Footloose multinationals: Extending the internalization theory

Correspondence Irina Gokh, Leicester Castle Business School, De Montfort University, Leicester LE1 9BH, UK. Email: irina.gokh@dmu.ac.uk Abstract Internalization theory explains how the multinational enterprise (MNE) makes decisions on new investments but does not explicitly address the restructuring (i.e., unbundling) processes that can take place within the MNE network of subsidiaries. Further extensions to internalization theory have considered the ways MNEs recombine, bundle, unbundle and orchestrate resources but have yet to address specifically the issue of divestments and relocations. We contribute to the literature by extending internalization theory through conceptualizing footloose behavior, a repeated relocation of divested operations over a period of time, and approaching MNEs behavior in a holistic way. Empirically, we use four case studies of mature, complex and diversified MNEs within the context of European Union (EU). These MNEs have engaged in several investments, divestments and relocations over several years. Through the analysis of their behavior we demonstrate how the inclusion of divestments and relocations can extend internalization theory in explaining the modern MNE.


| INTRODUCTION
Internalization theory of the multinational enterprise (MNE) has remained one of the most influential theories explaining the existence, evolution and strategic management of the MNEs (Buckley, 2016;Casson, 2015). Over the last 40 years, since its original introduction, the theory has evolved to remain current with changes in the external environment as well as the evolution of the modern MNE. Fundamentally, internalization theory builds on the work of Coase (1937) exploring market imperfections and suggesting that some transactions are more efficiently performed internally rather than externally. Initially, internalization theory aimed to explain why Foreign Direct Investment (FDI) was focused on knowledge-intensive industries (Casson, 2014) and offered explanations on the choice of the entry mode between market and non-market modes of operation.
Internalization theory suggests that the benefits of internalizing the market should outweigh the costs (Buckley, 2014;Buckley & Casson, 1976;Casson, 2014), i.e., "the advantages of internalizing a market are the obverse of outsourcing" (Buckley, 2014, p. 229). Casson (2014) argues that international business (IB) literature extensively covers the benefits of internalizing the market, but almost ignores the costs, which increase significantly with the size and international expansion of the firm.
According to Buckley and Casson (2019) MNEs can demonstrate high levels of versatility. This versatility is reflected in changes in the structure of the MNE through new investments and divestments as the MNE requires a consolidation of resources before expanding further.
Through the work of Rugman and Verbeke (2003) and Hennart (2009) internalization theory has been developed to add a dynamic dimension. In search of the optimal boundary, multinational due to different firm-specific advantages (FSAs) (e.g., new technological developments) and country-specific advantages (CSAs) (e.g., European integration). The interaction of FSA/CSA governs the "boundary" (i.e., the scale) of the firm. Hence, when these factors change, the cost/benefits balance may also shift, and consequently, the boundary of the firm may be pushed back (i.e., the firm will reduce its size). Activities that no longer contribute to the firms' strategic goals become valueless "empty calories" (IBM, 2014, p. 3) and should be divested if the firm wants to avoid potential failure. Buckley (2016) argues that a key avenue of extending internalization theory is through the dynamic nature of the theory. It is through adding and explaining divestments and relocations where the present study aims at extending internalization theory further. Divestments and relocations enable firms to rationalize their size once they have exceeded their optimum size (Casson, 2014) and to reallocate resources to more promising, or new, product lines and markets.
To extend internalization theory, we build on the work of Rugman andVerbeke (1992, 2003), Rugman (2010), Narula and Verbeke (2015), Hennart (2009) and Pitelis and Teece (2018). This study complements prior studies by Buckley and Casson (2019), Strange and Humphrey (2019) and Gaur, Pattnaik, Singh, and Lee (2019) published in a Journal of International Business Studies special issue on extending internalization theory. We use resource recombination, bundling and unbundling and orchestration as ways to extend internalization theory and conceptualize footloose behavior, as a repeated relocation of divested operations over a period of time. Through an orchestration approach (Pitelis & Teece, 2018) the MNE decides to drop out old and obsolete assets and bring in new assets to facilitate its functioning. Asset orchestration can span across firm boundaries through a variety of collaborative modes.
First, we approach internalization theory from a strategic perspective and build on arguments developed by Rugman andVerbeke (1992, 2003) to argue that footloose behavior, a repeated relocation of divested operations over a period of time can be seen as a process of constant bundling (i.e., investment), unbundling (i.e., divestment) and re-bundling (i.e., relocation) of resources to balance the optimal size and growth of the MNE. We argue that this process depends both on the nature of CSAs and their changes but primarily on the type of FSAs used, that is, location-bound versus non-location bound FSAs.
According to Narula and Verbeke (2015), FSAs sustainability comes through a constant search for locational characteristics that can lead to the creation of non-location bounded FSAs. For complex MNEs linkages between a variety of transactions can lead to differential outcomes in the investment/divestment behavior of the MNE (Narula & Verbeke, 2015).
We extend this argument by including Hennart's (2009) approach and looking at the cost and availability of accessing CSAs. We take this approach a step further and argue that MNEs might not only opt for lower control modes of operations, as argued by Hennart (2009) but completely divest should their FSAs be not strong enough.
In addition to the earlier extension, we integrate the investment motives proposed by Dunning (1993), Dunning and Lundan (2008) with the work of Rugman (2010) to explain how certain types of investment motives, such as resource, market and efficiency-seeking lead to higher levels of footloose activity when compared with strategic asset-seeking.
Despite the conceptual nature of the article, we offer an empirical discussion with some practical examples through a case study approach. Building on information from FDI Markets and Eurofound we have identified four MNEs. We use these MNEs as a basis for our discussion and practical examples. All four MNEs have had a number of divestments and investments as a response to changing CSAs and FSAs configurations. Our discussion demonstrates empirically how our extension of internalization theory can explain the modern MNE and its behavior holistically, accounting not only for new investments but also divestments and relocations.
The rest of the article is structured as follows: Next section provides our conceptual framework in extending internalization theory and developing two propositions. The following section offers a discussion and practical examples on the application of our proposed extension. Finally, the last section of the article concludes and discusses limitations as well as future research.

