in the context of MNEs from developed markets, as EMEs were generally considered to be lacking in innovation. However, over the last few decades, there is growing evidence of EMEs exploiting their innovation outputs arising from knowledge exploration and transformation in the form of new products, patents, or services to achieve cost and/or differentiation based competitive advantage in foreign markets (Oura et al., 2016). Early research suggests that internationalization will occur when firms possess strategic resources and superior skills that give rise to monopolistic or competitive advantage in foreign markets (Caves, 1971; Hymer, 1976). From a resource-based view (Barney, 2001; Wernerfelt, 1984), possessing rich resources helps firms to differentiate themselves from global rivals and stimulates their internationalization. Thus, innovation outputs serve as the “resources” needed to generate and sustain competitive advantage in international markets and facilitate internationalization in EMEs. As suggested by Oura et al. (2016), a firm’s ability to compete in international markets is usually based on its ability to produce new and improved innovation output that attracts foreign customers. Accordingly, we argue that innovation output is an important factor that will encourage future internationalization by EMEs.

Hypothesis 3 (H3) Innovation output has a positive effect on internationalization in EMEs.

Institution-Based View: The Moderating Role of State Ownership (SO)

So far, we have discussed about the dynamic relationship between internationalization and innovation in EMEs by applying the organizational learning theory. However, we believe that a more complete picture can emerge when institutional factors are taken into consideration. This is because the learning process in EMEs can be influenced by external institutions such as governments (Xie & Li, 2018; Zhou et al., 2017). According to the institution-based view, governments in emerging markets play a significant role in firms’ knowledge management and innovation strategy (Lu et al., 2008). A wide range of government measures, including taxation, regulation, and direct ownership, are put in place to stimulate and nurture innovation in emerging markets (Zhou et al., 2017). With crucial influence on regulatory policies and control over scarce resources, governments can influence EMEs’ innovation process and affect their decision-making in internationalization (Cui & Jiang, 2012; Hong et al., 2015; Zhou et al., 2017). Although there has been widespread privatization over the last few decades (Cuervo-Cazurra et al., 2014), there are many examples of successful state-owned international firms from emerging markets. It is therefore important to understand the role of state ownership on the relationship between internationalization and innovation in EMEs.

Although internationalization encourages more innovation input in EMEs in the knowledge exploration phase, we expect state ownership to positively moderate this relationship. Whereas innovation input is a powerful strategic tool to acquire, nurture, and maintain new knowledge, it also entails significant risk, requires a strong commitment of resources, and takes time to produce tangible outcomes (Crossan & Apaydin, 2010; Kumar, 2009). EMEs are also subject to various short-run constraints that may limit the investments they can commit to innovation activities (Zhou et al., 2017). This is especially true for international EMEs, which have to deal with multiple challenges associated with both internationalization and innovation. Therefore, although EMEs have the incentive to increase innovation input after going international, they may not be able to invest as much as they would like to in order to explore new opportunities. In such situations, state ownership can be an institutional advantage in seeking external resources and addressing the resource constraints in international EMEs.

Governments in emerging markets play a major role in allocating key factor resources such as funding, land, and technical infrastructure (Hong et al., 2015; Zhou et al., 2017). As government-owned entities, SOEs usually enjoy privileges granted by the government and related agencies to overcome resource constraints. For example, SOEs - especially those in emerging markets - often receive subsidized credit from the government (Musacchio & Lazzarini, 2014). In the context of China in particular, the local government of Guangzhou city invested 200 billion Chinese Yuan (equivalent to 30 billion US dollar at the time) to encourage innovation specifically in SOEs in 2016. With government support, SO EMEs can be more patient investors in innovation activities and can bear the high costs of innovation input despite the high risks of international activities. That is, although internationalization and innovation input may create a “trade-off” dilemma for resource constrained non-SO EMEs, SO EMEs have sufficient resources to support the knowledge exploration activities following internationalization.

Furthermore, as international SOEs are affiliated to the government, they are under higher institutional pressure to meet the governments’ objectives (Cuervo-Cazurra et al., 2014). In recent years, governments in emerging markets are trying to build an innovative nation and catch up with the developed world. According to the global innovation index (GII) in 2021, some of the emerging economies “have managed to catch up in innovation, by complementing successful domestic innovation with international technology transfer” (GII, 2021: 21). As extended arms of government, SO EMEs follow government instructions by investing intensively in innovation and internationalization activities (Zhou et al., 2017). Therefore, we posit that SOEs in emerging markets can afford and are motivated to invest more in innovation input compared with non-SOEs while internationalizing their activities. Hence,

Hypothesis 4a (H4a) State ownership positively moderates the relationship between internationalization and innovation input in EMEs.

While most researchers assume that a positive association exists between innovation input and innovation output, the “slope” of this relationship is likely to differ across firms because of the uncertainty and complexity of the knowledge transformation process (Duran et al., 2016). That is, firms may differ in how efficiently they transform their innovation input into innovation output. Innovation output emerges after several complicated tasks are completed successfully, and whether research effort leading to innovation outcomes is punctuated by uncertainties (Parthasarthy & Hammond, 2002). We posit that state ownership is a crucial resource that facilitates the process of transforming innovation input into innovation output.

Specifically, SOEs benefit from privileged network access that fosters innovation development. As argued in the innovation literature, “networks can provide access to knowledge and resources that are not readily available via market exchanges” (Rothaermel & Hess, 2007: 901). Since SOEs operate under soft budget constraints and are supported by the government (Cuervo-Cazurra et al., 2014; Kotabe et al., 2017; Zhou et al., 2017), they are regarded as trustworthy business partners. Therefore, SOEs can build strong connections with external stakeholders, which can further facilitate their knowledge transformation process. For instance, network partners such as firms, universities, and public research institutes possess valuable experience regarding innovation development. These organizations can share their innovation experience with SOEs, and help SOEs to avoid failure in the process of converting innovation input to innovation output. Accordingly, state ownership may increase the likelihood of successful knowledge transformation in EMEs and allows more innovative outcomes to emerge.

Moreover, SOEs are better able to orchestrate diverse innovation input resources to produce more innovation output. According to Sirmon and Hitt (2003: 341), to master the challenges associated with efficient resource utilization, firms need to excel in “resource orchestration”. A firm can realize the full value of its innovation input only when the input resources are structured, bundled, and managed effectively (Sirmon et al., 2011; Asiaei et al., 2021). In emerging markets, SOEs have privileged access to resources that can facilitate integration of scattered innovation input to produce more innovation output. For example, SOEs in China, especially those owned by the central government, are encouraged to build innovation platforms to orchestrate the resources from non-SOEs, universities, national laboratories, etc. (Xinhua, 2021). By doing so, SOEs can better leverage innovation input resources and overcome the high uncertainty and risk associated with the knowledge transformation process. Thus, we propose that:

Hypothesis 4b (H4b) State ownership positively moderates the relationship between innovation input and innovation output in EMEs.

Lastly, although we hypothesise that EMEs with more innovation output are more likely to internationalize, the impact of innovation output on internationalization may differ between EMEs with and without state ownership.

