Home country (in)stability and the locational portfolio construction of emerging market multinational enterprises

Emerging markets often experience instability due to rapid changes to the institutional environment, social changes like rapid urbanization, or even unrest. We argue that emerging market multinationals (EMNEs) manage such instability by constructing and changing locational portfolios, and qualitatively analyze six cases in South Africa over a period that included the entrenchment of Apartheid, increasing resistance to it, the immediate post-Apartheid era, and finally the period of state capture. The four periods of (in)stability – initial tenuous stability, extreme instability, comprehensive stability, and finally growing instability – differently affected EMNEs ’ location choices. EMNEs went to proximate developing countries when the home country was relatively stable, but left for host countries in the developed world once the home country became unstable. Few EMNEs capi-talized on their experience there once home-country stability returned, instead returning to emerging markets. These patterns are best explained by a portfolio logic that takes into account home-country environmental dynamism.


Introduction
Most of the operations of emerging market multinational enterprises (EMNEs), as indeed for MNEs generally, take place in their home country (Banalieva & Santoro, 2009;Rugman, 2005). What are the implications of dramatic and sudden changes in the stability of EMNEs' home countries for the host locations in which they operate? This paper provides evidence that EMNEs' internationalization is affected by the (in)stability in their home country. EMNEs use a portfolio logic in making location choices, opting for 'safe', stable host locations when the home country is unstable, and for riskier, less stable host locations when the home country is stable. This finding addresses a gap in existing scholarship and suggests at least two novel contributions.
First, we contribute to the growing body of literature on how dynamic changes in the macro-environment shape firm behavior, especially in emerging markets (Banalieva, Cuervo-Cazurra & Sarathy, 2018;Klein, Puck & Weiss, 2019). Our evidence suggests that the stability or not of home country conditions contributes to EMNEs' location choices and we explain why that may be the case.
The second novel finding relates to EMNEs' use of a portfolio logic in their location choices. Our case studies demonstrate that given the outsized importance of home country operations, EMNEs considered conditions in both home and host countries to construct a portfolio of locations with what they deemed an acceptable overall level of stability. They internationalized to more stable developed countries while their home country was suffering from extreme instability, even though home county instability makes it harder for EMNEs to develop the capabilities and resources to internationalize to such locations (Wu & Chen, 2014). As their home country stabilized, however, EMNEs expanded to less stable developing countries, quite often exiting developed countries (and all their experience and potentially learnings there) to secure the resources to do so. Thus, the relative (in)stability of the home country was important in explaining whether EMNEs sought out host locations in developed or in emerging countries.
Conceptualizing the MNE as a portfolio of operations across different locations is not new to international business. Aliber argued that the MNE can be conceptualized as a "special form of international portfolio investment " (1970:17), with scholars like Rugman (1976;1979) and Kogut (1983) expanding those ideas. Recently, a portfolio perspective has regained prominence as an explanation for the internationalization of MNEs both from developed (Zhao, Parente, Song & Wenger, 2020) and emerging markets (Huang, Xie & Wu, 2021). Yet the focus is typically on the "portfolio of overseas affiliates" (our emphasis, Belderbos, Tong & Wu, 2020:1206. We contribute to that body of work by showing how the conditions in the home country are central in shaping the overall locational portfolio. We focus on home country instability because one of the defining characteristics of emerging markets is instability (Zhao, Park & Zhou, 2014). " [F]undamental and comprehensive institutional transformations" (Ahlstrom & Bruton, 2006:299) are not only argued to be characteristic of emerging markets, but as Rottig (2016:7) points out, such change "is typically more sudden and unpredictable and so difficult for MNCs to manage" in emerging than in developed markets. Moreover, there is evidence that such instability presents challenges to firms not only when it is negative, but even when it represents positive reforms (Banalieva, Cuervo-Cazurra & Sarathy, 2018;Barnard & Luiz, 2018;Perez-Batres & Eden, 2008).
Our study covers firms in South Africa from 1956 to the present day. South Africa is a useful setting for our work, because of dramatic, welldocumented changes in social (in)stability in the country. We identify four periods with differing levels of instability. 1956 to about 1975 is characterized by tenuous stability. Although incidents of instability occurred (e.g. the 1961 Sharpeville massacre), they were effectively contained soon after breaking out. In contrast, the period from 1976 to 1990 (when Mandela was released from jail) was characterized by extreme instability. The decade post-Apartheid was the most stable on record, but since 2004 there has been growing instability, associated with the slow pace of transformation and crisis of state capture (Luiz, 2016).