| Definition of footloose behavior
We define footloose behavior as a repeated relocation of divested operations over a period of time. "Repeated" refers to the systematic character of divestments and relocations. A one-time relocation can refer to a random event. "Divestment" refers to a cessation of all or of a major part of existing active operations that reduces the presence in the foreign market (Belderbos & Zou, 2006;Boddewyn, 1979). We subdivide divestments into three groups: (1) exit from the market; (2) closure of a subsidiary; and (3) organizational restructuring. Divestment is termed as "relocation" when terminated activities in the subsidiary are relocated to another country: (1) by establishing a new subsidiary, and (2) by increasing a market scope, product scope or value-added scope of an existing subsidiary (White & Poynter, 1984).
Despite efforts to discuss divestment from a strategic perspective, studies by Boddewyn (1983) and Berry (2010) are isolated efforts. Boddewyn (1983) offers a discussion of the factors that drive divestments, summarizing them in return on assets, strategic and behavioral and resource allocation. Following the same line of argument, Berry (2010) argues that growth and expansion involve divestment and investment. Thus far, only a limited number of studies have addressed the issue of footloose behavior (Cowling & Sugden, 1999;Flamm, 1984;Görg & Strobl, 2003;Van Beveren, 2007). The focus of these studies is not on the MNEs' strategic behavior but on the negative impact footloose behavior has on the host country's economy.
This strand of the literature emphasizes that footloose behavior emerges only due to the adverse shocks in the host country's environment that bring changes to the CSAs; thus, these studies largely ignore the characteristics of the firm, in the form of FSAs, which may also drive footloose behavior.

