INTRODUCTION

How does a multinational enterprise’s (MNE) internationalization speed affect its financial performance? By internationalization speed, following Vermeulen and Barkema (2002), we mean the average pace at which the MNE builds a network of foreign subsidiaries across host countries after establishing its first foreign subsidiary. As boundedly rational actors (Cyert & March, 1963), MNEs may have to slow down internationalization speed in order to better understand and organizationally adjust to new foreign markets (Penrose, 1959; Vermeulen & Barkema, 2002). This is particularly evident when they enter host countries where they are less familiar (Johanson & Vahlne, 1977). Yet, some MNEs internationalize more rapidly than suggested above. This is especially true for Chinese MNEs (CMNEs), as evidenced by the fact that Chinese firms have climbed up fast on the Fortune Global 500 list over the past two decades or so.1 CMNEs are rising as powerful giants at a surprisingly fast pace and showing no sign of slowing down, even when they arguably have fewer firm-specific advantages (FSAs) than developed-market MNEs (DMNEs). This apparent anomaly has recently been identified as one of the “big questions” for international business (IB) research (Buckley, Doh, & Benischke, 2017). Yet, it is remarkable how little research exists on the financial performance implications of the rise of CMNEs through their increasingly fast internationalization and the distinctiveness of their FSAs.

To provide some insights, we draw upon the home regionalization literature (Rugman & Verbeke, 2004), which highlights the prominent role of geographic space in shaping firms’ operations and outputs. The core argument presented here is that when it comes to financial performance, geographic space is a contextual complement to internationalization speed. Using a set of CMNEs, we test the idea that the effects of internationalization speed on financial performance will depend, in part, on where the foreign investments occur: home and host regions. To determine home and host regions, we use continent-based regional grouping of countries based on geographic proximity, because this grouping has several advantages, including fixed boundaries (Flores, Aguilera, Mahdian, & Vaaler, 2013) and managerial relevance (Ghemawat, 2007; Rugman & Verbeke, 2007). For China, the home region is composed of a geographically proximate, continent-based group of countries throughout East and South Asia as well as Southeast Asia.2 We call these proximate and related regional neighbors intra-regional host countries. By contrast, we consider host regions to be those that include all non-Asian countries in North America, Western Europe, Central and Eastern Europe, Latin America and the Caribbean, Africa and the Middle East, and Oceania. Countries in host regions are classified as inter-regional host countries.

Specifically, we predict that fast-mover CMNEs, compared to slower-movers, are substantially more profitable when expanding into intra-regional host countries, but are less profitable when expanding into inter-regional host countries. That is, faster is not always better. Our rationale is that fast-mover CMNEs may find it not only less costly to operate but also much easier to exploit existing FSAs in intra-regional host countries. In contrast, fast-mover CMNEs may find that distance-induced costs stymie their ability to realize FSAs in inter-regional host countries. Our results show that considering regional spatial heterogeneity clarifies the internationalization speed-to-performance relationship.

Furthermore, we link our theorizing about the fit between the internationalization speed and its environmental context to the notion of home region-bound FSAs. As Rugman (2005) claims, this notion signifies the paradox of FSA vulnerability. On the one hand, an MNE’s FSAs, which are presumed to be home region-bound and path-dependent, can be a strategic asset in the home region. On the other hand, the same FSAs can become a liability in host regions, where they quickly lose relevance and sustainability. This is due to the lingering legacy of the MNE’s home region-bound administrative heritage, defined as its existing organizational attributes such as the configuration of FSAs and organizational routines and capacities, built up over a long period of time in the home region (cf. Bartlett & Ghoshal, 1989). Specifically, we posit that technological and marketing resources possessed by fast-mover CMNEs improve financial performance in intra-regional host countries but hurt financial performance in inter-regional host countries. A test using 767 CMNEs, publicly listed on the Shanghai and Shenzhen stock exchanges over the period 2002–2014, generally supports our hypotheses.

Our study makes two important contributions to the scholarly literature. First, we contribute to the internationalization speed literature by drawing insights from the home regionalization literature. These two streams of scholarship have evolved almost independently from each other, but we integrate them to provide novel insights into how the internationalization speed dilemma can be resolved. As noted, we find that the impact of internationalization speed on firm performance varies remarkably across entry into different geographic regions. Second, our study not only takes a small step toward “a general theory of the [MNE] in space” (Casson, 1987: 1) but also contributes to the theory of the regional MNE by revealing the home region boundedness of fast-mover CMNEs’ core FSAs. We show evidence that CMNEs’ technological and marketing resources can improve financial performance when quickly deployed into intra-regional host countries but can hurt financial performance when quickly deployed into inter-regional host countries. These results challenge conventional mainstream internalization theory of the MNE (Buckley & Casson, 1976) that overstates the global transferability of MNEs’ technological and marketing FSAs that are assumed to have characteristics of global public goods.

THEORETICAL BACKGROUND

The Internationalization Speed Dilemma: Fast-Mover Advantages and Disadvantages

The literature on speed focuses upon the notion of being fast to market. Fast-mover advantages are defined as the ability of the firm to be better off than its competitor as a result of being fast to market. Over past decades, some practitioners have touted speed as “the single most critical factor for success across all markets” (Vesey, 1991: 23) and “the next source of competitive advantage” (Stalk, 1988: 41). Strategy scholars have argued that fast-mover status confers an impressive array of positional advantages, such as early cash flow, market share, visibility, and legitimacy (e.g., Hawk, Pacheco-de-Almeida, & Yeung, 2013; Schoonhoven, Eisenhardt, & Lyman, 1990). The conceptual appeal of fast-mover advantages is also evident in the IB literature. IB scholars argue that the benefits of being a fast-mover generally outweigh the costs and risks that have fallen in recent years for all internationalizing firms (Chetty, Johanson, & Martín, 2014; Guillén & García-Canal, 2010; Ramamurti & Hillemann, 2018). For example, Guillén and García-Canal (2010: 122) note that “[i]n a fast-paced international economy, the risks of falling behind or failing by waiting too long to break out of the home market exceed the hazards inherent in any process of globalization.”

In contrast, the high regard for the thesis that speed benefits performance has long been challenged and resisted (Dierickx & Cool, 1989; Penrose, 1959; Scherer, 1967). Led by boundedly rational managers (Cyert & March, 1963; Johanson & Vahlne, 1977), firms that rush the internationalization process may face time compression diseconomies (Jiang, Beamish, & Makino, 2014; Vermeulen & Barkema, 2002), the notion that shortening the time of organizational learning causes inefficiencies, which in turn decrease firm performance (Dierickx & Cool, 1989). There are clearly other problems from fast internationalization. Picking up the pace of the internationalization process can lead to dynamic adjustment costs (Penrose, 1959). This is because fast-mover MNEs are constrained in their ability to quickly adjust their managerial resources to cope with increased organizational complexity. What is more is that MNEs that rapidly establish foreign subsidiaries may commit many errors because their understanding of host environments may be superficial (Morales-Raya & Bansal, 2015). Thus, the alternative thesis is that slower internationalization may lead to better performance.

Taken together, these two logics paint a messy picture as to how internationalization speed affects MNE performance, thereby resulting in the internationalization speed dilemma. In fact, prior research (e.g., Chang & Rhee, 2011; García-García, García-Canal, & Guillén, 2017; Vermeulen & Barkema, 2002; Yang, Lu, & Jiang, 2017) has implicitly assumed the homogeneity of the foreign locations that an MNE enters either rapidly or slowly, without much consideration to how the benefit–cost analysis is affected by the “contextual characteristics of the [host] environment” (Buckley et al., 2017: 1053). Studies taking the home regionalization perspective have shown that this contextual homogeneity assumption and its resulting one-size-fits-all approach are unwarranted (e.g., Patel, Criaco, & Naldi, 2018; Qian, Khoury, Peng, & Qian, 2010). More relatedly, the Uppsala model is widely believed to suggest that an MNE’s internationalization follow a slow, gradual process; however, Johanson and Vahlne indeed highlight the contingent nature of the internationalization process (Figueira-de-Lemos, Johanson, & Vahlne, 2011; Johanson, & Vahlne, 2009). In other words, the MNE is willing and able to adapt its internationalization speed to the host country conditions.3 Moreover, Vahlne and Johanson (2017) subscribe to Rugman and Verbeke’s (2004) home regionalization perspective. Hence, we next use a home regionalization perspective to theorize the foreign location aspect of MNE activity.