We examine six cases that represent about 20% of market capitalization on the Johannesburg Stock Exchange (JSE), framing our analysis of the EMNEs' location choices with an assessment of the levels of domestic instability and patterns of outward foreign direct investment (OFDI). Location choices are very different over the four periods. At first, the EMNEs internationalized to close neighbors, consistent with Johanson and Vahlne's (1977) explanation. During extreme instability, however, the EMNEs sought out the developed world with its more stable institutions. However, most of the cases in our study abandoned locations in the developed world when the home country stabilized, instead focusing on wider Africa and other emerging markets. In the final period of growing instability, the emphasis is on focusing their international portfolios.
We suggest propositions to explain these trajectories, highlighting the need to consider the level of (in)stability in the home country as an important consideration shaping the EMNE's location choices and overall locational portfolio. To the extent that home country operations constitute a substantial part of EMNE operations, the EMNE will internationalize to more stable and typically developed country locations when home country conditions are unstable. During periods of home country stability EMNEs will often exit developed countries and all their learnings there to seek out riskier higher growth (and more proximate and/or profitable) emerging market settings. This finding contributes an important new insight about EMNEs' location choices and locational portfolios. It suggests that managing instability, whether termed "institutional fragility" (Shi, Sun, Yan & Zhu, 2017), "institutional turbulence" (Luiz, Magada, & Mukumbuzi, 2021), or more positively "institutional dynamics" (Chen, Cui, Li & Rolfe, 2017) is a feature of EMNEs that deserves more theoretical attention.
Our work has important implications for the evolution of EMNEs. The often dramatic changes in home country conditions require of them to respond in ways that MNEs from advanced economies do not need to do. Our work suggests that EMNEs do not only develop capabilities, but how they view locations also evolves. In particular, the riskiness of EMNEs' home (and many host) locations results in them conceptualizing locations in portfolio terms. Although thinking in terms of a portfolio of locations was initially an integral part of international business scholarship (Aliber, 1970, Rugman, 1976, the theory later faded from view. However, the requirement for EMNEs to hedge risk as a key component of managing performance shows the value of revisiting and further extending this body of work.
Another important theme in the literature is how the conditions in the home country shape internationalization. Again, the theme of capabilities recurs, because the capabilities for internationalization are developed not only abroad, but also at home, whether as a spillover from inward direct investment (Li, Li & Shapiro, 2012), as the results of deliberate domestic efforts to learn and improve the flexibility of their resources (Li, Yi & Cui, 2017) or because of learning opportunities allowed by the diversity of sub-national institutions (Chen et al., 2015;Pattnaik, Singh & Gaur, 2021).
To provide a stylized and simplified summary of the current consensus on the state of the art on EMNEs and their location choices: EMNEs are not 'exempt' from the challenges of operating in other emerging markets, but their experiences in their own less developed home countries tend to result in them having capabilities that allow them to succeed there. When they locate in developed markets, EMNEs often need to upgrade their capabilities, and indeed, learning and capability upgrading are key motives for going to those advanced markets. EMNEs' home country conditions are again relevant. On balance it seems as if supportive home country conditions make it somewhat easier to internationalize to developed countries, whereas problematic home country conditions make it less easy but more desirable to go to those locations. The importance of capabilitiesactual or desiredin extant scholarship is clear.
It is also worth pointing out that scholars as a rule presume country conditions in emerging markets to be quite static and stable, albeit dysfunctional in various ways. Yet it does happen that conditions dramatically change. For example, although Chile from the early 1990s onwards was quite stable, from 2019 to 2021 it suddenly experienced the so-called "Estallido Social", a range of severe protests. Scholars have increasingly been considering how those dynamics affect firm behavior. To better understand how, we next discuss instability in emerging markets.