| The combination of CSAs and FSAs and footloose behavior
According to Rugman and Verbeke (1992) internalization of activities reflects the market failure. This market failure exists, in the vast majority of cases, due to natural or government-created market imperfections. The MNE owns internationally transferable FSAs (nonlocation bound) and non-transferable FSAs (location bound). The first ones represent technological, marketing or administrative (governance-related) knowledge that allows the MNE to reduce the costs of doing business abroad. This knowledge can either be codifiable which can be transferred cheaply, but it is also easy to be copied and replicated or tacit with more challenges when transferring to the host location but more difficult to imitate from competitors. The value of these non-location bound FSAs could differ substantially from country to country depending on the goals of the MNE. The second set of FSAs represent access to a network of standalone resources (e.g., a network of distributors), local marketing knowledge or reputational resources, local best practices (e.g., routines) and some domestic recombination capability of resources. These FSAs can only be created by linking with existing third parties operating in host markets.
As the MNE enters new locations, it is important to develop new location-bound FSAs in the host country. These will act in a complementary way to the non-location bound FSAs and will enable the firm to maximize the utility from locational characteristics. When seen in the context of internalization theory, Rugman and Verbeke (1992) argue that the core FSAs are non-location bound and originate from the parent company, whilst CSAs are exogenous, cannot be influenced directly by the MNE and therefore can be used in a local and static sense. It, therefore, depends on the MNE whether resources could complement CSAs in a way that can generate long-term FSAs. It is also worth pointing out that internalization advantages will depend on a MNE's transactional FSAs to operate foreign subsidiaries. The MNE will create value through the recombination of resources. This recombination is, at the same time, a key driver but also a constraint and can take a variety of forms. Rugman's view (Narula & Verbeke, 2015) of resource recombination can lead not only to new investments but also to the change of role in existing subsidiaries as well as divestments.
According to Verbeke and Kano (2016), there are four combination processes: fast bundling, principles driven bundling, adaptive bundling, and entrepreneurial resource orchestration. These processes are from lower to higher-order, and as they approach the entrepreneurial resource orchestration they require entrepreneurial skills, slack resources and willingness and capacity to let go of existing resources in order to substitute them with higher value-creating potential. This leads to the creation of the highest order FSA for an MNE which is the recombination capability. The Rugman and Verbeke (1992) approach to internalization theory assumes that the MNE can access CSAs at no cost and can recombine those seamlessly across national boundaries. Responding to this approach, Hennart (2009) has developed a framework showing that this assumption of cost-free bundling between FSAs and CSAs is not always correct. There might be cases where local CSAs are controlled by actors and therefore accessing them does not come without a cost. Hennart's (2009) approach supports the evolution of an MNE's engagement in a location and shows that the MNE can move between equity and non-equity modes. We take this argument a step further and argue that when this cost of accessing CSAs is high or reduced in alternative locations, the MNE might decide to change the subsidiary's mandate or even divest.
To optimize the internalization in a growing firm, MNEs may start to relocate existing operations, which involve the movement of nonlocation bound FSAs from one country to another, leading to an increase of the efficiency of the network (i.e., when benefits of internalizing activities outweigh the costs). Internalization theory assumes a linear process of increasing engagement in the host market as the MNE gains experience and therefore, faces reduced transaction costs. Building on our previous argument, this is not always linked with a switch to higher control modes. We argue that it is the strength of the FSAs that will determine whether the MNE will decide to remain in the host market even through a lower control mode or the decision will be made to relocate/divest its assets. We also argue that higher levels of nonlocation bound FSAs will lead to footloose behavior. The relocation and recombination capability of resources is associated with the necessity to rationalize the portfolio of existing operations in such a way that it will increase the efficiency gains of the differentiated MNE network.
Based on the earlier, we put forward our first proposition:

Proposition 1. The insufficient bundling of non-location bound FSAs
with CSAs, reduces the ability of the MNE to generate strong location-bound FSAs and increases the footloose behavior.