Unlike SOEs that are generously supported with institutional resources and have “soft” budget constraints, non-SOEs usually have limited resources and are burdened with hard budget constraints (Zhou et al., 2017). As SOEs innovate to fulfil the expectation and mandates of the government, SOEs may have larger number of patent applications or new products. However, SOEs may have lower imperative to profit from these innovations quickly via internationalization due to resource abundance. In contrast, non-SOEs are likely to be more cautious in their innovation activities, and also try to maximise their return on innovation outputs by rapidly exploiting them in international markets.

In addition, managers of SOEs may lack the knowledge to exploit their innovation outcomes in the international market. It is suggested that managers from non-SOEs are better able to spot and take advantage of global opportunities to exploit their innovation outputs (Huang et al., 2017). On the contrary, SOE managers are often appointed by the government more for political reasons rather than their managerial expertise, and may have limited experience and knowledge about how to expand firm’s boundary overseas to create value from their innovation outputs. In China, for example, the top managers of SOEs are also Communist Party members, and their promotion/career success depend more on their political loyalty to the Party rather than their capability to commercialize innovation outputs or to identify international opportunities (Guo, 2019; Hu & Xu, 2022). According to the annual reports of listed Chinese firms in 2021, only 6.944% of top management team (TMT) members in SOEs have worked/studied overseas, while 11.389% of TMT members in non-SOEs have international experience. Therefore, although SOEs are motivated and better able to produce innovation outputs, SOE managers are less able to identify new opportunities to exploit their innovation outputs in international markets. Thus, we hypothesize that:

Hypothesis 4c (H4c) State ownership negatively moderates the relationship between innovation output and internationalization in EMEs.

Figure 1 summarises our theoretical model and hypotheses. The next section presents the data and the methods we use to test our model.

Fig. 1.

Theoretical model and hypotheses

Methods

Data

The empirical setting for this study is firms from China. After China opened its borders to international trade by adopting an open-door policy and became a member of the World Trade Organization (WTO) in 2001, the international activities of Chinese firms have grown exponentially. Like other developing-economy firms, Chinese firms are known to lag developed markets in technological and managerial domains. Through internationalization, Chinese firms address some of the disadvantages in managerial expertise and technological capabilities and compensate for their “latecomer” status in the global economy. Moreover, although China is moving towards a market-based economy, the Chinese government still plays a significant role in its economic activities (Liang et al., 2015). Therefore, China is an ideal setting to explore the relationships among internationalization, innovation, and state ownership.

The data for this study were collected from the China Stock Market & Accounting Research (CSMAR) Database. The CSMAR database provides reliable data on the China stock market and the financial statements of listed firms. Researchers in strategy and international business often use this database as a reliable source of information about firms operating in China (Ren et al., 2015; Tse et al., 2017). Our study sample is an unbalanced panel of 3449 firms covering the period from 2007 to 2018, with 17,653 firm-year observations. We focus on the period from 2007 onwards as some of the key information (e.g., R&D expenditure) is not available for earlier years. We deal with missing data in two ways. Firstly, it is compulsory for listed firms in China to report information such as total sales, total assets (from which we can calculate firm size), ownership structure, and financial performance. If the information about these variables is missing in the dataset, we treat them as missing values and drop the observation. Secondly, it is not compulsory to report information such as foreign sales or patent numbers in annual reports. If the information about these two variables is missing in the dataset, we assume the firm does not apply for any patent or does not have foreign sales in a given year, and we replace the value with 0.

The descriptive statistics for our sample are reported in Table 2. Section A shows that over 70% of the data comprises non-SOEs, with about 30% being SOEs. Section B shows that the sampled Chinese firms are clustered in high-tech industries such as manufacturing (62%) and software & IT services (20%). These industries are more dynamic and provide greater opportunities for EMEs to internationalize and innovate.

Table 2.

Ownership structure and industry distribution of sample firms

	No. Observations	Percentage (%)
Section A: Ownership structure
Non-SOEs	12,599	71.370
SOEs	5054	28.630
Total	17,653	100
Section B: Industry distribution
General manufacturing	10,919	61.854
Software and IT services	3572	20.235
Business service	787	4.458
Civil Engineering Construction	449	2.543
Wholesale	400	2.266
Mining	396	2.243
Production and supply of electric power and thermal power	266	1.507
Professional technological service	266	1.507
Agriculture	226	1.280
Transportation agency	201	1.139
Real estate	171	0.969
Total	17,653	100

Measures

Main Variables

Innovation Input (RDI) The traditional, and still the most popular, input indicator of innovation is expenditure on research and development (R&D). On the input side of innovation, a firm’s investment in R&D activities reflects the resources committed to innovation activities (Ren et al., 2015; Zhou et al., 2017). Following previous research with similar topics (e.g., Chittoor et al., 2015; Freixanet & Churakova, 2018), we use a firm’s yearly R&D intensity to measure R&D input, calculated as percentage of R&D expenses to total sales (Piperopoulos et al., 2018). In the robustness test using propensity score matching (PSM) approach, we measure innovation input as a dummy variable (RDI dummy) to estimate the impact of internationalization on innovation input. RDI dummy takes the value of 0 if a firm does not have any R&D expenditure in a given year, and 1 otherwise.

Innovation Output (Patent) R&D expenditure is a useful indicator of input into the innovation process, but it tells us nothing about the outcome of innovation activities. For our study, we use patent applications as the measure of innovation output. Extensive research on the economics of technological change has shown that patenting activity reflects the quality and extent of firm innovation, and the use of patent applications as a measure of innovation output is widely accepted (Salomon & Jin, 2010). Importantly for this study, information about patent applications is available for the listed Chinese firms. We use patent application rather than granted patent as the measure of innovation output, as granted patent cannot capture the effectiveness of internationalization on innovation in a given year. For example, it may take 3–5 years for a patent to be granted in China, and that is a large time gap. A patent granted in 2020 could have been under review since 2015, and cannot be regarded as the outcome of innovation input in 2020. We measure innovation output as the logarithm of 1 plus the count of the number of patent applications, for firm i in year t. We use a logarithmic form for patent numbers since the number of yearly patent application ranges widely. As suggested by Hair et al. (2013), logarithmic form is appropriate for scattered data. The particular year t is assigned as the application date for each granted patent. For example, a patent filed in 2016 but granted in 2018 is treated as a 2016 patent. Likewise, in the PSM estimation, we measure innovation output using a dummy variable (Patent dummy) to estimate the impact of innovation input on innovation output. Patent dummy takes the value of 0 if a firm does not apply for any patent in a given year, and 1 otherwise.

Internationalization (DOI, Status) We measure internationalization in two ways. In the baseline analysis, internationalization is measured as the degree of internationalization (DOI), which is the percentage of foreign sales to total sales in a given year. Many studies (e.g., Ren et al., 2015; Shearmur et al., 2015) have implicitly or explicitly suggested that sales generated outside the home country can be referred to as internationalization. Hence, the percentage of foreign sales to total sales is the most common measure of internationalization. In our robustness tests, we also use a dummy variable to indicate the status of internationalization in a given year (Status), which is coded as 1 if a firm has any foreign sales in a given year, and 0 otherwise.