Instability in emerging markets
One of the defining characteristics of emerging markets is their instability (Ahlstrom & Bruton, 2006;Zhao, Park & Zhou, 2014). Institutional instability has been defined as the extent to which an institutional environment changes rapidly over time (Li, Poppo & Zhou, 2008;Wu & Chen, 2014). Such instability has multiple manifestations and also multiple origins, with inadequate and transforming institutions one key source (Adams & Luiz, 2022). Institutions determine the parameters within which social and economic activities occur (North, 1990). One of the key purposes of formal institutions is to safeguard the investments of firms in a given location (Cavusgil et al., 2020;Henisz, 2000). Especially in emerging markets, institutions do not always fulfil that function, with Khanna and Palepu (2010) speaking of "institutional voids" which they describe as missing or inefficient market-supporting institutions necessary to complete transactions in an economy. It is to remedy such gaps that countries undertake reforms.
Reforms introduce "institutional friction" (Kim, Kim & Hoskisson, 2010) or "institutional fragility" (Shi, Sun, Yan & Zhu, 2017). Both latter terms refer to the challenges experienced when newly introduced formal reforms create tension in the existing institutional context. Reforms are typically planned and intended to improve the institutional environment, yet the resulting uncertainty and instability challenge firms. Chari and Banalieva (2015), using India as their empirical setting, explain that "monitoring vacuums" emerge during periods of reforms when previous monitoring systems (e.g. socialist governance) are no longer valid, but the newly introduced market-based monitoring systems have not yet taken hold.
Perez-Batres and Eden (2008) coin the term "liability of localness" to refer to the disconnect firms experience while navigating the "then" and "now" of reforms. Although pro-market reforms in emerging markets have been found to eventually support firm learning and thus performance (Cuervo-Cazurra & Dau, 2009), firms do better when reform speeds are synchronized (Banalieva, 2014) and when they are slower (Banalieva, Eddleston & Zellweger, 2015). In both cases, the core explanation revolves around the benefits of limiting the instability that firms need to manage. The more predictable the situation, the better firms perform. However, as Rottig (2016:7) points out, the institutional changes faced by firms are "typically more sudden and unpredictable and so difficult for MNCs to manage" in emerging than in developed markets.
There are also unplanned sources of instability in emerging markets. The hoped-for growth of emerging markets stems from their young population, expanding workforce, and rapid urbanization (Cavusgil, 2021), but all of these also carry with them both the need for social change and potentially also social upheaval. Moreover, as Marquis and Raynard (2015) point out, governments in emerging markets are often vulnerable to external conflicts, coups and internal tensions, further increasing the possibility of instability. Austin, Dávila and Jones (2017) elaborate: Political instability, expropriation, violence, and extreme macroeconomic fluctuations-often a function of dependence on exports of primary commodities and abrupt policy reversals-have been the norm rather than the exception in the modern history of Africa, Asia, and Latin America.
Grosse in 2006 commented that "perhaps there is a possibility that the old saying [Brazil is the country of the future, and always will be] may be contradicted in the new century" (2006:183). Fifteen years later, Brazil is still characterized not by the realization of its potential, but by uncertainty. Brazil is far from alone: Cavusgil (2021) highlights that many emerging markets have encountered "speedbumps" they could not overcome. He specifically emphasizes the poverty, hyperinflation, distorted distribution of income, government oppression, and generally chaotic conditions (Cavusgil, 2021) that characterize these emerging markets.
These are the conditions characterizing the home countries of MNEs from countries like Brazil, South Africa and Turkeyall well-recognized source countries of EMNEs. In making sense of how EMNEs respond to not only the institutional weakness of their home countries, but specifically the instability that they need to deal with, we draw on portfolio thinking.

Portfolio logic in an emerging market context
Early international business scholars often used a portfolio logic to explain the existence of MNEs: MNE were argued to exist because they could minimize the risk of a given location and achieve greater stability in earnings if they diversified operations across different locations (Aliber, 1970;Rugman, 1976;1979). Although other explanations for the existence of the MNE, e.g. internalization (Buckley & Casson, 1976) and transaction cost economics (Hennart, 1982) soon became more prevalent, portfolio-type explanations have recently again grown in importance. Scholars increasingly highlight how the MNE's presence across different locations can help it deal with situations of downside risk, including war and terrorism (Belderbos, Tong & Wu, 2014;Dai, Beamish & Eden, 2017;Li, Tong, Xiao & Zhang, 2021). There is also increasing work on the returns that MNEs can realize from a more diverse portfolio of locations (Jiménez, Luis-Rico & Benito-Osorio, 2014;Zhao, Parente, Song & Wenger, 2020).