Geographic Space: Home and Host Regions

In recent years, IB scholars have increasingly been concerned with how and why geographic regions matter, especially when it comes to understanding the regional strategies of MNEs (e.g., Buckley & Ghauri, 2004; Ghemawat, 2007; Kim & Aguilera, 2015; Rugman & Verbeke, 2004). Notably, Ghemawat (2007) suggests that MNEs have shifted attention away from country-by-country adaptation and moved it towards a regional basis for geographic aggregation. In fact, the global economy is largely driven by economic and political forces that promote greater regional economic integration (Buckley & Ghauri, 2004). Hence, MNEs may find it more beneficial to integrate their foreign subsidiaries at the regional level. Echoing this view, Arregle, Miller, Hitt, and Beamish (2013) use regions as the basic units of geography and argue that MNEs are able to develop and exploit region-bound FSAs and organizational learning, which by and large are tied to a specific region.

The concept of region, as used here, is defined as “a grouping of countries that are relatively similar to each other and relatively dissimilar to countries in other regions” (Verbeke & Asmussen, 2016: 1054, italics in original). However, this concept can be defined along various criteria – for example, in terms of geographic proximity (Arregle et al., 2013), cultural proximity (Ronen & Shenkar, 1985), and institutional development (Hoskisson, Eden, Lau, & Wright, 2000). As noted earlier, our study joins and extends the home regionalization perspective (Rugman & Verbeke, 2004), and hence uses the regional grouping of countries based on geographic proximity. In particular, we dichotomously distinguish between home and host regions. Mudambi, Li, Ma, Makino, Qian, and Boschma (2018) refer to the use of such a home–host region dichotomy as a zoom-out approach, which is indeed instrumental in advancing existing theories of the MNE such as the Uppsala model (Vahlne & Johanson, 2017).

More importantly, such a grouping scheme has been widely used in the home regionalization literature (e.g., Banalieva & Dhanaraj, 2013; Goerzen & Asmussen, 2007; Kim & Aguilera, 2015; Patel et al., 2018; Qian et al., 2010), and further, has better international corporate strategy and performance implications for the vast majority of MNEs than other grouping schemes noted above (Rugman & Verbeke, 2007). The vast majority of MNEs, including CMNEs (Buckley et al., 2007; Child & Rodrigues, 2005), are not global but regional (Rugman & Verbeke, 2004). As geographic regions are far from homogenous and the geographic reach of the MNE’s FSAs may be limited to its home region, the costs and benefits associated with internationalization differ between home and host regions (Goerzen & Asmussen, 2007; Patel et al., 2018; Qian et al., 2010; Rugman & Verbeke, 2004). Hence, the home regionalization perspective offers a fertile contextual ground for theorizing about the boundary conditions under which internationalization speed contributes, either positively or negatively, to financial performance. With these two distinct regions as the primary geographic unit of analysis, we next hypothesize how the financial performance implications of internationalization speed depend upon where CMNEs internationalize and then investigate the moderating effects of technological and marketing resources that constitute two specific parts of their FSAs.

HYPOTHESES

The Speed of Intra-regional Versus Inter-regional Internationalization

If there are spatial contexts under which fast-mover advantages can outweigh fast-mover disadvantages, the home region (in our case, Asia) is clearly one of them because of geographic proximity and because of cultural, economic, and/or institutional similarities among intra-regional host countries. From this perspective, we begin by predicting that a faster speed of intra-regional foreign expansion by a CMNE leads to higher financial performance because its benefits are likely to outweigh its costs.

On the cost side of the cost–benefit equation, we expect that the total costs of faster intra-regional internationalization are likely to increase, but at a diminishing rate by country. MNEs may be able to establish foreign subsidiaries in intra-regional host countries rapidly, since their home region is intuitively viewed as “the easiest region into which to make an expansion” (Delios & Beamish, 2005: 33). This view fits comfortably with “a model of rational internationalization” (Johanson & Vahlne, 2009: 1413), given that MNEs can expect more similarities than differences among countries within a home region (Ghemawat, 2007) – for example, in terms of business practices (Hastings, 1999) and customer preferences (Jones, Kanno, & Egawa, 2013).

In light of this, Johanson and Vahlne (1977: 29) point out that “[a lack of experiential knowledge] is an important reason why the internationalization process often proceeds slowly.” At the same time, however, they acknowledge that if the firm pursuing internationalization “has much experience from other markets with similar conditions” (1977: 30–31), its experiential learning advantages – ones that stem from accumulated experience and existing knowledge within the home region in our study – will likely enable the MNE to “speed up this process.” This description implies that because of the small distance between the home country and the rest of the home region, the fast-mover MNE may readily overcome the liability of intra-regional foreignness, or at least significantly reduce the costs of doing business intra-regionally (Rugman & Verbeke, 2007). It follows that the same MNE will likely encounter fewer problems of time compression diseconomies. It is, of course, inevitable that as fast-mover CMNEs establish more foreign subsidiaries in intra-regional host countries within short periods of time, their coordination costs will rise. However, given the similarities within the home region, it seems likely that these same firms will become increasingly adept at coordinating the complexity of their subsidiaries and hence, for every additional subsidiary, the marginal costs of coordination will decrease (Patel et al., 2018; Qian et al., 2010).

On the benefit side of the cost–benefit equation, we maintain that compared to slow-mover ones, fast-mover CMNEs are better positioned to leverage their home region-based FSAs to achieve superior financial performance (Chetty et al., 2014; Rui & Yip, 2008). The substantial benefits enjoyed by faster-mover CMNEs are expected to stem from what Kim and Aguilera (2015) call “intra-regional exploitation,” by which they modify and adapt their existing FSAs for the home region. Specifically, Kim and Aguilera argue that intra-regional internationalization is geared toward exploitation rather than exploration. As March (1991) notes, the essence of exploitation is to refine and extend existing knowledge and the returns from such exploitation are positive, proximate, and predictable. More importantly, it is the considerable overlap in the knowledge base between MNEs’ existing internal knowledge and the external knowledge that is required to adapt to intra-regional host countries that is likely to enhance the benefits from the exploitation of home region-based FSAs. In addition, for every additional subsidiary, performance is further enhanced through the repeated use of the exploitation routines.

Of course, intra-regional host countries are not a large, homogeneous group.4 It is important to recognize the existence of sub-regions, defined as “groups of countries within a region that have much more in common with one another than with other countries in the region” (Osegowitsch & Sammartino, 2008: 191). In Asia, which is CMNEs’ region of origin, cultural sub-regions can exist so that some intra-regional host countries are more culturally similar to one another than others (Ronen & Shenkar, 1985). Here, the existence of ethnic networks in intra-regional host countries matters a great deal (Estrin, Meyer, & Pelletier, 2018), especially for CMNEs (Buckley et al., 2007). This may be why Hong Kong, Macao, Taiwan, Indonesia, Thailand, Malaysia, and Singapore are the major foreign direct investment (FDI) destination countries for CMNEs, where Chinese migrants have emerged as the principal economic players (Buckley et al., 2007; Child & Rodrigues, 2005). As there are more similarities among these seven Asian countries based on Chinese ethnicity, CMNEs become insiders in their ethnically based foreign networks and, therefore, benefit from proximity advantages (Johanson & Vahlne, 2009).

Although CMNEs tend to highly concentrate their intra-regional international activities in these seven countries, some also venture into culturally less similar intra-regional host countries (Buckley et al., 2007). However, intra-regional psychic distance is decreasing in Asia, as intra-regional trade and integration grow rapidly between Asian countries (Ghemawat, 2007). McKinsey Global Institute (2019: 18) reports that considering the increasing trend toward home regionalization within the Asian region, firms in all industries “[p]rioritize speed to market and proximity to customers…. Speed to market enables faster responses to what customers want and less product waste from forecasting errors.”5 In fact, home-region similarities and intra-regional proximities may enable fast-mover CMNEs to have more rapid knowledge absorption and exploitation, which would, in turn, boost performance more quickly. By contrast, slow-paced intra-regional internationalization, with fewer foreign subsidiaries in intra-regional host countries over a longer period of time, may decrease experiential learning (Yang et al., 2017). As a result, slow-mover CMNEs are likely to have positional disadvantages in Asia, since they fail to speed the exploitation of home region-specific FSAs before market opportunities disappear.

The above discussion leads us to propose the following hypothesis:

Hypothesis 1a:

A faster speed of internationalization into intra-regional host countries by CMNEs increases financial performance.