State Ownership (SOE) A firm with full, majority, or largest minority state ownership is regarded as an SOE. We code state ownership as 1 if the firm is state-owned in a given year and 0 otherwise. Notably, we measure state ownership as a dummy variable rather than the percentage of ownership. The dummy variable is a better match with the objective of this study wherein we examine how government as the ultimate controlling shareholder of a firm moderates the internationalization-innovation relationship.

Control Variables

Firm Size Following Schumpeter (1942), researchers examine whether large or small organizations are more likely to innovate. Research on innovation has extensively emphasized that innovation is a resource-intensive activity. As larger firms tend to have more resources, they are more likely to conduct innovation activities. Likewise, larger firms are better able to overcome the entry barriers for internationalization. We measure firm size with total assets (in Chinese Yuan). Following previous studies (e.g., Piperopoulos et al., 2018; Xie & Li, 2018), the natural logarithm transformation is used as a measure of firm size to control for possible non-linearity.

Firm age With rich complementary assets accumulated over years, older firms are more likely to benefit from R&D investment and more motivated to internationalize (Liu, 2017). However, older firms are also more likely to suffer from organizational inertia, which makes it difficult for them to adapt their resources, processes, and strategies to new markets, technologies, or products (Zahra et al., 2000). Thus, we control for firm age as measured by the number of years elapsed since the establishment of the firm (Kotabe et al., 2017).

Marketing Capability Strong marketing capability can help firms utilize limited resources more effectively and even enhance customer cooperation in bringing together complementary assets to foster innovation (Ren et al., 2015). Firms with strong marketing capability are also better at exploiting new market knowledge and capturing the opportunity to internationalize (Ren et al., 2015). We therefore include marketing capability as one of the control variables, measured by the ratio of marketing expenses to sales revenue.

Leverage There is a clear negative relationship between leverage and R&D spending (Lerner et al., 2011). However, the direction of causality is unclear. It is difficult to determine whether high debt leads to R&D reductions or if struggling firms simply have high debt and low spending on innovation. Therefore, we control for leverage in the research model and measure it with the ratio of total assets divided by total assets minus total liabilities.

Organizational Slack Organizational slack is also likely to influence a firm’s innovation and internationalization in different ways. On the one hand, it enables firms to invest in long-term R&D projects, while on the other hand, it may diminish the incentive to innovate (Nohria & Gulati, 1996). Previous research has found mixed results about the impact of organizational slack on firms’ innovation (Dan & Geiger, 2015; Nohria & Gulati, 1996). Firms with more slack resources are better able to internationalize (Carneiro et al., 2018). In this study, we use current ratio as the measure of slack for each firm, with higher ratio indicating more organizational slack.

Government Grant Firms may invest less in innovation and internationalization activities than would be “socially desirable” (Zhou et al., 2017). In such situations, governments often provide significant subsidies to facilitate firm activities and promote national economic growth and development. Public subsidy helps in mitigating high, uninsurable risk of decisions to innovate or internationalize, and therefore is an important supplementary resource for firms to pursue innovation and internationalization activities. For example, Le and Jaffe (2017) suggest that government subsidies have significantly increased innovation inputs. Therefore, we include government grant as a control variable, and measure it with the natural logarithm of the government’s yearly subsidy given to a firm for research and technological development activities.

Financial Performance Financial performance is a crucial determinant of innovation (Crossan & Apaydin, 2010). It has also been argued that firms with better performance are more likely to invest in internationalization, as they can easily cover the sunk costs of international activities. Following previous studies (Chang et al., 2019), we use return on assets as the measure of financial performance.

Industry Type Industry is an important component of the context, and which may help or hinder a firm's investment decisions in innovative activities. Firms from knowledge- and technology-intensive industries invest more in innovation activities than those in other sectors (Zhou et al., 2017). Furthermore, some industries within a country may appear intensely competitive, which forces more firms to innovate. In addition, the Chinese government's “Go-Global” strategy has prioritized international activities for selected industries. Therefore, we control for industry effect, with industry coded as a categorical dummy variable such as A (Agriculture) and B (Mining) according to the classification of the industry in China. Table 3 contains a summary of all variables used in our study.

Table 3.

Summary of variables

Variable name	Measurement
Main variables
Innovation input (RDI)	R&D intensity, measured with the ratio of R&D expenditure to total sales
Innovation output (Patent)	The natural logarithm of 1 plus the count number of yearly patent applications in China
Internationalization (DOI, Status)	Internationalization is measured with (1) the degree of internationalization (DOI), which is the ratio of foreign sales to total sales, and (2) a dummy variable (Status), coded as 1 if the firm has foreign sales, and 0 otherwise
Moderating variable
State ownership (SOE)	A dummy variable with the value of 1 if the firm is state owned, and 0 if the firm is non-state owned
Control variables
Firm size	The natural logarithm of the firm’s total assets
Firm age	The number of years elapsed since the establishment of the firm
Marketing capability	The ratio of marketing expenses to sales revenue
Leverage	Financial leverage reported in annual report, which is the ratio of total assets divided by total assets minus total liabilities
Organizational slack	Current ratio reported in annual report, which is the ratio of current assets to current liabilities
Government grant	The natural logarithm of the government’s yearly subsidy paid to the firm for research and technological development
Financial performance	Return on assets (ROA)
Industry type	A categorical/dummy variable coded as A, B, etc

Analytical Methods

We use a system-of-equations approach to estimate the relationship between internationalization and innovation, as well as the moderating role of state ownership on the internationalization-innovation link. The regression equations for the main effects models are as follows:

$${\mathit{RDI}}_{i,t+1}={\alpha }_1+{\beta }_{1,1}\ast {\mathit{DOI}}_{i,t}+{\sum }_{m=1}^m{\gamma }_{1,m}\ast {\mathit{Control}}_{i,t}+Year+{\mathit{Ind}}_i+{\epsilon }_{1,i,t}$$ 1

$${\mathit{Patent}}_{i,t+2}={\alpha }_2+{{\beta }_{2,1}\ast RDI}_{i,t+1}+{\sum }_{n=1}^m{\gamma }_{2,m}\ast {\mathit{Control}}_{i,t+1}+Year+{\mathit{Ind}}_i+{\epsilon }_{2,i,t+1}$$ 2

$${\mathit{DOI}}_{i,t+3}={\alpha }_3+{\beta }_{3,1}\ast {\mathit{Patent}}_{i,t+2}+{\sum }_{m=1}^m{\gamma }_{3,m}\ast {\mathit{Control}}_{i,t+2}+Year+{\mathit{Ind}}_i+{\epsilon }_{3,i,t+2}$$ 3