In spite of the central role of risk in emerging markets, research conceptualizing the actions of the EMNE in terms of portfolios is only starting. Huang, Xie and Wu (2021) find that EMNEs stand to benefit not only from a portfolio with higher quality institutional environments, but also, and specifically over the longer term, from a portfolio characterized by a greater diversity of institutional environments. In other words, having a presence in a range of locations, both developed and developing, is of value to EMNEs. But this situation is complicated by the nontrivial possibility of EMNEs experiencing instability in their home country. As Rugman (2005) pointed out, most of the operations of MNEs take place in their home country. When there are sudden institutional changes and resulting instability, EMNEs' locational portfolio can easily become unbalanced with a very large proportion of operations located in a highly unstable home country. To counter the risks associated with instability and support firm growth, the locational portfolio may need to be adjusted.
Both risk diversification and real options theories use the notion of portfolios to explain how firms deal with risk. Real options theory is argued to involve more active firm intervention than risk diversification (where only occasional adjustments are made), and real options theory also emphasizes production rather than revenues, typically the focus of risk diversification theory (Chi, Li, Trigeorgis & Tsekrekos;Ioulianou, Leiblein & Trigeorgis, 2021). However, not only is there evidence that both matter in realizing returns (Belderbos, Tong & Wu, 2020;Yeo & Lee, 2020), but real options in emerging markets may not function as in developed markets.
For example, contrary to the expectations of real options theory, EMNEs tend to increase rather than reduce their commitment in risky markets (Yoon, Peillex & Buckley, 2021). In addition, it is known that an environment less obscured by ambiguity and greater managerial ability both increase the value of options (Driouchi, Chen, Lyu, Bennett & Sol, 2021). But not only do emerging markets tend to be quite informationpoor and thus ambiguous, EMNEs also tend to have a more limited capability base, including managerial capability, than their counterparts from developed regions (Luiz, Stringfellow & Jefthas, 2017;Ramachandran & Pant, 2010). Recognizing that much more needs to be done to tease out how real options theory functions relative to risk diversification theory, we nonetheless argue that a portfolio logic is more central in understanding EMNEs location choices than has hitherto been recognized.
To consider the location choices of EMNEs through the lens of portfolio thinking, Fig. 1 compares the horizon of host locations to which a firm from a stable versus an unstable home country can consider internationalizing if a portfolio logic is applied.
When the home country is stable, a balanced portfolio will typically include some host country locations that are institutionally stable but also some that are less stable but typically with greater growth potential (Witt & Lewin, 2007). Indeed, most papers on locational portfolios (e.g. Belderbos, Tong & Wu, 2020;Ioulianou, Leiblein & Trigeorgis, 2021;Zhao, Parente, Song & Wenger, 2020) examine a portfolio of host locations with differing institutional characteristics from the vantage point of MNEs from developed economies with their quite stable home country conditions. That body of work, as well as the limited work on EMNEs' locational portfolio construction (Huang, Xie & Wu, 2021) both confirm the central premise of portfolio thinking: There is value in having a diverse portfolio of locations.
But what happens when the home country of the EMNE is itself unstable? The bottom image shows the much narrower horizon of potential host locations that an MNE with an unstable home country can consider when it seeks to construct a balanced locational portfolio. Because home country operations constitute such an important part of the overall operations of the EMNE, there is a high risk of constructing an imbalanced locational portfolio if the firm does not prioritize stable host locations. According to portfolio logic, the EMNE has little choice but to exit most other less stable locations and instead to seek out more stable and typically developed host locations.
In the next section, we empirically examine the location choices made by South African EMNEs as they responded to the (in)stability in their home country.

Contextual setting
We situate our study in South Africa. South Africa has produced the largest number of EMNEs on the African continent and even within the broader emerging market category, it has been a prolific source of EMNEs, outperforming many of its larger counterparts in not only the number of EMNEs produced, but also their relatively early internationalization (Luiz & Ruplal, 2013).
South Africa has also seen very different levels of (in)stability in its national institutional conditions over the past decades. We identified four distinct periods.
• Tenuous stability: Until about 1975, instability occurred, but incidents were effectively contained soon after breaking out. • Extreme instability: The period from 1976 to the early 1990s (Mandela was released from jail in 1990) was characterized by extreme instability. • Comprehensive stability: In the period from 1994 to 2003 there was relative stability. • Growing instability: 2004 to the current period is characterized by an increase in incidents of violent protest.