From a home regionalization perspective, it is tempting to think of inter-regional internationalization as unwise geographic overstretching (Verbeke & Asmussen, 2016), too risky a strategy for CMNEs that generally lack international experience (Estrin et al., 2018). Notably, however, Ghemawat (2007: 35) argues, somewhat counterintuitively, that “distance isn’t always bad,” touting the benefits of setting up foreign subsidiaries outside the home region. Referring to the regional portfolio strategy, Ghemawat contends that MNEs focusing too much on their home region may run out of room to grow and may find it necessary to pursue inter-regional expansion. Delios and Beamish (2005) document that Japanese MNEs with a host region-oriented strategy, inter-regional in our terminology, outperform their counterparts with a home region-oriented strategy.

Against this backdrop, CMNEs choose to enter inter-regional host countries. Furthermore, it has been observed that largely due to their tendency to “focus on speed and growth (a revenue mindset)” (Whitler, 2019: 78), some CMNEs venture swiftly beyond the confines of their home region into inter-regional host countries (Ramamurti & Hillemann, 2018). They may have to speed up the inter-regional internationalization process to catch up with earlier moving competitors, especially DMNEs (Guillén & García-Canal, 2009). However, we argue below that faster inter-regional internationalization of CMNEs, albeit purposeful, diminishes their financial performance.

With regard to costs, rapidly establishing foreign subsidiaries in inter-regional host countries may incur substantial distance costs. As Verbeke and Asmussen (2016) note, inter-regional distance represents a quantum leap when compared to intra-regional distance. On the entry market side, the MNE incurs significant adjustment costs arising from engaging in business in inter-regional host countries, as home and host regions differ in important ways – for example, in terms of national cultures, business practices, and customer tastes (Ghemawat, 2007; Rugman & Verbeke, 2007). On the organizational side, an MNE will likely find it more difficult and expensive to acquire full and accurate information about subsidiaries’ operations when these subsidiaries are located far from its home country (Hennart, 2007). In addition, Hitt, Hoskisson and Kim (1997) and Qian et al. (2010) have suggested that as the MNE sets up its foreign subsidiaries in inter-regional host countries, the coordination costs of these widely dispersed subsidiaries increase significantly. Our study goes beyond these studies by contending that compared to a slow-mover, a fast-mover CMNE’s coordination costs grow due to time constraints in governing and integrating its foreign subsidiaries. Accordingly, the total costs of faster speed of inter-regional expansion are likely to rise at an increasing rate.

On the other side of the cost–benefit equation, the benefits of internationalization speed may be at best marginal, and at worst leveling off, at least in the short run, when CMNEs expand quickly into inter-regional host countries. As we argued in developing Hypothesis 1a, while the exploitation of homegrown FSAs is critical to a CMNE’s performance, inter-regional distance may impede its ability to automatically or easily translate these advantages into inter-regional host countries. Additionally, rearranging their existing FSAs often involve organizational unlearning or forgetting (Bettis & Prahalad, 1995), given that so many MNEs sometimes find it necessary to reshape products that are developed at home and start with a zero-based assessment of customer needs due to substantial differences across countries outside the home region (Immelt, Govindarajan, & Trimble, 2009).

Of course, when rapidly expanding into emerging-market inter-regional host countries, such as those in Eastern Europe and Latin America, CMNEs may expect “more similarities than differences in terms of resource conditions, the approaches to doing business, and their understanding of environmental conditions” (Hitt, Li, & Worthington, 2005: 369–370). In a sense, they possess non-traditional FSAs such as know-how to compete in emerging markets where institutional environments are weak and market volatility and political instability are hazardous (Cuervo-Cazurra & Genc, 2008; Guillén & García-Canal, 2009). Because of such advantages, some CMNEs may pursue fast inter-regional expansion into emerging markets. However, as these same CMNEs face market competition, they may not only run into trouble due to distance costs but also increase their speed costs associated with time compression diseconomies. These combined costs may increase much faster than any anticipated benefits when CMNEs accelerate their expansion into emerging-market inter-regional host countries (Johanson & Vahlne, 1977, 2009), especially given that they still have to engage in costly and time-consuming “inter-regional exploration,” which refers to the pursuit of “new possibilities and learning of distant knowledge in order to build the necessary capabilities to operate in a new region” (Kim & Aguilera, 2015: 118). In a sense, inter-regional exploration represents “long jumps in time and space from the company’s knowledge base” in the sense that a knowledge gap exists between its current knowledge base and the new knowledge needed to compete in inter-regional host countries (Barkema & Drogendijk, 2007: 1136; Cohen & Levinthal, 1990). In such a case, fast internationalization may neither be efficient nor profitable.

We next argue that CMNEs may face many challenges when venturing into developed-market inter-regional host countries such as North America and Western Europe in order to transfer new FSAs acquired in these developed markets back to the home country (Cuervo-Cazurra, Newburry, & Park, 2016; Luo & Tung, 2018; Meyer, 2005). Clearly, their fast internationalization into developed-market inter-regional countries is not just about exploratory learning, but also about strategic asset-seeking (Luo & Tung, 2007; Ramamurti & Hillemann, 2018). In other words, they purport to augment their inferior home-grown FSAs, move up the value chain (from manufacturing to R&D and branding), and enhance their long-term competitive advantage by acquiring and absorbing strategic resources such as sophisticated technology and brands (Buckley et al., 2007; Child & Rodrigues, 2005; Deng, 2007). As argued by Madhok and Keyhani (2012), however, CMNEs may suffer from the liability of emergingness, that is, the extra costs incurred by emerging-market MNEs (EMNEs) that expand into developed markets. Indeed, CMNEs are very likely to face strong competition from local firms with better FSAs and be subject to expensive entry costs (Kim, Hoskisson, & Lee, 2015).

Synthesizing our discussion, we propose the following hypothesis:

Hypothesis 1b:

A faster speed of internationalization into inter-regional host countries by CMNEs decreases financial performance.

Resource Heterogeneity: Moderating Effects of Technological and Marketing Resources

While predicting that fast-mover advantages and disadvantages for CMNEs depend on heterogeneous regional contexts, the above-mentioned hypotheses do not address firm resource heterogeneity (Barney, 1991; Luo & Tung, 2018). In this regard, our study focuses specifically on CMNEs’ technological and marketing resources, which represent two critical types of core FSAs, to draw the important implications of regional MNEs for “research adopting a resource-based approach to MNE functioning” (Rugman & Verbeke, 2004: 13). The former refer to technological FSAs, including technological knowledge, patents, and so forth that are generated as a byproduct of its history- or path-dependent R&D activity (Cohen & Levinthal, 1990), whereas the latter refer to marketing FSAs, including brands and reputation for quality (Goerzen & Asmussen, 2007).

According to conventional internalization theory (Buckley & Casson, 1976), MNEs’ technological and marketing resources are non-location-bound FSAs, under the assumption that they are easily transferred to global markets “at little or no cost” (Caves, 1971: 6). Hence, such FSAs positively moderate the relationship between internationalization and financial performance (e.g., Lu & Beamish, 2004; Morck & Yeung, 1991). However, if this global transferability assumption holds true, why should the overwhelming majority of MNEs, even ones with strong technological and marketing FSAs, be regional rather than global (Rugman & Verbeke, 2004)? To understand why, we seek to provide more nuanced and enriched insights into the moderating financial performance effects of fast-mover CMNEs’ technological and marketing resources, especially outside their home region.

Some scholars claim that firms achieve high profits not because they possess more or better resources, but because they deploy these resources more effectively in given environmental contexts (Priem & Butler, 2001). This highlights the importance of the fit between the firm’s resources and specific markets in which those resources are used effectively. Thus, in order for MNEs to improve financial performance, they may need to enter foreign markets where they are able to exploit similarities across such markets because they have less need to significantly adapt such resources to fit the new environment (Szulanski & Jensen, 2006). Following this fit logic, fast-mover CMNEs may find it sensible to quickly apply their resources, especially those that are a critical part of their FSAs, to the needs and preferences of actors within their home region (Rugman & Verbeke, 2007). Thus, we predict that when being better off with a faster speed of intra-regional internationalization, faster-mover CMNEs are likely to enjoy better performance from deploying their technological and marketing resources intra-regionally than slower-mover CMNEs. Stated formally:

Hypothesis 2a:

CMNEs’ technological resources are likely to positively moderate the relationship between the faster speed of internationalization into intra-regional host countries and financial performance.

Hypothesis 2b:

CMNEs’ marketing resources are likely to positively moderate the relationship between the faster speed of internationalization into intra-regional host countries and financial performance.