The regression equations to test the moderating effect of state ownership are as follows:

$${\mathit{RDI}}_{i,t+1}={\alpha }_4+{\beta }_{4,1}\ast {\mathit{SOE}}_{i,t}\ast {\mathit{DOI}}_{i,t}+{\beta }_{4,2}\ast {\mathit{DOI}}_{i,t}+{\beta }_{4,3}\ast {\mathit{SOE}}_{i,t}+{\sum }_{m=1}^m{\gamma }_{4,m}\ast {\mathit{Control}}_{i,t}+Year+{\mathit{Ind}}_i+{\epsilon }_{4,i,t}$$ 4

$${\mathit{Patent}}_{i,t+2}={\alpha }_5+{\beta }_{5,1}\ast {\mathit{SOE}}_{i,t+1}\ast {\mathit{RDI}}_{i,t+1}+{\beta }_{5,2}{\ast RDI}_{i,t+1}+{\beta }_{5,3}\ast {\mathit{SOE}}_{i,t+1}+{\sum }_{m=1}^m{\gamma }_{5,m}\ast {\mathit{Control}}_{i,t+1}+Year+{\mathit{Ind}}_i+{\epsilon }_{5,i,t+1}$$ 5

$${\mathit{DOI}}_{i,t+3}={\alpha }_6+{\beta }_{6,1}\ast {\mathit{SOE}}_{i,t+2}\ast {\mathit{Patent}}_{i,t+2}+{\beta }_{6,2}\ast {\mathit{Patent}}_{i,t+2}+{\beta }_{6,3}\ast {\mathit{SOE}}_{i,t+2}+{\sum }_{m=1}^m{\gamma }_{6,m}\ast {\mathit{Control}}_{i,t+2}+Year+{\mathit{Ind}}_i+{\epsilon }_{6,i,t+2}$$ 6

where $\mathrm{i}$ stands for firm i, t stands for year t, ${\mathrm{Control}}_{\mathrm{i},\mathrm{t}}$ includes control variables that are potential determinants of innovation and internationalization, such as firm size, firm age, marketing capability, leverage, organizational slack, government grant, and financial performance. Following previous studies with panel data (e.g., Xie & Li., 2018), generalized least squares (GLS) technique is used to estimate the models above. Radom-effects regression is used to ensure that the industry dummies would not be omitted from the regressions.

Potential Endogeneity Problem

Endogeneity arises when an explanatory variable is related to the error term (Wooldridge, 2016). As the error term cannot be observed, there is no direct way to check whether an explanatory variable is correlated with the error term statistically (Angrist & Pischke, 2008). As a result, it is almost impossible to statistically ensure that an endogeneity problem is completely addressed. Therefore, we have to first identify the potential causes of endogeneity, and then take proper actions to reduce the negative impact of endogeneity (Cameron & Trivedi, 2010). Generally, endogeneity may arise due to reverse causality, selection bias, and omitted variables (Greene, 2003; Wooldridge, 2016). For example, whereas internationalization may encourage more innovation input, firms investing in innovation input may also be more motivated to spread their innovation input costs through internationalization. To address the potential reverse-causality concern, we lag all explanatory variables for one period to ensure that the direction of causality runs from the independent variable to the dependent variable. Secondly, we apply the propensity score matching approach that allows us to correct for sample selection bias due to observable differences between the treatment and control group. Thirdly, we mitigate issue of omitted variables by including a range of control variables while estimating our theoretical model (Wooldridge, 2016).

Propensity Score Matching

To find evidence of the causality from internationalization to innovation input (phase 1: knowledge exploration model), an ideal experiment would be to compare the innovation input of a firm after it starts internationalizing with the innovation input of the same firm that never internationalizes. However, it is not possible to observe the counterfactual outcome. Thus, there is no direct counterpart of the treatment effect computed as the mean difference between the outcomes of the treated and untreated groups. Matching estimators provide an interesting solution to this problem (Abadie & Imbens, 2016). In short, matching estimators impute the missing potential outcomes of treated individuals (i.e., what would have happened had the treatment occurred) by using the outcomes of observations with similar values of “relevant” pre-treatment variables or covariates, but which are not exposed to treatment. The average outcome for the untreated matched group identifies the mean counterfactual outcome for the treated group in the absence of the treatment (Cameron & Trivedi, 2010).

The matching method has been used to address potential endogeneity problems in international business studies. For example, Golovko and Valentini (2011) used the matching estimator to study the relationship between innovation and exports. A major strength of the matching approach is that it allows researchers to clearly identify the non-random component of the treatment effect and to identify the suitable counterfactual firms. Matching approach includes pairing treated and untreated observations that are similar in terms of their observable characteristics (Abadie & Imbens, 2016). When the relevant differences between any two observations are captured in the observable (pretreatment) covariates, which occurs when outcomes are independent of assignment to treatment conditional on pretreatment covariates, matching approach can provide an unbiased estimate of the treatment effect (Abadie & Imbens, 2016). To perform propensity score matching, we first estimate a probit model by using international EMEs and non-international EMEs in the sample, with control variables of firm size, firm age, marketing capability, leverage, organizational slack, government grant, and financial performance. We then use the propensity score to perform one-to-one matching. Such a matched sample controls for differences between international EMEs and non-international EMEs other than international involvement and ensures comparability between these two groups of firms. We then re-run the analysis to check whether the results are consistent with the baseline results. We use the same method to estimate our three models, namely the knowledge exploration model of internationalization to innovation input, the knowledge transformation model of innovation input to innovation output, and the knowledge exploitation model of innovation output to internationalization.

Results

Main Results

The descriptive statistics and correlations are reported in Table 4. As shown in the table, the R&D intensity (RDI) in EMEs is 0.050 on average. The number of patent applications is also considerable, with yearly average number of 27 (before taking logarithms). Moreover, the average degree of internationalization (DOI) is 0.147, ranging from 0 to 1. The average firm age in the sample is 16.426, which indicates that the sampled firms are relatively experienced. Internationalization (DOI), innovation input (RDI), and innovation output (Patent) are found to be positively and significantly related with each other, which suggests possible relationships among the key variables. A review of the correlation coefficients among the independent variables shows that multicollinearity is not a concern, with low variance inflation factor (VIF) ranging from 1.07 to 2.21 (Hair et al., 2013).

Table 4.