To understand these phases, it is useful to reflect on country conditions. The domestic policy of Apartheid, involving the systematic discrimination against the black majority population, dated from early in the twentieth century, became law in 1948, and was entrenched after various political and legal challenges by 1956.
Extensive and increasing social instability followed. We use an index for social instability during the Apartheid era (see Fig. 2) drawn from the work of Fedderke, De Kadt and Luiz (2001). The index shown uses the most unstable year on record (1985) as the base year and is based on indicators like political fatalities, banned organizations and publications, political prisoners, states of emergency and so on. By 1994, Apartheid was dismantled and a democratic and politically stable South Africa emerged under the leadership of Nelson Mandela.
Subsequent to that, especially from 2004, locally organized protests started emerging to place demands on people who held or benefited from political power (Alexander, 2010). By 2008, more than half of such protests were violent; by 2013, more than 90% were (Chigwata, O'Donovan & Powell, 2017). Examples of such violent protests include the intentional injuring of police, foreigners or government officials, the burning down of houses or municipal buildings, looting shops, instances where police disperse protesters with tear gas, rubber bullets or water cannons, when rocks are thrown at passing motorists or tires burned to blockade roads (Powell, O'Donovan & De Visser, 2015).
Scholars place the start of growing instability in 2004, the year when the first community protests (initially called service delivery protests) started (Alexander, 2010;Powell, O'Donovan & De Visser, 2015). We followed Alexander et al. (2018) and used the ACLED database to determine the number of violent demonstrations or riots where government forces were involved. The data we present represent somewhat of an undercount, as the (growing) size or level of violence for a given event is not captured. For example: data from the South African Police Service record 1091 incidents of unrest in 2012, compared to the 379 in the ACLED database, with the discrepancy ascribable to the number of incidents of violence during a given event (Powell, O'Donovan & De Visser, 2015). Instability is on the increase, with 2021 seeing a new high in the number of daily incidents. We express all the years using the daily number of incidents from January to July 2021 as denominator (see Fig. 3).
The two datasets on social instability are not directly comparable, and it is important to note that the Apartheid era instability far exceeded what has been experienced after Apartheid. For example, deaths due to unrest numbered 3 in 2010; in 1985 over 600 people died in "township protests" alone, with more executed, dying in police custody or disappearing. The two indices do overlap for a period, and although they differ, the change in social (in)stability is clearly visible.
Figs. 2 and 3 reflect not only levels of social instability using the two sets of indicators, but also how they relate to OFDI, expressed as a proportion of GDP. We see evidence of the continued liberalization and opening up of the South Africa economy. However, there is little  evidence of dramatic shifts in the level of OFDI taking place in response to the instability, except for during the violent 1980s during which the government adopted extreme policy levers to limit capital flight (Havemann, 2014).
The South African Reserve Bank started tracking the destination of OFDI from 1956 onwards. The data show marked shifts in the destination of OFDI over more than six decades (see Fig. 4). When the share of OFDI to each location is expressed as a proportion of total OFDI for that year, there are clear and dramatic shifts in the destination of OFDI. In addition to the expected growth in importance of Asia as a host location for OFDI in the twenty-first century, two host locations stand out: Neighboring Africa and Europe. Wider Africa was unstable during much of this time as it was experiencing a transition from being colonized to selfgovernance, often accompanied by a rise of socialism and conflict during the second half of the twentieth century. In turn, Europe was a stable region with which South Africa had prior colonial and economic ties.
The dramatic shift in the location of OFDI from primarily wider Africa to primarily Europe took place during the period of extreme instability of the 1980s. On a prima facie basis it seems as if MNEs from South Africa responded to social and institutional instability by changing the location of OFDI when confronted with home country instability.

Methodology
Because of the challenges in finding objective evidence of the choices made by EMNEs , and especially the basis for those choices, we opted for a historical approach. In particular, we use what Kipping and Üsdiken (2014) term "historical cognizance", an approach where history is used as a means to develop theory. However, data are used "not just to test or develop universal theory, but [to] make that theory more contingent on the changing context" (2014: 542). This approach allowed us to longitudinally map location choices and institutional conditions.