The home regionalization perspective (Rugman & Verbeke, 2004, 2007) assumes that an MNE’s core FSAs can become deployable in the entire home region but benefit the MNE only in the home region. If this perspective is correct, the aforementioned logic may not hold when CMNEs rapidly cross-regional borders, which are viewed as discrete discontinuities in space (Aguilera, Flores, & Kim, 2015), by entering inter-regional host countries. Rather, we expect the opposite due to the lingering effects of the MNE’s home region-bound administrative heritage (Rugman, 2005).

An MNE’s administrative heritage refers to its existing organizational attributes such as the configuration of FSAs, built up over the decades in the home country (Bartlett & Ghoshal, 1989). As predicted in Hypotheses 2a and 2b, this heritage can be a strategic asset for the MNE. However, it can be a significant liability in the MNE’s foreign expansion, because “it resists change and thereby prevents realignment or broadening of strategic capabilities” (Bartlett & Ghoshal, 1989: 38). Collis (1991) observes that an MNE’s administrative heritage constrains strategic choice, which in turn hampers its ability to economically optimize its global asset configuration. The dark side of administrative heritage is also consistent with previous IB research (e.g., Birkinshaw & Ridderstråle, 1999; Monteiro, 2015; Song & Shin, 2008), which has revealed paradoxical patterns – and disruptive consequences – associated with an MNE’s technological resource accumulation efforts. For example, Song and Shin (2008) find that the most technologically capable MNEs are the ones that are most prepared but least motivated to source novel knowledge from host countries. The reason is that because these MNEs have already established their idiosyncratic technological trajectories based on their strong existing knowledge base, they tend to engage in greater exploitative change, while also rejecting externally sourced knowledge. This finding confirms that the MNE would make a type I error because the corporate immune system (Birkinshaw & Ridderstråle, 1999) or the not-invented-here syndrome (Cohen & Levinthal, 1990) tends to reject many promising technological initiatives conducted in host countries and that technological exploitation drives out exploration (Levinthal & March, 1993).6

Extending the notion of administrative heritage to the home regionalization perspective, Rugman (2005: 197) bluntly argues that “the administrative heritage of most MNEs, undoubtedly conducive to home-region market success, may well constitute an administrative rigidity when attempting to penetrate host regions.” Anecdotes show that MNEs’ technological resources that have been developed and contextually embedded in their home regions are made less transferable to host regions, where cross-regional differences in consumer preferences, business practices, government regulations, and sociocultural and institutional norms, may render technological FSAs more vulnerable, as well as less sustainable (e.g., Immelt et al., 2009; Sőlvell, 2006). Of course, MNEs may be compelled to immediately engage in technological exploration to learn what is required for the development of host region-specific, technological knowledge. More often than not, however, anticipated outcomes can be suboptimal, particularly in the short run, because the MNE’s home region-bound administrative heritage may constrain its capacity to absorb host region-specific, technological knowledge. This constraint may also explain why MNEs find it more difficult to develop dynamic capabilities than suggested by Teece (2014).

A major part of the challenge to the development of an MNE’s dynamic capabilities that is especially well suited to the technological needs of host regions may arise from the tendency of the MNE to credit – and thus cling to – established sets of routines that work so well for technological knowledge in its home region (cf. Levinthal & March, 1993; Levitt & March, 1988; Nelson & Winter, 1982). The mechanism behind this paradox may be deeply rooted in an MNE’s home region-bound administrative heritage (Rugman, 2005) that is difficult to abandon. For example, Monteiro (2015) asserts that senior MNE managers tend to accept externally sourced technologies that simply fit well with existing ones or confirm their existing dominant logic, while also screening out dissonant external technologies, especially those sourced from more distant host countries. Furthermore, Banalieva and Dhanaraj (2013), using a sample of U.S., European, and Japanese MNEs, document that while technological FSAs enable MNEs to expand into host regions, it is these FSAs that diminish rather than enhance their financial performance.

The key implication of prior studies is that if MNEs’ core FSAs are embedded in their home region-bound administrative heritage (Rugman, 2005), MNEs with more technological resources (as a determinant of absorptive capacity [Cohen & Levinthal, 1990]) are likely to find it more difficult and time-consuming to unlearn and leave behind much of home region-specific technological knowledge that may, in reality, inhibit the host region-specific technological knowledge accumulation process than MNEs with less technological resources.7 It is by now conventional wisdom that MNEs with more technological resources are better able to value, assimilate, and apply new technical knowledge even sourced from developed-market inter-regional host countries to commercial ends (e.g., Delios & Beamish, 2005). However, this conventional wisdom is not necessarily true for EMNEs. For example, Luo and Tung (2018: 134) point out that “[o]ne particular challenge for EMNEs pertains to their weak skills in organizing the transfer, diffusion, and integration of what they have acquired abroad with what they already possess at home.” In particular, as EMNEs pursue fast internationalization into developed-market inter-regional host countries, they are likely to have “little time to evaluate their foreign experience, assimilate it, and apply it to commercial ends” (Vermeulen & Barkema, 2002: 640). Moreover, there are enough examples of EMNEs, including CMNEs, seriously struggling to absorb new technological resources acquired in these host countries (Cuervo-Cazurra et al., 2016; Meyer, 2005).

In a situation where CMNEs’ absorptive capacity looks doubtful, their accelerated innovation (Williamson, 2016) can do more harm than good because as time is compressed to complete R&D projects, the total R&D costs grow at an increasing rate due, in part, to less information and more mistakes, thereby yielding diminishing marginal returns (Dierickx & Cool, 1989; Scherer, 1967). The detrimental consequences of accelerated innovation may become more severe when CMNEs penetrate a far-away region. As the home region-bound administrative heritage of MNEs may handicap their timely, effective adjustments to inter-regional customers’ technical needs, the accelerated pace of inter-regional internationalization may make matters worse. TCL, a Chinese maker of electronics and home appliances, is a particularly painful example (Cuervo-Cazurra et al., 2016). In 2004 alone, the company spearheaded quick forays into inter-regional acquisitions of two European giants of Thomson SA and Alcatel Mobile Phone. However, these investments were not successful and TCL ran up huge losses until 2009, which threatened its very existence.8

The above discussion suggests that when the inter-regional internationalization process may need to proceed slowly to better understand the technical needs of inter-regional customers (Johanson & Vahlne, 1977, 2009), faster-mover CMNEs are likely to gain fewer performance benefits from their technological resources than are slower-mover CMNEs.

Hypothesis 3a:

CMNEs’ technological resources are likely to negatively moderate the relationship between the faster speed of internationalization into inter-regional host countries and financial performance.

Conventional internalization theory (Buckley & Casson, 1976) assumes that MNEs’ marketing resources are non-location-bound or globally transferable FSAs so that marketing FSAs can be deployed with little adaptation to complement their technological FSAs – and hence exploited profitably – outside their home regions. However, a home regionalization perspective questions this assumption by taking a stance that MNEs’ marketing resources, which maintain a particularly strong home-region focus, experience “a rapid decay when penetrating a host region” (Rugman & Verbeke, 2007: 204). If so, CMNEs that possess a home region-based marketing advantage may see such an advantage erode rapidly as they meet new customers in host regions such as North America, Europe, and South America. More severely, inter-regional customers may simply be too costly to acquire, retain, and serve. For example, “[f]ew Chinese brands have so far managed to make a name for themselves in Western Europe despite heavy advertising and a visible presence in major airports and city centers. A key issue is that Europeans still do not know much about Chinese companies.”9 China’s stated-owned automaker FAW Group Corporation, which produced reliable, low-cost cars and established its brand reputation in China, experienced the same issue and thereby shut the operation down in Mexico (Cuervo-Cazurra et al., 2016).

CMNEs may be required to find new ways to cater to inter-regional customers with different tastes and preferences. However, CMNEs’ administrative heritage can constrain their ability to learn about new customers in inter-regional markets or even encourage superstitious learning whereby cause and effect are misspecified (Levitt & March, 1988). For example, cultural marketing is prevalent in China. As China, like many Asian countries, represents a strong collectivist culture, CMNEs tend to encompass many cultural aspects (e.g., customs, traditions, values, etc.) of collectivism rooted in Confucianism in their marketing efforts to build successful brands (Melewar, Meadows, Zheng, & Rickards, 2004). This heritage can be an asset in their home region, especially in Southeast Asia, where Chinese brands can utilize profitably cultural marketing by deepening their relevance and connection with local audiences of the Chinese diaspora. As Rugman (2005) argues, however, CMNEs’ home region-bound marketing FSAs can be a liability in host regions characterized by individualist cultures, Hispanic cultures, or Islamic cultures, thereby becoming irrelevant and disconnected with those customers.