Descriptive statistics and correlation matrix

Variable	Mean	S.D	1	2	3	4	5	6	7	8	9	10	11
1. DOI	0.147	0.218	1.000
2. RDI	0.050	0.060	0.061	1.000
3. Patent	1.566	1.624	0.091	0.085	1.000
4. Status	0.674	0.469	0.468	0.017	0.179	1.000
5. Firm size	21.710	1.156	− 0.067	− 0.205	0.133	0.026	1.000
6. Firm age	16.426	5.866	− 0.046	− 0.122	− 0.173	− 0.045	0.200	1.000
7. Marketing capability	0.085	0.098	− 0.151	0.224	− 0.002	− 0.143	− 0.147	− 0.027	1.000
8. Leverage	0.389	0.203	− 0.046	− 0.324	− 0.011	0.025	0.473	0.199	− 0.303	1.000
9. Organizational slack	2.905	3.317	0.018	0.322	0.011	− 0.050	− 0.316	− 0.183	0.219	− 0.625	1.000
10. SOE	0.286	0.452	− 0.114	− 0.192	0.003	− 0.055	0.369	0.201	− 0.171	0.327	− 0.204	1.000
11. Grant	15.860	2.393	− 0.019	0.010	0.198	0.063	0.329	0.001	0.007	0.157	− 0.106	0.109	1.000
12. Financial performance	0.043	0.064	− 0.005	0.157	0.084	− 0.030	− 0.071	− 0.113	0.220	− 0.351	0.209	− 0.118	0.024

Correlations with p < 0.050 are in bold

Table 5 presents the results for our three models: knowledge exploration (model 1), knowledge transformation (model 2), and knowledge exploitation (model 3). In our knowledge exploration model, the degree of internationalization (DOI) has a positive and statistically significant effect on innovation input (R&D intensity, RDI) (${\beta }_{{\mathit{DOI}}_t}=0.006,p=0.041$). This result supports hypothesis 1 and shows that internationalization encourages EMEs to raise their R&D spending to pursue innovation activities. In the knowledge transformation model 2, innovation input (R&D intensity, RDI) has a positive and statistically significant effect on innovation output (patent application) (${\beta }_{{\mathit{RDI}}_{t+1}}=0.878,p=0.000$), indicating that innovation input contributes to more innovation output in general. Thus, hypothesis 2 is supported. Finally, in our knowledge exploitation model 3, innovation output (patent application) has a positive and significant effect on internationalization in EMEs (${\beta }_{{\mathit{Patent}}_{t+2}}=0.004,p=0.000$). This result supports hypothesis 3 that EMEs with more innovation output resources are more likely to internationalize. In addition, the control variables are by and large in the expected direction in the regression models. For example, organizational slack, government grant, and financial performance are found to have a positive effect on innovation input.

Table 5.

Random-effects GLS estimates for the knowledge exploration, transformation, and exploitation models

Variable	$\mathrm{Model}1:{\mathrm{RDI}}_{\mathrm{t}+1}$			$\mathrm{Model}2:{\mathrm{Patent}}_{\mathrm{t}+2}$			$\mathrm{Model}3:{\mathrm{DOI}}_{\mathrm{t}+3}$
	Est	S.E	p	Est	S.E	p	Est	S.E	p
${\mathrm{H}1:\mathrm{DOI}}_{\mathrm{t}}$	0.006	0.003	0.041
${\mathrm{H}2:\mathrm{RDI}}_{\mathrm{t}+1}$				0.878	0.210	0.000
${\mathrm{H}3:\mathrm{Patent}}_{\mathrm{t}+2}$							0.004	0.001	0.000
Firm size	− 0.002	0.001	0.000	0.218	0.016	0.000	− 0.006	0.002	0.001
Firm age	− 0.001	0.000	0.000	− 0.025	0.004	0.000	− 0.001	0.001	0.043
Marketing capability	0.075	0.007	0.000	0.011	0.170	0.947	− 0.154	0.021	0.000
Leverage	− 0.016	0.003	0.000	− 0.096	0.086	0.268	− 0.002	0.009	0.841
Organizational slack	0.002	0.000	0.000	− 0.006	0.004	0.113	0.000	0.000	0.313
Grant	0.001	0.000	0.004	0.007	0.005	0.167	0.000	0.000	0.944
Financial performance	0.059	0.008	0.000	1.517	0.219	0.000	− 0.057	0.021	0.007
Intercept	0.058	0.015	0.000	− 3.851	0.365	0.000	0.291	0.054	0.000
Year dummies	Included			Included			Included
Industry dummies	Included			Included			Included
N	14,482			14,482			14,482
Wald Chi-squared	1859.24			9390.55			218.42
R2	0.271			0.312			0.050

The control variables are in year t, t + 1, t + 2 for model 1, model 2, and model 3, respectively. All tests are two-tailed

Table 6 shows the results of the moderating effect of state ownership on the internationalization-innovation links in EMEs. As shown in the table, for model 4, the interaction term between internationalization and the SOE dummy variable is not statistically significant (${\beta }_{{\mathit{DOI}}_t\times {\mathit{SOE}}_t}=-0.004,p=0.569$). Thus, hypothesis 4a is not supported. This does not mean that SOEs do not increase innovation input after going international. Rather, it simply suggests that there is no significant difference between SOEs and non-SOEs in the effect of internationalization on innovation inputs. In fact, additional analysis with the SO EME subsample shows that internationalization has a significant and positive relationship with innovation input (${\beta }_{{\mathit{DOI}}_t}=0.010,p=0.061$) in SOEs. Next, for model 5, the interaction term between innovation input and SOE dummy is positive and significant (${\beta }_{{\mathit{RDI}}_{t+1}\times {\mathit{SOE}}_{t+1}}=2.429,p=0.000$). Therefore, hypothesis 4b is supported, namely, SOEs are more efficient in transforming innovation input into innovation output. Finally, in model 6, the interaction term between innovation output and SOE is negative and significant (${\beta }_{{\mathit{Patent}}_{t+2}\times {\mathit{SOE}}_{t+2}}=-0.004,p=0.020$). Therefore, Hypothesis 4c is supported, that is, state ownership negatively moderates the relationship between innovation output and internationalization. Further analysis shows that whereas innovation output does not have a significant impact on internationalization in SO EMEs (${\beta }_{{\mathit{Patent}}_{t+2}}=0.002,p=0.133$), it has a significant and positive effect in non-SO EMEs (${\beta }_{{\mathit{Patent}}_{t+2}}=0.005,p=0.000$). Therefore, the overall positive relationship between innovation output and internationalization in EMEs (in Hypothesis 3) is mainly driven by non-SO EMEs.

Table 6.

Random-effects GLS estimates for the moderating effect of state ownership

Variable	$\mathrm{Model}4:{\mathrm{RDI}}_{\mathrm{t}+1}$			$\mathrm{Model}5:{\mathrm{Patent}}_{\mathrm{t}+2}$			$\mathrm{Model}6:{\mathrm{DOI}}_{\mathrm{t}+3}$
	Est	S.E	p	Est	S.E	p	Est	S.E	p
$\mathrm{H}4\mathrm{a}:{\mathrm{DOI}}_{\mathrm{t}}\times {\mathrm{SOE}}_{\mathrm{t}}$	− 0.004	0.007	0.569
$\mathrm{H}4\mathrm{b}:{\mathrm{RDI}}_{\mathrm{t}+1}\times {\mathrm{SOE}}_{\mathrm{t}+1}$				2.429	0.531	0.000
$\mathrm{H}4\mathrm{c}:{\mathrm{Patent}}_{\mathrm{t}+2}\times {\mathrm{SOE}}_{\mathrm{t}+2}$							− 0.004	0.002	0.020
${\mathrm{DOI}}_{\mathrm{t}}$	0.006	0.003	0.056
${\mathrm{RDI}}_{\mathrm{t}+1}$				0.539	0.223	0.016
${\mathrm{Patent}}_{\mathrm{t}+2}$							0.005	0.001	0.000
SOE	− 0.006	0.002	0.001	0.008	0.044	0.860	− 0.018	0.007	0.013
Firm size	− 0.002	0.001	0.001	0.221	0.016	0.000	− 0.005	0.002	0.006
Firm age	− 0.001	0.000	0.000	− 0.024	0.004	0.000	− 0.001	0.001	0.166
Marketing capability	0.074	0.007	0.000	0.018	0.169	0.915	− 0.157	0.021	0.000
Leverage	− 0.015	0.003	0.000	− 0.089	0.086	0.302	0.001	0.009	0.913
Organizational slack	0.002	0.000	0.000	− 0.006	0.004	0.125	0.000	0.000	0.292
Grant	0.001	0.000	0.004	0.008	0.005	0.128	0.000	0.000	0.885
Financial performance	0.057	0.008	0.000	1.496	0.219	0.000	− 0.059	0.021	0.005
Intercept	0.051	0.015	0.001	− 3.872	0.369	0.000	0.270	0.054	0.000
Year dummies	Included			Included			Included
Industry dummies	Included			Included			Included
N	14,482			14,482			14,482
Wald Chi-squared	1870.04			9409.26			241.39
R2	0.270			0.315			0.058