A historical approach can be pursued with different types of evidence and research approaches (Decker, Kipping & Wadhwani, 2015). Given our interest in a complex social phenomenonhow home country conditions affect the construction of the portfolio of locations of EMNEs -we decided on a case study approach as it is the preferred approach for dealing with "how" and "why" questions, and when the boundaries between the context and the phenomenon are blurred (Yin, 2003). We especially value the fact that case studies allow for a longitudinal investigation (Ghauri, 2004). We used an embedded case design (Yin, 2003), selecting South Africa as the focal emerging market country, and then six South African EMNEs.
An abductive approach, described as an approach that involves iterating between theory and empirical evidence (Dubois & Gadde, 2002), fit well with our research question. Wadhwani and Decker (2017) argue historical analysis is almost inevitably abductive as the researcher seeks to interpret the past from a position in the present. In turn, Timmermans and Tavory (2012) foreground the important role of "surprises" in abductive research. The key surprise triggering our research interest was Fig. 4, with the very stark shift in the location of South African OFDI over what we knew was a period with substantial institutional changes.
In describing the analytic process, Saetre and Van de Ven (2021) highlight the importance of identifying and then confirming anomalies, and then generating and evaluating "hunches" for what could explaining them. This process describes well how we went about in generating and making sense of our evidence.

Case selection
In their review of case study research in international business, Fletcher, Zhao, Plakoyiannaki and Buck (2018) differentiate between theory-driven and phenomenon-driven case selection. They also identify a third path with elements of both. Although they did not use the term "abduction" in their paper, they describe a process of simultaneously considering and systematically combining (Dubois & Gadde, 2002) both theory and the phenomenon. This is the path we chose.
In keeping with our research question, we focused on South African MNEs that were or had been operating at least in wider Africa and Europe. Within that population, we also sought as much variation as possible. In particular, we selected firms whose internationalization spanned the full period under study (South African Breweries, Barloworld, Nampak) and cases where firms were either founded or internationalized during or after the turbulent 1980s (Naspers, BCX, Shoprite). We examined firms that were seeking global recognition (South African Breweries, Naspers) and also firms that explicitly positioned themselves as Africa-focused (Barloworld, Shoprite).
Given that location choice is more constrained for mining and natural resource firms, we omitted the large number of mining South African MNEs, even when there was evidence of them operating in both wider Africa and Europe. The six cases we selected represent almost a fifth (19.4%) of the market capitalization of the JSE (two of the companies were recently acquired and delisted and incorporated into larger holding companies and so for those we use the last point at which they were individually listed -BCX (delisted in 2015) and SAB (delisted in 2016). Table 1 provides an overview of the six cases we selected.
We follow Miles and Huberman (1994) and present within-case analysis first, before proceeding to a cross-case comparison of patterns observed. We start with the early internationalizers. Although liberalization meant that more firms started operating across borders later in the twentieth century, we believe that much can be learnt from the firms that were actively seeking to operate internationally even before then. We then discuss three late internationalizers.
The evidence on which the case discussion is drawn from a range of sources. In some cases, previous scholarship details relevant points, and then the sources are cited in the regular way. For most of the information, we aggregate materials from annual reports, company histories, archival material and news reports as our evidence base. Table 2 provides more detail about the data sources we used.

South African Breweries (SAB)
SAB, now acquired by AB InBev, was founded in 1895, but in 1956 became the lead brewer in South Africa through the acquisition of its major competitors. Over the next few decades, it entered a range of Southern African countries like Angola, Botswana, Lesotho, Mozambique, Namibia, Zambia and Zimbabwe. Political turmoil and in some cases nationalization resulted in SAB leaving a number of Southern African countries (e.g. Angola, Mozambique, Zambia) before the onset of severe instability in South Africa.
As home conditions deteriorated with increased anti-Apartheid protest, economic sanctions across the globe also meant that SAB's opportunities for international expansion were limited. Instead SAB focused on the South African market, and in addition to completely dominating the South African beer market, by the end of the 1980s had also diversified into a range of unrelated South African industries such as retail and tourism.
In the early 1990s, as South Africa started its political transition and liberalizing its economy, SAB renewed its focus on internationalization. In 1993, SAB started actively pursuing opportunities in the transition economies in Europe such as the Czech Republic, Hungary, Poland, and Romania. A joint venture established in China in the mid-1990s saw it become the largest brewer in China and through continued acquisitions in Eastern Europe and in Africa it assumed the same title in each of those regions. In 2000 it moved into India and the following year started a rapid expansion into Latin America. SAB subsequently used its experience to support the acquisition of US-based Miller in 2002 (establishing it as the second largest brewer in America), followed by acquisitions of breweries in other European countries like Italy (in 2003) and the Netherlands (2008) and in 2012 in Australia.