The above discussion suggests that accumulating valuable marketing resources – brands, customer bases, and distribution channels – in host regions can be greatly time-consuming; accordingly, as Johanson and Vahlne (1977) point out, CMNEs may need to slow down their inter-regional internationalization to learn the requirements of inter-regional markets. In this situation, some CMNEs may seek to rapidly accumulate new marketing resources by even acquiring locally obsolete brands, as exemplified by TCL’s acquisitions of the Thomson and RCA brands in Europe and North America. However, their spending may contribute less due to time compression diseconomies. That is, as CMNEs choose to shorten the time needed to develop marketing FSAs that are host region-specific, their marketing costs are likely to increase exponentially but may be unable to immediately achieve a marketing advantage. If so, CMNEs’ faster inter-regional internationalization, which is posited to yield the negative performance consequences of time compression diseconomies, can be more damaging to financial performance as they may gain fewer returns and incur greater cost penalties associated with the rapid accumulation of their marketing resources. Stated formally:

Hypothesis 3b:

CMNEs’ marketing resources are likely to negatively moderate the relationship between the faster speed of internationalization into inter-regional host countries and financial performance.

METHODS

Data

To test the hypotheses, we construct a panel data sample from multiple sources. We begin with all firms listed on the Shanghai and Shenzhen stock exchanges, the two stock exchanges in mainland China. The panel data for the period 2002–2014 give us a sufficiently long window to systematically trace Chinese firms’ international expansions. We chose 2002 as the starting year because Chinese firms in general and private firms (i.e., non-state-owned enterprises) in particular began to internationalize rapidly after December 2001 when China became a member of the World Trade Organization. To maintain consistency with prior studies on the relationship between the speed of internationalization and financial performance, we focus on non-financial firms (Vermeulen & Barkema, 2002). As stipulated by the China Securities Regulatory Commission, all companies listed on the stock exchanges must disclose their annual financial and investment information. This allows us to collect Chinese firms’ financial data and identify their intra- and inter-regional expansion.

We obtained financial information of publicly listed Chinese companies from the China Stock Market Financial Statements Database provided by the China Stock Market & Accounting Research (CSMAR).10 We then downloaded annual reports from the two stock exchanges and reviewed them to identify the FDI information of each publicly listed Chinese company, including the foreign markets it entered, the year in which each of its subsidiaries was established in that market, and so forth. We extracted the information on political connections from CSMAR China Listed Firm’s Corporate Governance Research Database. After removing the missing observations, we construct our panel data sets on 2780 Chinese publicly listed firms, of which 2013 are domestic firms (these firms report no FDI), and 767 are CMNEs.

Measures

Dependent variable

Our study uses return on assets (ROA), an accounting-based measure of financial performance most commonly used in the literature (e.g., Hitt et al., 1997; Lu & Beamish, 2004; Vermeulen & Barkema, 2002).11 We lag all independent and control variables by 1 year. For example, when ROA is measured in 2014, the independent and control variables are measured in 2013.

Independent and moderating variables

Our two independent variables include the two aspects of internationalization: speed and geographic space. Previous studies (e.g., Chang & Rhee, 2011; Vermeulen & Barkema, 2002; Yang et al., 2017) have measured the speed of internationalization as the cumulative number of a focal MNE’s foreign subsidiaries established in each year, irrespective of where they were located, divided by the number of years since its first foreign subsidiary was established. A large average number of foreign subsidiaries per year indicates a fast-paced internationalization process. Clearly, this measure as is does not incorporate geographic space: home and host regions. Following Rugman and Verbeke (2004, 2007), we divide host countries in which CMNEs’ foreign subsidiaries were located into intra-regional versus inter-regional host countries. For China, all Asian (non-Asian) countries are classified as intra-regional (inter-regional) host countries. However, there are dangers in lumping them all together because this can potentially mask some important differences, given that not all intra-regional or inter-regional host countries are the same (Osegowitsch & Sammartino, 2008) – for example, in terms of cultural and institutional distance. As a region is too broad a geographic context, Arregle et al. (2013) suggest testing the regionalization hypothesis by splitting home and host regions into several home and host sub-regions. Accordingly, we follow Arregle et al. (2013) to capture potential variations within the home and host regions.

In Asia, which is CMNEs’ region of origin, our study thus splits the home region into culturally more or less similar intra-regional host countries. In fact, this classification scheme corresponds more closely to our earlier description of the internationalization activities carried by CMNEs (Buckley et al., 2007).12 More specifically, we measure the speed of internationalization into culturally more (less) similar intra-regional host countries using the cumulative number of a focal CMNE’s foreign subsidiaries established in Hong Kong, Macao, Taiwan, Indonesia, Thailand, Malaysia, and Singapore (established in Asian countries other than these seven countries) in each year, divided by the number of years since its first foreign subsidiary was established. In parallel with CMNEs’ actual internationalization efforts (Buckley et al., 2007), we also measure the speed of internationalization into emerging-market (developed-market) inter-regional host countries as the cumulative number of the focal CMNE’s foreign subsidiaries established in emerging markets (developed markets) outside the home region in a given year, divided by the number of years since its first established subsidiary. We identify a list of these emerging markets through Hoskisson et al. (2000). Developed markets are primarily located in North America and Western Europe.

In measuring a Chinese firm’s technological resources and marketing resources, we use industry-adjusted R&D and advertising intensity to remove any industry-specific effects. A focal firm’s R&D and advertising intensity is the ratio of its R&D and advertising expenditures to total sales (e.g., Banalieva & Dhanaraj, 2013; Chang & Rhee, 2011; Lu & Beamish, 2004; Morck & Yeung, 1991). Like prior studies, our study uses total R&D and advertising expenditures because information about the amount of R&D and advertising expenditures in each region is unavailable. We normalize the focal firm’s technological and marketing resources by taking the average R&D and advertising intensity of all firms in the same industry that are within the same two-digit SIC code and subtracting this average from the focal firm’s R&D and advertising intensity, which then is divided by the industry’s standard deviation of R&D and advertising intensity.

Control variables

We control for the effect of firm size, measured by the natural logarithm of total assets (Chang & Rhee, 2011). Because debt can influence firm performance, we control for a firm’s debt ratio, measured by the firm’s debt over its total assets (Chang & Rhee, 2011; Vermeulen & Barkema, 2002). Because liquidity is likely to affect firm performance, we control for a firm’s cash flow measured by its cash from operations (Almeida, Campello, & Weisbach, 2004). We also control for product diversification measured as the Herfindahl index: 1 − \(\sum p_{i}^{2}\), where \(p_{i}\) is the proportion of a firm’s sales reported in product group i (Lu & Beamish, 2004).

Because foreign ownership can affect firm performance, we include this variable measured by the percentage of shares owned by foreign investors (Gul, Kim, & Qiu, 2010). We control for the effect of the state-owned enterprise (SOE) status using the classification scheme developed by La Porta, Lopez-de-Silanes and Shleifer (1999). If the government has more than 20% equity ownership in our sample firm, the firm is classified as an SOE and takes the value 1, and 0 otherwise. In addition, because political connections have been related to performance in China, we control for their effect (Zheng, Singh, & Mitchell, 2015). Following Fan, Wong, and Zhang (2007), we review a CEO’s curriculum vitae to identify whether s/he served as a current or former government bureaucrat for the central or local governments, or their agencies. If the CEO served in such positions, the firm is coded 1, and 0 otherwise.

In addition, to control for the multinationality of the firm (Chang & Rhee, 2011; Vermeulen & Barkema, 2002), we use the number of host countries in which the firm has established foreign subsidiaries. As early internationalization (how early a firm becomes an MNE) can affect financial performance, we control for its age at internationalization by calculating the time, in years, between the firm’s founding and its first FDI (Autio, Sapienza, & Almeida, 2000). Following Vermeulen and Barkema (2002), we include rhythm of internationalization as a control variable to partial out the effect of the rhythm or regularity of the internationalization process. This variable is measured as the kurtosis of the first derivative of the number of the firm’s foreign subsidiaries over time. We also control for international experience, which is measured by the number of years since the firm established its first foreign subsidiary (Zahra, Ireland, & Hitt, 2000). To estimate the ownership control effect (e.g., joint ventures are only partly controlled by the parent), we calculate the average subsidiary ownership held by a focal firm. Because the foreign modes of entry can affect MNE performance, we control for the ratio of foreign acquisitions measured as the number of foreign acquisitions divided by the number of total foreign subsidiaries established by the firm (Vermeulen & Barkema, 2002). Using data compiled by Berry, Guillén, and Zhou (2010), we also control for the average cross-national distance between China and host countries that focal Chinese MNEs have entered in terms of Ghemawat’s (2007) cultural, administrative, geographic, and economic dimensions.13

Finally, we control for year and industry fixed effects. As our data cover multiple years and industries, we include 11 year dummy variables and 16 industry dummy variables at the three-digit level of industrial classification categorized by the China Securities and Regulatory Commission.