The control variables are in year t, t + 1, t + 2 for model 1, model 2, and model 3, respectively. All tests are two-tailed

Figure 2 illustrates the moderating effect of state ownership on EME internationalization-innovation linkages. As shown in Fig. 2a, internationalization has a positive relationship with innovation input for both SO EMEs and non-SO EMEs. However, the slope of these two lines do not show significant difference as indicated in the results for model 4. Figure 2b shows that innovation input is positively related with innovation output in both SOEs and non- SOEs, but the slope of the regression line is steeper in SOEs than in non-SOEs, which suggests that the innovation input–output relationship is much stronger in SOEs than in non-SOEs. Figure 2c illustrates the moderating effect of state ownership on the innovation output-internationalization path. The figure shows that the positive relationship between innovation output and internationalization is mainly driven by non-SOEs. As also suggested by the results from model 6, the moderating effects of state ownership on the innovation output-internationalization relationship is negative and significant.

Fig. 2.

Interaction plots of the moderating effect of state ownership in the knowledge a exploration, b transformation, and c exploitation models

Sensitivity and Robustness Tests

To deal with potential reverse causality, we estimated our models by lagging all explanatory variables including control variables by one period to account for the delay between internationalization and innovation activities. In addition, we use a propensity score matching (PSM) approach to check whether the baseline results continue to hold, and to provide a more rigorous empirical support for the observed relationship. As shown in Table 7, internationalization (status) has a significant and positive effect on innovation input. The results are consistent across different matching methods such as nearest neighbor method (${\beta }_{{\mathit{Status}}_t}=0.004,p=0.000$) and radius method (${\beta }_{{\mathit{Status}}_t}=0.002,p=0.001$), and they are qualitatively unchanged and corroborate hypothesis 1 (Cameron & Trivedi, 2010). Table 7 further shows the PSM results of knowledge transformation model (model 2). As shown in the table, the decision to invest in innovation input (R&D dummy) has a significant positive effect on innovation output (patent), which further support hypothesis 2. Similarly, the results from Table 7 show that patent has a positive and significant effect on internationalization in EMEs. The results are consistent across regression models using nearest neighbour approach (${\beta }_{{\mathit{Patentdummy}}_{t+1}}=0.035,p=0.000$) and radius approach (${\beta }_{{\mathit{Patentdummy}}_{t+1}}=0.039,p=0.000$). Therefore, hypothesis 3 is further supported.

Table 7.

PSM estimates for the knowledge exploration, transformation, and exploitation models

Variable	$\mathrm{Model}1:{\mathrm{RDI}}_{\mathrm{t}+1}$						${\mathrm{Model}2:\mathrm{Patent}}_{\mathrm{t}+2}$						${\mathrm{Model}3:\mathrm{DOI}}_{\mathrm{t}+3}$
	Nearest neighbor			Radius			Nearest neighbor			Radius			Nearest neighbor			Radius
	Est	S.E	p	Est	S.E	p	Est	S.E	p	Est	S.E	p	Est	S.E	p	Est	S.E	p
${\mathrm{Status}}_{\mathrm{t}}$	0.004	0.001	0.000	0.002	0.001	0.001
${\mathrm{RDI}\mathrm{dummy}}_{\mathrm{t}+1}$							0.051	0.029	0.081	0.138	0.033	0.000
${\mathrm{Patent}\mathrm{dummy}}_{\mathrm{t}+2}$													0.035	0.004	0.000	0.039	0.004	0.000
Firm size	− 0.004	0.000	0.000	− 0.005	0.000	0.000	0.506	0.010	0.000	0.544	0.012	0.000	− 0.007	0.002	0.000	− 0.007	0.002	0.000
Firm age	− 0.001	0.000	0.000	− 0.001	0.000	0.000	− 0.016	0.002	0.000	− 0.010	0.003	0.000	− 0.001	0.000	0.001	− 0.001	0.000	0.116
Marketing capability	0.086	0.005	0.000	0.101	0.006	0.000	− 0.494	0.116	0.000	− 0.346	0.136	0.011	− 0.471	0.018	0.000	− 0.458	0.019	0.000
Leverage	− 0.014	0.003	0.000	− 0.016	0.003	0.000	− 0.211	0.076	0.006	− 0.286	0.091	0.002	− 0.032	0.013	0.011	− 0.041	0.013	0.002
Organizational slack	0.004	0.000	0.000	0.004	0.000	0.000	− 0.003	0.004	0.430	− 0.003	0.005	0.612	0.001	0.001	0.051	0.000	0.001	0.635
Grant	0.002	0.000	0.000	0.002	0.000	0.000	0.108	0.005	0.000	0.104	0.006	0.000	− 0.001	0.001	0.444	− 0.001	0.001	0.459
Financial performance	0.068	0.008	0.000	0.064	0.008	0.000	3.750	0.207	0.000	3.827	0.238	0.000	− 0.059	0.035	0.091	− 0.046	0.037	0.220
Intercept	0.067	0.009	0.000	0.068	0.009	0.000	− 11.801	0.220	0.000	− 12.787	0.254	0.000	0.304	0.041	0.000	0.309	0.044	0.000
Year dummies	Included			Included			Included			Included			Included			Included
Industry dummies	Included			Included			Included			Included			Included			Included
N	18,536			17,436			21,457			16,261			16,486			14,783
F	264.32			244.87			500.01			372.56			40.79			35.98
R2	0.299			0.296			0.411			0.407			0.068			0.066

The control variables are in year t, t + 1, t + 2 for model 1, model 2, and model 3, respectively. All tests are two-tailed

As the PSM procedure tries to find a comparison group that is statistically equivalent to the treated group, we need to check whether the sample of treated and untreated firms is balanced. In a randomized experiment, the treated and untreated group should show insignificant difference in matching variables (Cameron & Trivedi, 2010). The quality of matching is shown in the appendix in Tables 8, 9, 10. Overall, the matching is of good quality.