| CSA/FSA, investment motives, and footloose behavior
According to Rugman (2010), the CSA/FSA framework can be used to complement Dunning's (1993), Dunning and Lundan's (2008) motivations of multinationals. FDI takes place in those cases where FSAs are combined with high CSAs. According to Rugman (2010), three types of motives, that is, resource-seeking, market seeking, and efficiencyseeking emerge in cases where the MNE combines a high level of CSAs with a low level of FSAs. These motives are reflected in the operations of the subsidiary in the host location. According to Gaur et al. (2019) internalization theory should focus on the MNE's subsidiary to provide reasonable explanations to the modern MNE's evolution.
In resource-seeking motivation, the MNE invests because the location can offer a number of CSAs such as an abundance of natural resources and suitable infrastructure to transport these resources (Castiglione, Gorbunova, Infante, & Smirnova, 2012;Wilson & Baack, 2012). In the case of the natural resource-seeking motivation, footloose behavior is unlikely to occur unless a firm can source these resources somewhere else (Benito, 2015). In resource-seeking behavior associated with access to low-cost resources, footloose behavior might emerge if these resources (usually the cost of the labor force) become relatively expensive in comparison to other locations. Economic development, for example, can lead to an increase in input costs. In marketseeking investments, key CSAs are the size of the market, its geographical proximity to other markets, the level of economic development and the availability of skilled, but reasonably priced labor force (Benito, 2015;Castiglione et al., 2012;Ramasamy & Yeung, 2010;Wilson & Baack, 2012). Market size and opportunities for future growth are crucial (Ramasamy & Yeung, 2010), but "if there is a trade-off between market size over the market growth, manufacturing investors would prefer market size" (Ramasamy & Yeung, 2010, p. 590). In efficiency-seeking investments, CSAs mostly reflect the potential incentives a location can offer to the firm. For example, government involvement in removing trade restrictions (Wilson & Baack, 2012), facilitation of infrastructure development (Benito, 2015;Wilson & Baack, 2012), development in the area of human resources (Wilson & Baack, 2012), more favorable business environment such as government incentives, country's economic and political stability are key CSAs. It is worth highlighting though that government actions such as removing trade barriers or advancing regional integration do not attract the FDI on their own because open borders do not imply that locations can offer other complementary factors such as strong local capabilities (Loll and Narula, 2004). A key CSA is also the reasonable price of labor (Ramasamy & Yeung, 2010) paired with reasonable purchasing power (Castiglione et al., 2012). Divestment can also be complemented by outsourcing when an MNE switches from internal production to production through an external partner. This switch is common in manufacturing, an activity that is predominantly driven by the efficiency-seeking motive (Strange & Humphrey, 2019).
The earlier CSAs are complemented with location bound FSAs. Gaur et al. (2019) argue that the host location environment plays a significant role in the way MNEs leverage their FSAs. In their study, they find that MNEs tend to leverage their FSAs better in countries with weaker institutional environments (Gaur et al., 2019). In resource-seeking FDI and specifically natural resource-seeking, investments tend to flow into developing countries where the rule of law tends to be weak; thus, internalization helps the MNE to secure the stability of production output in highly volatile environments. In the case of market-seeking, FSAs such as property rights, that is, brand name, trademark or an innovative product (Benito, 2015) are crucial.
Within market-seeking motivation, we need to highlight the importance of local adaptation. Local adaptation could require additional financial inflows or specific technology and knowledge. In the case of efficiency-seeking investments, the key FSA is the firm's ability to reorganize the network. Investments are performed to build better (more efficient) production facilities and consequently locate the production facilities in a way that increases the overall efficiency (specialization or clustering) of the MNE (e.g., more efficient logistics) (Benito, 2015). A core FSA used in efficiency-seeking driven investments is the firm's ability to maximize the benefits of activities under common governance. Common governance can have two aims: to coordinate different activities in different parts of the world and to coordinate the same activity in diverse environments.
Strategic asset-seeking, on the other hand, results from a more advanced combination of CSAs with FSAs. In this case, high-level CSAs are combined with a high level of FSAs, in the vast majority of nonlocation bound FSAs. Strategic asset-seeking is a forward-looking investment motivation, and "it is about developing new resources and capabilities that can generate future streams of revenue, not exploiting already existing ones" (Benito, 2015, p. 6). From a CSA perspective, the location should offer a number of factors to be attractive. High levels of intellectual property protection, reduced volatility of economic conditions and well-developed infrastructure and a highly skilled labor force are amongst them. These CSAs are linked with high-level FSAs such as R&D, customer service and innovation (Wilson & Baack, 2012). Considering that footloose behavior is a systematic activity (repeated relocations) we argue that in the case of strategic asset-seeking investments, the nature of such investments (i.e., long-term investment) makes them less prone to footloose behavior. As Benito (2015, p. 8) argues: "Companies that venture abroad for strategic asset-seeking motivations are likely to prioritise control over their foreign operations, perhaps overriding other relevant concerns". On the basis of the earlier arguments, we put forward our second proposition: Proposition 2. Resource-seeking, Market-seeking and Efficiency-seeking investments will be more sensitive to changes in the CSAs and FSAs combinations than Strategic asset-seeking investments and will lead to more frequent footloose behavior by the MNE.