Analysis

The data are constructed as a panel and our sample firms were repeatedly observed. Cross-sectional heteroscedasticity and within-firm serial autocorrelation that violate the basic assumption of ordinary least square (OLS) regression may exist. Following prior studies (e.g., Kim et al., 2015; Lu & Beamish, 2004), we adopt generalized least square (GLS) models. To decide between the random- versus fixed-effects approach, we perform the Hausman test with the null hypothesis, suggesting that a systematic difference does not exist between the two approaches. The result could not reject the null hypothesis, indicating that the random-effects approach is preferred over the fixed-effects approach and produces efficient and consistent estimates (Greene, 2003). In analyzing data, we also specify the cluster option in Stata 14 to produce standard errors that are robust to cross-sectional heteroscedasticity and within-firm serial autocorrelation.

To address selection bias and endogeneity, our study uses Bascle’s (2008: 287) criteria and employs the Heckman-type two-stage estimation procedure in conjunction with exclusion restrictions (Heckman, 1979; cf. Certo, Busenbark, Woo, & Semadeni, 2016; Hamilton & Nickerson, 2003), which apply the same criteria for valid instrumental variables.14 Estimates for the first-stage ordered probit model are used to generate the inverse Mills ratio (λ), which is included as an additional control in the second-stage outcome regression. In the first-stage selection model, a full sample consisting of both domestic firms and CMNEs is used, with a total of 25,885 firm-year observations among 2780 Chinese publicly listed firms. However, the second-stage outcome model only employs 2692 firm-year observations from 767 CMNEs because internationalization speed is not measurable for 2013 domestic firms. Methodological details are described in the Appendix.

RESULTS

Table 1 reports descriptive statistics and pairwise correlation of the variables used in the second-stage outcome model. We calculate the variance inflation factor (VIF) for multicollinearity. The maximum value of VIF is 3.80, the minimum value of VIF is 1.04, and the mean value of VIF is 1.43. In addition to the VIF tests, a careful examination of correlations between variables shows that multicollinearity is not a severe problem for this study.

Table 1 Descriptive statistics and correlation
Full size table

Table 2 reports the results of the first-stage selection model. As expected, a provincial governor’s tenure is added to the first-stage selection model, and it is positive and statistically significant (β = 0.075, p = 0.012). Table 3 reports the results of hypothesis tests for ROA. Hypothesis 1a predicts that a faster speed of internationalization into intra-regional host countries by CMNEs will increase financial performance. As shown in Model 2 in Table 3, the coefficient of speed of internationalization into culturally more similar intra-regional host countries is positive and statistically significant (β = 0.543, p = 0.000). Ceteris paribus, a one standard deviation increase of speed of internationalization into culturally more similar intra-regional host countries increases ROA by 0.255 (0.543 × 0.47). With average ROA of 4.31 in our sample, this figure represents a 5.92% increase in ROA. In addition, we find that the coefficient of speed of internationalization into culturally less similar intra-regional host countries is positive and statistically significant (β = 2.096, p = 0.000). Ceteris paribus, a one standard deviation increase of speed of internationalization into culturally less similar intra-regional host countries increases ROA by 0.440 (2.096 × 0.21), representing a 10.21% increase in ROA. Overall, Hypothesis 1a is supported.

Table 2 First-stage Heckman selection model
Full size table
Table 3 Second-stage Heckman outcome model: effects of the speed of intra-regional and inter-regional internationalization on ROAt+1
Full size table

Hypothesis 1b predicts that a faster speed of internationalization into inter-regional host countries by CMNEs will decrease financial performance. As shown in Model 2 in Table 3, the coefficient of speed of internationalization into emerging-market inter-regional host countries is negative and statistically significant (β = − 1.272, p = 0.000). Ceteris paribus, a one standard deviation increase of speed of internationalization into emerging-market inter-regional host countries decreases ROA by 0.280 (− 1.272 × 0.22), representing a 6.50% decrease in ROA. In addition, we find that the coefficient of speed of internationalization into developed-market inter-regional host countries is negative and statistically significant (β = − 0.461, p = 0.000). Ceteris paribus, a one standard deviation increase of speed of internationalization into developed-market inter-regional host countries decreases ROA by 0.327 (− 0.461 × 0.71), representing a 7.59% decrease in ROA. Hence, Hypothesis 1b receive support.

Hypotheses 2a and 2b predict that CMNEs’ technological resources and marketing resources positively moderate the relationship between their faster speed of internationalization into intra-regional host countries and firm financial performance. As reported in Model 3 in Table 3, the coefficient of speed of internationalization into culturally more similar intra-regional host countries × technological (marketing) resources is positive and statistically significant (β = 31.072, p = 0.002; β = 16.408, p = 0.001). In addition, the coefficient of speed of internationalization into culturally less similar intra-regional host countries × technological (marketing) resources is positive and statistically significant (β = 65.123, p = 0.000; β = 37.209, p = 0.004). Hence, Hypotheses 2a and 2b are both supported. We, following Meyer, van Witteloostuijn and Beugelsdijk (2017), plot their respective interaction effects that are displayed in Figures 1, 2, 3, and 4, respectively. The two outer lines give 95% confidence intervals around the interaction line, and when they are both above (or below) the horizontal zero line, the interaction effect is statistically significant. For example, the interaction line in Figure 1 illustrates that the positive interaction effect, which shows the marginal effect of speed of internationalization into culturally more similar intra-regional host countries, is significant only after the normalized value of technological resources reaches approximately 0.007.

Figure 1
figure 1

The marginal effect of speed of internationalization into culturally more similar intra-regional host countries on ROA (technological resources as a moderator).

Full size image
Figure 2
figure 2

The marginal effect of speed of internationalization into culturally less similar intra-regional host countries on ROA (technological resources as a moderator).

Full size image
Figure 3
figure 3

The marginal effect of speed of internationalization into culturally more similar intra-regional host countries on ROA (marketing resources as a moderator).

Full size image
Figure 4
figure 4

The marginal effect of speed of internationalization into culturally less similar intra-regional host countries on ROA (marketing resources as a moderator).

Full size image

Using Model 4 in Table 3, we next test Hypotheses 3a and 3b, which predict that CMNEs’ technological resources and marketing resources negatively moderate the relationship between their faster speed of internationalization into inter-regional host countries and financial performance. The coefficient of speed of internationalization into emerging-market inter-regional host countries × technological resources is negative and statistically significant (β = − 46.123, p = 0.000), and the interaction effect is plotted in Figure 5. However, we find that the coefficient of speed of internationalization into developed-market inter-regional host countries × technological resources is not statistically significant (β = − 4.074, p = 0.566). Hence, Hypothesis 3a is supported only for emerging-market inter-regional host countries. We also find that Hypothesis 3b is supported only for emerging-market inter-regional host countries. While the coefficient of speed of internationalization into emerging-market inter-regional host countries × marketing resources is negative and statistically significant (β = − 22.504, p = 0.037), the coefficient of speed of internationalization into developed-market inter-regional host countries × marketing resources is not statistically significant (β = − 1.298, p = 0.633). The former’s interaction effect is plotted in Figure 6.

Figure 5
figure 5

The marginal effect of speed of internationalization into emerging-market inter-regional host countries on ROA (technological resources as a moderator).

Full size image
Figure 6
figure 6

The marginal effect of speed of internationalization into emerging-market inter-regional host countries on ROA (marketing resources as a moderator).

Full size image

Robustness Checks

We conduct several additional analyses. We first examine whether a faster speed of internationalization into intra-regional host countries has an inverted U-shaped relationship with firm performance. Although we did not aim to explore such a curvilinear effect, we cannot rule out the possibility that faster intra-regional internationalization may be overwhelming if CMNEs, on average, establish too many foreign subsidiaries in intra-regional host countries in a short amount of time. However, we fail to find a genuine inverted-U relationship, given the fact that an inflection point, a point of the curve where the curvature changes its sign, is “[well] beyond two standard deviations from the mean” (Meyer, 2009: 191).15 This implies that the relationship between the speed of internationalization into intra-regional host countries and financial performance is predominantly positive. Second, while our study uses accounting-based performance, we alternatively use Tobin’s Q (Himmelberg, Hubbard, & Palia, 1999) as a proxy of market-based performance (e.g., García-García et al., 2017; Morck & Yeung, 1991). The results are reported in Table 4. Although we do not discuss these results, they still support our idea that the faster the intra-regional (inter-regional) internationalization, the better (poorer) the firm’s performance and that fast-mover CMNEs’ technological and marketing resources are valuable in intra-regional host countries, but vulnerable in inter-regional host countries. Finally, as the coefficient of the inverse Mills ratio in the second-stage Heckman models is not statistically significant, there is no evidence that sample selection bias is quantitatively critical. Accordingly, we might be able to just run the GLS random-effects model instead of the Heckman two-stage procedure. As reported in Table 5, the results are very similar to those reported in Table 3.