Table 8.

Quality of matching by nearest neighbor method and radius method (knowledge exploration model)

Variable	Nearest neighbor		Radius
	Mean		Mean
	Treated	Control	Treated	Control
Firm size	21.728	21.779	21.718	21.781
Firm age	16.175	16.242	16.198	16.303
Marketing capability	0.075	0.075	0.068	0.067
Leverage	0.393	0.399	0.397	0.404
Organizational slack	2.789	2.702	2.692	2.628
Grant	15.965	15.974	16.023	16.008
Financial performance	0.041	0.041	0.041	0.040

Table 9.

Quality of matching by nearest neighbor method and caliper method (knowledge transformation model)

Variable	Nearest neighbor		Radius
	Mean		Mean
	Treated	Control	Treated	Control
Firm size	21.709	21.910	21.719	21.902
Firm age	16.377	17.003	16.082	16.553
Marketing capability	0.085	0.078	0.083	0.085
Leverage	0.388	0.437	0.398	0.429
Organizational slack	2.914	2.537	2.617	2.247
Grant	15.863	16.105	15.878	16.138
Financial performance	0.043	0.043	0.042	0.043

Table 10.

Quality of matching by nearest neighbor method and caliper method (knowledge exploitation model)

Variable	Nearest neighbor		Radius
	Mean		Mean
	Treated	Control	Treated	Control
RDI	0.054	0.055	0.051	0.050
Firm size	21.661	21.698	21.680	21.704
Firm age	15.233	15.136	15.627	15.507
Marketing capability	0.086	0.085	0.086	0.084
Leverage	0.369	0.375	0.378	0.380
Organizational slack	3.152	3.033	2.934	2.862
Grant	16.098	16.060	16.151	16.086
Financial performance	0.047	0.048	0.046	0.046

In addition to addressing the endogeneity problem by using lagged terms and the PSM approach, we also used different measures of key variables for robustness test (Table 11). For example, we use invention patents only (instead of all types of patents) as the measure of innovation output, and internationalization status (instead of DOI) as an alternative measure of internationalization, and the results remain consistent.

Table 11.

Random-effects GLS estimates by using alternative measurements

Variable	$\mathrm{Model}1:{\mathrm{RDI}}_{\mathrm{t}+1}$			$\mathrm{Model}2:{\mathrm{Invention}}_{\mathrm{t}+2}$			$\mathrm{Model}3:{\mathrm{Status}}_{\mathrm{t}+3}$
	Est	S.E	p	Est	S.E	p	Est	S.E	p
${\mathrm{H}1:\mathrm{Status}}_{\mathrm{t}}$	0.004	0.001	0.000
${\mathrm{H}2:\mathrm{RDI}}_{\mathrm{t}+1}$				1.085	0.180	0.000
${\mathrm{H}3:\mathrm{Invention}}_{\mathrm{t}+2}$							0.012	0.002	0.000
Firm size	− 0.003	0.001	0.000	0.212	0.013	0.000	0.027	0.005	0.000
Firm age	− 0.001	0.000	0.000	− 0.016	0.003	0.000	− 0.006	0.001	0.000
Marketing capability	0.076	0.007	0.000	− 0.007	0.143	0.958	− 0.428	0.050	0.000
Leverage	− 0.016	0.003	0.000	0.071	0.074	0.337	0.018	0.021	0.391
Organizational slack	0.002	0.000	0.000	0.001	0.003	0.759	− 0.004	0.001	0.000
Grant	0.001	0.000	0.006	0.013	0.004	0.002	0.003	0.001	0.014
Financial performance	0.059	0.008	0.000	1.296	0.189	0.000	− 0.045	0.051	0.382
Intercept	0.061	0.015	0.000	− 4.319	0.307	0.000	− 0.102	0.118	0.387
Year dummies	Included			Included			Included
Industry dummies	Included			Included			Included
N	14,482			14,482			14,482
Wald Chi-squared	1869.86			6403.69			634.44
R2	0.271			0.267			0.095

The control variables are in year t, t + 1, t + 2 for model 1, model 2, and model 3, respectively. All tests are two-tailed

Discussion and Conclusion

Internationalization and innovation provide EMEs with unique opportunities to catch up with developed market enterprises, as well as overcome their latecomer liability (Buckley & Hashai, 2014; Xie & Li, 2018). As both strategies are built upon existing and new resources and capabilities, growth along one dimension is expected to be systematically related to growth along the other one (Chiva et al., 2014; Kumar, 2009). While extant research has highlighted the potential association between internationalization and innovation, our study extends this research by examining the dynamic, two-way relationship between internationalization and innovation in EMEs while broadening our understanding of the role of state ownership on this phenomenon.

Using a panel dataset of Chinese listed firms from 2007 to 2018, we find that internationalization contributes to innovation input in EMEs. Innovation input then leads to more innovation output, which further contributes to a higher level of internationalization. State ownership positively moderates the relationship between innovation input and innovation output, but negatively moderates the link between innovation output and internationalization. We also conduct robustness tests using the PSM method, and the results remain consistent.

Theoretical Contributions

Our study makes a number of key contributions to the IB literature. First, we develop a dynamic internationalization-innovation model for EMEs. Drawing upon organizational learning theory, our study captures all three phases of knowledge, namely, exploration, transformation, and exploitation within a single integrated framework. Compared with MNEs from developed markets, international EMEs lack innovation resources at the early stage of internationalization. EMEs therefore embark on internationalization to drive their knowledge exploration activities, resulting in growing innovation inputs driven by internationalization. Next, the innovation inputs are transformed into innovation outputs by undertaking large-scale basic and applied R&D activities. These innovation outputs are next exploited by EMEs to accelerate their internationalization. By providing empirical support for our proposed theoretical model, our study makes a unique contribution to research about internationalization and innovation by revealing the dynamic nature of the relationship.

Second, the study incorporates knowledge transformation as the bridge between the knowledge exploration and knowledge exploitation phases of organizational learning. Exploration and exploitation have long been regarded as two complementary approaches for organizational learning (Lavie et al., 2010; March, 1991). However, researchers have suggested that it is important to understand how firms make the transition between exploration and exploitation (Gobbo & Olsson, 2010; Lane et al., 2006). Although knowledge transformation is implicitly recognized, it has rarely been tested formally with empirical data. Incorporating the knowledge transformation link in our model enables us to demonstrate how EMEs iterate between the two phases of knowledge exploration and knowledge exploitation and create a dynamic upward spiral of internationalization and innovation. Our study shows that exploration, transformation, and exploitation complement one another to support the positive dynamics between internationalization and innovation. The transformation activity serves as an intermediate step where firms deploy the R&D resources to translate the knowledge that they explored from international sources into something that they are able to exploit in the next phase of internationalization. As suggested by Gobbo and Olsson (2010), the transformation movement between exploration and exploitation is clockwise, and this is empirically supported by the results of our study. By doing so, we contribute to organizational learning theory by providing theoretical and empirical support for the mediating role of knowledge transformation in the knowledge exploration to exploitation process.