| DISCUSSION AND PRACTICAL EXAMPLES
In order to provide some empirical support to our conceptual arguments we decided to explore footloose behavior within the European Union (EU) context. Two reasons can be used to support our decision to focus on the EU, one conceptual and one empirical. Conceptually, the EU has, over the last 30 years, gone through a transformational process of integration, accession of new member states and introduction of common policies and regulations. It offers a stable economic and political environment which also reflects a degree of change that we cannot find in any other regional integration group of developed countries (Barrell & Pain, 1999;Filippaios & Papanastassiou, 2008).

| IBM and the case of Ireland
International Business Machines (IBM) was founded in 1911 (Marketline, 2017) and has gone through a lot of transformations in its 106 years of existence. The company has managed to survive in a changing environment and remains a leader in the Software and IT industry. IBM is a large, mature and diversified MNE. We argue that IBM has dealt a few times with the optimum network size reflecting the boundaries of the firm set by internalization. According to Casson (2014) when a firm reaches its size limit (which is its margin), it either breaks down and fails, or the firm will make rationalization decisions that involve a careful exit strategy from locations that no longer serve the company's goals. We extend this argument by suggesting that rationalization of activities is not a one-time event; it is a systematic activity (i.e., footloose behavior). We argue that IBM constantly adopts a footloose behavior (i.e., divestment of "empty calories" in the company's own language), as it is the only way to sustain efficiency and remain competitive.
An example that supports the earlier argument is IBM's divestment from Ireland (2009) and subsequent relocation to Singapore with a market-seeking purpose. IBM has a long history to its presence in Ireland; however, the market in Ireland, and its relatively small size, does not allow IBM to take advantage of expansion opportunities. Therefore, the company decided to relocate to Singapore. This divestment comes shortly after an investment in Ireland, in 2007, because it was an "ideal" location with a welcoming environment reflecting rapid economic growth and high technological activity. The 2007 investment captures an efficiency-seeking investment motive (i.e., the location has some attractive features and fits MNE efficiency goals) but in the years that followed other locations emerged (e.g., Singapore) that offered a better strategic

| Johnson controls in Portugal and Spain
Johnson Controls (JC) was established in 1885, and has, since, remained a leading company in the automotive components industry operating in more than 150 countries (Johnson Controls, 2017). In 2006, JC divested two manufacturing plants in Portugal due to the need to "adjust the output capacity in Europe" (Eurofound, 2017). This may suggest that JC overproduced in Europe and they had to restore efficiency in their operations. They decided to relocate Portuguese activities to Germany and Slovenia.