Table 4 Second-stage Heckman outcome model: effects of the speed of intra-regional and inter-regional internationalization on Tobin’s Qt+1
Full size table
Table 5 GLS Random-effects model: effects of the speed of intra-regional and inter-regional internationalization on ROAt+1
Full size table

DISCUSSION AND CONCLUSION

Our study offers a compelling case to rethink how an MNE’s financial performance relates to the speed and geographic space of internationalization. Examining a set of CMNEs, we find that it is not internationalization speed per se that matters for financial performance, but instead a requisite fit between the speed and geographic space of foreign expansion. We specifically posit and discover that CMNEs’ fast speed interplays with intra-regional foreign expansion to increase financial performance, while it also interplays with inter-regional foreign expansion to decrease financial performance. Furthermore, our study considers how the strategic fit between the speed and space of internationalization interacts with CMNEs’ technological resources and marketing resources to affect financial performance. We find that CMNEs’ investments in these two types of resources improve financial performance for fast-movers intra-regionally, but hurt performance for fast-movers inter-regionally. Overall, our findings show that the impact of internationalization speed and firm technological and marketing resources on financial performance varies across host country regional environments.

Our theoretical contribution is twofold. First, we help to resolve the dilemma in the decision speed literature about the advantages and disadvantages of fast or slow internationalization. Some studies show that firms which internationalize quickly perform the best (e.g., Chetty et al., 2014; Guillén & García-Canal, 2010), while other studies show that slower internationalization yields the best performance (e.g., Morales-Raya & Bansal, 2015). Compared to these studies, we offer an important contextual element that conditions this relationship: the home and host regions an MNE enters. In doing so, we integrate research on internationalization speed with insights from the home regionalization literature, which is broadly accepted in IB research, but has not previously been applied to the study of the pace of internationalization, to allow for a more refined understanding of where speedy internationalization may or may not lead to superior financial performance.

It is important to link these two literatures that have separately examined temporal (how fast) and spatial (how far) dimensions of firm internationalization because managers must make joint decisions regarding speed and space of foreign expansion. Previous research (e.g., Chang & Rhee, 2011; García-García et al., 2017; Vermeulen & Barkema, 2002) tends to place MNEs on a continuum from fast- to slow-paced, and accordingly downplays the competing demands of speeding up and slowing down that often coexist within the same firm. Compared to such research, we argue that to remain more profitable, the MNE needs to set up or acquire intra-regional subsidiaries at a faster speed, while establishing inter-regional subsidiaries at a slower speed. Using this linkage, our study provides initial support for a “double-edged sword” effect of fast internationalization on firm performance: fast-mover advantages prevail in the home regional context, while fast-mover disadvantages manifest and slow-mover advantages reveal themselves in host regions away from home.

Relatedly, our study has implications for research on new venture internationalization and international entrepreneurship. International new ventures (Oviatt & McDougall, 1994) or born globals (Cavusgil & Knight, 2015) exemplify early and rapid internationalization.16 That is, these start-up firms begin exporting shortly after their inception, or with little preexisting business experience and knowledge at home, and rapidly increase their international sales growth thereafter. Autio et al. (2000) introduce the concept of learning advantages of newness to argue that younger new ventures that internationalize at an earlier age have several advantages of becoming faster learners in foreign markets. These advantages accrue, not only because they have fewer established routines to unlearn, but also because they are more flexible to rapidly embrace new foreign knowledge and entrepreneurial opportunities. However, not all start-ups are able to move beyond their home region, as observed in Oviatt and McDougall (1994). In fact, so many start-ups are actually born regional, not global (Lopez, Kundu, & Ciravegna, 2009). Nonetheless, Autio et al. (2000) fail to recognize the role of geographic space in understanding the benefits and costs associated with early internationalization (see Zahra, Zheng, & Yu, 2018, for a recent critique). Specifically, while Autio et al. (2000) assume that the learning advantages of newness are likely to trump the liabilities of newness and foreignness, this and other studies (Patel et al., 2018) suggest that the opposite may be the case for younger start-ups that engage in accelerated export activities in inter-regional host countries. Hence, unless scholars in the field of international entrepreneurship seek to fully utilize insights from the home regionalization perspective, they may be in danger of offering misguided advice regarding international new ventures’ learning advantages.

Second, our study directly contributes to the home regionalization literature by taking the interplay between internationalization speed and FSA reach seriously. In doing so, we redirect an ongoing conversation by demonstrating the home region-bound nature of fast-mover CMNEs’ technological and marketing resources. Despite increasing inquiries on EMNEs, relatively little is known about the speed at which, and the degree to which, their core FSAs transfer across regional borders (Luo & Tung, 2018). Our study is one of the first to empirically test whether and how fast-mover EMNEs’ technological and marketing resources matter. Focusing exclusively on CMNEs, we find that CMNEs’ technological and marketing resources that accentuate the positive effect of faster intra-regional internationalization on financial performance aggravate the negative effect of faster inter-regional internationalization on financial performance. What emerges from our empirical evidence is that on average, fast-mover CMNEs are unable to simultaneously leverage their core FSAs, captured by their technological and marketing resources, within and beyond the home region.

In our study, some (non)findings deserve more attention than others. For example, our findings reported in Model 3 in Tables 3 and 4 fail to confirm Verbeke and Kano’s (2015) view that technological and marketing resources of EMNEs do not act as a critical source of their competitive advantage. When it comes to the internationalization of CMNEs, their strong technological and marketing resources serve as an important driver of financial performance associated with faster intra-regional internationalization. However, we do not find evidence that CMNEs’ technological resources aggravate the negative effect of faster developed-market inter-regional internationalization on financial performance. Concomitantly, this nonfinding casts doubt on the notion (Guillén & García-Canal, 2009; Luo & Tung, 2007, 2018; Meyer, 2004) that the technological ability of CMNEs to catch up with their more advanced competitors depends on a combination of accelerated strategic asset-seeking FDI and absorptive capacity. Ambos, Ambos and Schlegelmilch (2006) find that knowledge transfers from foreign subsidiaries located in intra-regional host countries to European MNE headquarters are positively related to the latter’s absorptive capacity. However, these benefits may not be obtained “quasi-automatically” (Meyer, 2004: 267), especially when CMNEs, which generally lack absorptive capacity (Luo & Tung, 2018), speed up their expansion into developed-market inter-regional host countries.

Before proceeding further, we wish to raise the critical issue of whether the study of CMNEs requires a new theory or can borrow from existing theories about EMNEs or MNEs. It is our view that CMNEs can potentially extend, rather than replace, an accepted theory of MNE internationalization process in general and internationalization speed in particular, which has been applied to Swedish MNEs (Johanson & Vahlne, 1977), Dutch MNEs (Vermeulen & Barkema, 2002), Spanish MNEs (Casillas & Moreno-Menéndez, 2014), Japanese MNEs (Jiang et al., 2014), and Korean MNEs (Chang & Rhee, 2011). In fact, a home regionalization perspective can serve as a building block that helps increase the predictive power of an extant fast-mover (dis)advantage theory. Here it is instructive to revisit the original and revised Uppsala model (Johanson & Vahlne, 1977, 2009; Vahlne & Johanson, 2017). Overall, the generalizability of the model to CMNEs has been seriously questioned (Guillén & García-Canal, 2009; Luo & Tung, 2007; Mathews, 2006). As articulated earlier, however, the Uppsala model reveals context-dependent approaches to how firms adjust the pace of internationalization over geographic space (Figueira-de-Lemos et al., 2011; Johanson & Vahlne, 2009). The empirical results of our study generally validate the descriptive and prescriptive value of the Uppsala model.