Third, given the significance of governments in emerging markets, we examine the moderating role of state ownership in influencing the trajectory of firms on the internationalization-innovation path. Our results show that despite the large amount of resources available for SO EMEs, SO EMEs and non-SO EMEs are relatively equally effective in establishing the internationalization to innovation input link. That is, SO EMEs and non-SO EMEs do not show significant difference in the relationship between internationalization and innovation input. A possible reason for this is that although SOEs in emerging markets receive more financial support from the government, they also aim to achieve other strategic goals such as social responsibilities (Cuervo-Cazurra et al., 2014). As a result, their financial resources get diverted towards addressing social issues such as poverty alleviation, quality education, climate change, etc. Moreover, non-SOEs have stronger entrepreneurial incentives and management compared to SOEs, which may compensate for the resource disadvantages of non-SOEs in innovation input/investment following internationalization. Taken together, the net result is no significant difference in the strength of relationship between internationalization and innovation input between SOEs and non-SOEs in emerging markets such as China. This is an important finding that has significant managerial and public policy implications, as discussed in the next section.

This study further finds that state ownership serves as a crucial institutional advantage in the knowledge transformation phase. Specifically, we find that SO EMEs are more efficient in the process of transforming innovation input into innovation output than non-SO EMEs. This is an interesting result, as state ownership is often considered to be incompatible with efficiency according to previous research (Ramaswamy, 2001; Zhou et al., 2017). A possible reason for our surprising finding is that SO EMEs have privileged networks and resources that can help SOEs to better utilize their innovation input. Our study thus extends the institution-based view by providing an alternative perspective to understand the role of state ownership on firm efficiency.

However, we find that SO EMEs are less able to establish the innovation to internationalization link. A plausible explanation could be that SO EMEs are unable to commercialize their innovation outputs globally, perhaps due to relatively weak capability of SO EME managers to secure attractive international opportunities for exploiting their innovation outputs. Since the SOE managers are appointed by the government more for political reasons rather than their managerial capabilities and skills, they may not be as successful as non-SOE managers in exploiting the innovation outputs in international markets. This is in line with previous research that suggests non-SOE managers to be more business-orientated than SOE managers (e.g., Hu & Xu, 2022; Zhou et al., 2017). Furthermore, SOEs’ success in protected domestic markets may further undermine their ability and incentive to seek new markets abroad to utilize their innovation outputs.

Overall, our study integrates organizational learning theory and institutional theory to capture both firm- and institutional-level phenomena within a single, holistic model of EME internationalization and innovation.

Managerial and Policy Implications

Our study has important implications for EME managers and policymakers interested in developing their internationalization and innovation strategies. Our results show that EME managers can enhance firms’ knowledge exploration through internationalization, which in turn encourages more innovation input. Therefore, internationalization provides the impetus and opportunities for latecomer EMEs to enhance innovation activities and catch up with global technology leaders. The study also shows that SOEs are more efficient in transforming their innovation input to produce more innovation output compared with non-SOEs, possibly because of the superior network resources of SOEs. An important implication of this finding for non-SOE managers is to develop connections with policymakers and seek government support to enhance their networking resources and capabilities. On the other hand, we find non-SOEs are more efficient in utilizing innovation output for internationalization. This result indicates that governments in emerging markets should develop incentives, policies, and programs that encourage and support non-SOEs to enter foreign markets to leverage and exploit their innovation outputs. Conversely, we find that SOEs are relatively weaker in exploiting their innovation outputs via internationalization. SOEs could train and develop their managers to become more adept and agile in internationalization, and redesign their organizations, systems, and processes to motivate them to commercialize their innovation outputs more effectively and efficiently worldwide.

Limitations and Directions for Future Research

Notwithstanding our study’s contributions to the literature on internationalization and innovation in the context of EMEs, there are several interesting directions that may be explored in future research. First, besides the degree and status of internationalization, there are other aspects of internationalization, such as location choice and entry mode, etc., which may be related with innovation development in EMEs (e.g., Piperopoulos et al., 2018; Xie & Li, 2018). Due to data unavailability, these aspects of internationalization could not be examined here, and would be worthy of investigation in future research. Second, our study focuses on technological innovation and does not theorize and test other types of innovation, such as management innovation and process innovation (Azar & Ciabuschi, 2017; Chiva et al., 2014). Future research may extend this research and examine other forms of firm innovation (e.g., business model innovation) and their linkages with EME internationalization strategies. Third, we introduced the idea of knowledge transformation as the bridge between knowledge exploration and knowledge exploitation in the context of EMEs. We believe this is a critical intermediate stage in the knowledge creation and commercialization journey undertaken by all firms, large and small, domestic and international. Notwithstanding its importance, knowledge transformation has received limited attention in the IB literature. We suggest scholars to open the knowledge transformation black box, and dive deeper into the myriad and complex organizational systems, staff, and resources that are orchestrated inside this black box.

Fourth, we tested our theoretical model with data from Chinese firms. Since China is an exemplary country among emerging market economies, a useful extension of our study would be to test our theoretical model in the context of other emerging economies, especially the major ones such as Brazil, Russia, India, and Mexico. The different institutional settings in these economies could potentially produce different result with regard to the determinants and consequences of internationalization and the role of state ownership (De Beule & Duanmu, 2012; Hu et al., 2019). Such country-level differences may be further complicated by the likelihood that EMEs from different countries may vary in their motivations, resources, strategies, and capabilities for innovation and international expansion. Cross-country comparative studies would also enable us to tease out the firm-level versus country-level sources of differences in the internationalization and innovation strategies of EMEs from different countries. Finally, although we took multiple steps to overcome endogeneity bias in our model estimation, due to the non-availability of appropriate firm-level instrumental variables that are both conceptually and empirically related to the explanatory variables but not with the dependent variable, we could not formally conduct extensive endogeneity bias tests for our empirical model. In future research, scholars are encouraged to include appropriate instrumental variables and test the robustness of our findings against endogeneity bias.

Conclusion

In the last few years, there has been exponential growth in examining the internationalization and innovation strategies of EMEs, and in comparing and contrasting their strategies with developed economy firms. While these studies have greatly improved our understanding of the strategy and performance of EMEs, the baseline for most of these studies continue to be the theoretical models created with firms from developed countries. Given the relative growth and maturity of EMEs in recent times, there is a need for management researchers to switch the lenses through which we see the world of international business. In other words, the theoretical models based on firms from emerging markets (such as the one proposed and tested in this study) could be used as the baseline, and these EME based models could then be tested and replicated with firms from other developed and developing economies. Such an approach will enable us to expand and enlarge our conceptual toolkit, and allow us to see, understand, appreciate, and accept the world of business through diverse and multiple lenses.

Appendix

See Tables 8, 9, 10, 11.

Funding

This study is supported by research grants (National Natural Science Foundation of China, Grant/Award Numbers: 71972056; Natural Science Foundation of Guangdong Province, Grant/Award Number: 2020A1515011036).

Declarations

Conflict of interest

None.