| Procter & Gamble and a Pan-European restructuring
Procter & Gamble (P&G) is a multinational company and a manufacturer of consumer packaged products of 50 (Reingold, 2016)  The firm through its investment and divestment/relocation behavior adjusts to changes in the external environment. However, this adjustment is a reactionist behavior (i.e., the firm has no control over it and must adapt/react to survive). In the case of Portugal, the CSAs are no longer aligned to JC's unchanged FSAs and this leads to a divestment. The same argument holds for Spain, but here there is an additional element-The new presence of JC's client (Renault). This reflects a change to the Spanish CSAs and consequently to an investment back to Spain in order to support the client. leadership brands. It was founded in 1837 (P&G, 2017) and has grown brand-wise during its 180 years of operation. Some of its products require local adaptation due to different consumer preferences. P&G faces a competing need for global integration and local adaptation of products. Addressing this need requires the investment of substantial resources.

Procter & Gamble
The company decided to discontinue 116 brands over a period of 5 years (2012)(2013)(2014)(2015)(2016)(2017). We argue that this is a reflection of efficiencyseeking motive driving Pg.'s restructuring activities and investment patterns. P&G mercilessly divests assets that prevent development and affect the efficiency balance. P&G does not only divest activities but also restructures the MNE network, creating tighter links between subsidiaries and focusing towards increasing the overall efficiency.
Yannis Skoufalos, global product supply officer, stated that he is "reworking the [supply chain] system as if he is building it from scratcha monumental task for a company of P&G's size" (Reingold, 2016, p. 180

| Electrolux and the move to Poland
Electrolux belongs to the household appliance segment of the consumer electronics industry. Electrolux was founded in 1901 (Marketline, 2016)

| Summary of the four cases
The earlier four cases relate to large, mature, and diversified MNEs that have operations spanning across several countries and with different subsidiaries' mandates. In all cases we have seen that the evolution of the multinational network is a series of new investments, divestments, relocations and re-investments or changes in subsidiaries mandates. These case studies suggest that the footprint of the modern MNE does not stay still and is subject to constant change through new investments, changes in subsidiaries mandates but also relocations and closures. Our extension of internalization theory discussed in the conceptual section provides a useful tool in understanding the behavior of the MNE from a holistic approach.

| CONCLUSION
The main aim of this article is to extend internalization theory by explaining not only investments but also divestment and relocation activity. The literature does not always make a distinction between the expansion investment and relocation while evaluating the determinants of the "investment" (Sleuwaegen & Pennings, 2006). This logic limits our understanding of both processes because the investment exemplifies the expansion related activity, but relocation exemplifies the move of production facilities from one country to another (Mucchielli & Saucier, 1997). In this article, we explore a new way of investigating the MNE's expansion activity through the lens of systematic "repeated relocations," which we approach as footloose behavior. Instrumental in our extension of internalization theory is the work of Rugman and Verbeke (1992)  From a practical perspective, our study clarifies how to manage large, mature MNEs and how to keep an optimal balance between the size of the firm and future growth, that is, how to manage a large organization and maintain efficiency while responding to changes in CSAs and FSAs. Casson (2014) argues that even those who are closely involved do not usually foresee failures of the firms. Thus, it is crucial for managers to embrace and understand the complexity of the multinational enterprise, and to understand that the "balance and the boundaries of firms are subject, as always, to conflicting pressures" (Buckley, 2014, p. 231 This has been the case of Ireland and IBM. IBM has transferred low value-added activities to Singapore and China whilst upgrading its activities in Ireland. This has been clearly a beneficial activity for the Irish economy. MNEs, therefore, through footloose behavior and the change in subsidiaries' mandates could create positive effects for host locations. Further research is necessary to understand the exact conditions where an MNE can bring positive effects to the host location through footloose behavior.
In conclusion, for the MNE itself, footloose behavior can be a matter of survival, and this should be acknowledged in future research. Lessons can be drawn that not all investments will be "new investments," as some may be simply "relocation investments" and could follow very different investment motivations.