We next specify the boundary conditions under which our theorizing on CMNEs is more or less likely to hold. All MNEs, regardless of their geographic origin, have to evaluate which foreign markets to enter and how rapidly to simultaneously enter them, meaning that these combinative phenomena are not CMNE-specific. Hence, our theorizing and results may extend to EMNEs and DMNEs alike, albeit to varying degrees. Nonetheless, our predictions are likely to apply more to infant MNEs (and EMNEs generally) than to mature MNEs (and DMNEs generally). For example, while many Japanese MNEs are regional MNEs due to home region boundedness of FSAs (Collinson & Rugman, 2008), some might have already developed and possessed non-location-bound FSAs through possession of technological and marketing resources (Delios & Beamish, 2005). If this is the case, a great deal of caution is warranted due to contextual differences between Chinese and Japanese MNEs in terms of their stage of evolution as MNEs. Moreover, it is worth noting that there is no reason to believe, a priori, that predictive values observed in our study apply universally to all infant EMNEs. Many CMNEs have grown into regional MNEs, with a strong focus on culturally proximate intra-regional host countries, or intra-regional host countries, where there are fewer cultural barriers (Buckley et al., 2007; Child & Rodrigues, 2005). This pattern is more similarly observed among Latin American MNEs (Aguilera, Ciravegna, Cuervo-Cazurra, & Gonzalez-Perez, 2017) than among Indian MNEs (Sauvant, Pradhan, Chatterjee, & Harley, 2010). Hence, our theorizing and results can be generalizable more to the former than to the latter.

Besides theoretical implications, our study generates practically relevant knowledge with prescriptive implications for managers. Our findings show that the choice of internationalization speed might be a tricky task because it requires a more balanced mix of fast internationalization into intra-regional host countries with slow internationalization into inter-regional host countries. Hence, top managers should consider together both the speed and geographic space of internationalization. The fast pace of internationalization is not universally desirable because it is critical to take advantage of similarities across countries within the home region in order for MNEs to be successful. Therefore, managers are advised to rapidly venture into and focus on intra-regional host countries, where they can utilize their FSAs, including technological and marketing resources. However, they should be cautious when rapidly expanding into inter-regional host countries, not only because they may have difficulty transferring home-grown FSAs successfully (Rugman & Verbeke, 2007), but also because it is challenging to address large differences across inter-regional host countries (Ghemawat, 2007). Our study shows that as MNEs swiftly expand into inter-regional host countries, these distant countries may tax their technological resources in particular. Consequently, MNEs may need to be patient when moving technological resources into distant regions.

Our study has several limitations that in turn offer opportunities for future research. First, while our study found relatively short-term, fast-mover (dis)advantages to exist, future research might examine whether longer-term fast-mover (dis)advantages actually exist. Second, future research might replicate and extend our theoretical and empirical analyses to understand whether they can generalize to MNEs headquartered in other emerging markets and even developed markets. Third, although a home regionalization perspective is based primarily on the simple dichotomy between home and host regions, future research needs to break up a broader grouping of countries into ones with distinct aspects – for example, in terms of national culture and institutions – to further the development of this perspective in important ways. Finally, while R&D and advertising intensities are the most robust measures of technological and marketing resources, the use of these archival proxies has been criticized for failing to capture the intangible nature of these resources (e.g., Ketchen, Ireland, & Baker, 2013; Meyer, 2004; Verbeke & Kano, 2015). Hence, future research may employ other measures of FSAs with higher measurement quality across regions.

In conclusion, our study provides convincing evidence that CMNEs perform well financially when rapidly expanding intra-regionally but perform poorly when rapidly expanding inter-regionally. Considering the heterogeneity of these CMNEs’ technological and marketing resources, we additionally find that these resources yield value to fast-moving firms in intra-regional hosts yet fail to yield value to inter-regional hosts. As fast internationalization is a fascinating yet poorly understood phenomenon, we urge scholars to explore additional contexts under which fast-mover MNEs may realize better financial performance.

Notes

  1. 1

    In 2018, China was home to 120 Fortune Global 500 companies – just a handful fewer than the U.S.’s 126. In 2001, only ten Chinese firms made the Fortune Global 500 list. The list complied and published annually by Fortune magazine is available at http://fortune.com/global500/.

  2. 2

    Following Buckley et al. (2007), we refer to China as the People’s Republic of China (PRC), excluding the two special autonomous regions of Hong Kong and Macau and the Republic of China (Taiwan).

  3. 3

    We are indebted to an anonymous reviewer for his or her comment on the Uppsala model’s contingent nature.

  4. 4

    We are grateful to an anonymous reviewer for bringing this issue to our attention.

  5. 5

    Available at https://www.mckinsey.com/featured-insights/innovation-and-growth/globalization-in-transition-the-future-of-trade-and-value-chains (accessed 10 April 2019).

  6. 6.

    It is important to highlight that Cohen and Levinthal (1990: 137) acknowledge the “inertia” aspect of “the cumulative quality of absorptive capacity,” for instance, through the outcome of “the not-invented-here syndrome.”

  7. 7

    Conventional wisdom holds that the more resources the firm possesses, the better the firm performance (Barney, 1991). However, research now shows that more is not always better than less, particularly when organizational environments change. For example, Miller and Shamsie (1996) find that the very qualities of resources that create value in one environmental context paradoxically also accelerate value destruction in another context, precisely because such resources become more vulnerable to changes in the environment. Le Breton-Miller and Miller (2015) refer to this phenomenon as the paradox of resource vulnerability. In a similar vein, Leonard-Barton (1992) acknowledges that core capabilities can turn into core rigidities that inhibit technological innovation and change.

  8. 8

    Available at http://www.chinadaily.com.cn/a/201612/30/WS5a2b8b8ba310eefe3e9a1192.html (accessed 26 February 2019).

  9. 9

    Available at https://asia.nikkei.com/Opinion/Chinese-companies-must-Europeanize-for-Europe (accessed 10 July 2019).

  10. 10

    CSMAR is a leading data provider focusing on Chinese companies publicly listed on major Chinese stock exchanges (see http://us.gtarsc.com/p/sq/). These data are available at Wharton Research Data Services (WRDS) (see http://www.whartonwrds.com/datasets/csmar-2/ as well as http://www.whartonwrds.com/wp-content/uploads/2011/08/GTA-CSMAR-One-Sheet.pdf).

  11. 11

    In the robustness checks, we use Tobin’s Q known as a market-based measure of financial performance that is forward-looking and risk-adjusted.

  12. 12

    Arregle et al. (2013), in their study of Japanese MNEs, split Asia into three parts: East Asia (China, Taiwan, Hong Kong, and South Korea), Northwest Asia (India and Pakistan), and South East Asia (Thailand, Singapore, Vietnam, Malaysia, Philippines, and Indonesia). However, our grouping scheme may be more meaningful in advancing a realistic understanding of CMNEs than Arregle et al.’s (2013) grouping scheme, which yields too little variance among CMNEs.

  13. 13

    Distance data were downloaded in August 2017 from https://lauder.wharton.upenn.edu/resources-publications/. However, Berry et al.’s database does not include all host countries our sample Chinese MNEs had entered during the period 2002–2014. In particular, the values of cultural distance between China and host countries are either unreported or inconsistently reported. For countries with missing values, we use historical data and the trend extrapolation method (Rhyne, 1974) to yield its values for subsequent years. If no data are available on a given host country, we assign the value of a host country closest to the focal host country. For example, the value of cultural distance between the United Arab Emirates and China is missing for the whole observation period. However, the value of Qatar’s cultural distance from China is available for the year 2010. Thus, we use that year’s value of cultural distance between Qatar and China to capture the value of cultural distance between the United Arab Emirates and China.

  14. 14

    We acknowledge other econometric techniques such as two-stage least squares and general method of moments have been used to address the endogeneity issue. However, when these techniques are deployed, sample-induced endogeneity cannot be corrected for because they have to use a restricted sample of MNEs only.

  15. 15

    When the linear and squared terms of speed of internationalization into culturally more similar intra-regional host countries are included together, their signs are positive and statistically significant (β = 0.773, p = 0.001) and negative and statistically significant (β = − 0.100, p = 0.001), respectively. However, an inflection point for the predicted inverted U-shaped relationship is 3.865 [0.773/(2 * 0.100)], which is equal to 7.649 standard deviations above the mean (7.649  * 0.47 + 0.27). Hence, although the inflection point falls within the range of our data, it is practically appropriate to conclude that a tail of this inverted-U does not exist. We believe that our conclusion echoes the view that “a clearly predicted and empirically supported effect provides more powerful evidence than a curvilinear effect motivated only by the existence of opposing effects” (Meyer, 2009: 190). The results are available upon request.

  16. 16

    We thank an anonymous reviewer for bringing this issue to our attention.