By 2013 SAB was the second largest brewer in the world with operations in over 80 countries and over 200 products. By 2014, the year before AB InBev made a $104 billion bid for the company, the regional contribution to the company's earnings (EBITDA) was 19% South African, 12% the rest of Africa, 32% Latin America, 13% Asia, 12% Europe, and 12% North American (Luiz et al., 2017). In the space of two decades, the South African company transformed itself into an international giant with strategic investments in both advanced and emerging economies. It can therefore be classified as a "global consolidator" (Ramamurti, 2009), and evidence suggests that it did use its international expansion as an opportunity for learning.
SAB expanded to the less stable African markets during the early Apartheid era when it seemed as if the South African government was successfully containing the resistance to Apartheid (Barnard & Luiz, 2018). During the period of extreme instability SAB suspended its international aspirations. As Apartheid was collapsing, SAB focused on buying assets in the transition European countries, followed by a foray into China. Once the conditions in a post-Apartheid South Africa started stabilizing, SAB rapidly expanded into other emerging markets. It nonetheless continued seeking to balance its exposure to institutional risk by making major investments in advanced economies in Europe, the USA, and Australia. As a global player, growth came from emerging markets and stability from its portfolio in advanced economies.

Barloworld
In the 1980s, Barloworldthen named Barlow Randwas the largest non-mining firm on the JSE. It had been founded by an immigrant from Notes: R15 is equivalent to about US$1. Where the date of first internationalization appears to predate the founding of the firm, it refers to internationalization by a precursor of the firm.  1986-19891991-199419991986-19891991-19941999;2004-2010Lelyveld, J. (1982 . The many faces of Barlow Rand Ltd. https://www.nytimes. com/1982/04/11/business/the-many-faces-of-barlow-rand-ltd.htm 1986-1989; 1991-1994; 1999  Both the man who oversaw the listing of the firm, Punch Barlow, and his successor, Mike Rosholt, were social progressives who campaigned to end Apartheid, frequently explaining the cost of not undertaking reforms (Lelyveld, 1982). Although the actions taken by these men were similar to what were (eventually) taken by many South African EMNEs, they were among the first to respond to the increasing institutional instability in South Africa. For example, although the EMNE had listed (as Barlows) on the JSE in 1942, it in 1969 also listed on the London Stock Exchange, in 1972 in Brussels and Antwerp, in 1973 in Paris, and in 1980 on the Frankfurt Stock Exchange. Since the late 1990s, several South African companies have followed the pattern set by Barloworld by seeking a primary or secondary listing in European countries.
During the 1960s Barloworld had expanded to South Africa's direct neighbors (Botswana, Namibia and Zimbabwe) where it had numerous operations. International censure of South Africa's Apartheid policies limited the international expansion of Barloworld, and like many South African EMNEs, Barloworld became a conglomerate with operations in multiple industries. Ongoing and intensifying foreign exchange controls from South Africa during the 1980s meant that it was hard to use funds generated inside South Africa for international expansion. Eventually various operations in neighboring African countries were sold in order to fund the geographical diversification of the firm to the more stable USA and Europe.
Using the combination of funds raised though its foreign listings and the sales of African interests, Barloworld acquired Wrenn Brothers in the USA in 1979, in 1984 Bibby & Sons in the UK, and in 1992 made acquisitions in Belgium, Spain and Portugal (Barnard & Luiz, 2018). By 1989, Barlow Rand (as it was then termed) was the second largest South African EMNE (after Anglo American) and 79th on the global Fortune 500 ranking.
The very strong emphasis on especially Europe from 1969 to 1992 is noteworthy in the light of Barloworld's current operational footprint: Barloworld switched from its institutionally less stable African neighbors to the more stable European institutional conditions as its home country institutional conditions deteriorated, but was quick to reverse course. In 1994, shortly after Mandela's release, Barloworld (re-)entered Zambia, Angola, Mozambique, and Malawi. Over the next decade, it expanded its footprint in wider Africa, e.g. to Central Africa in a joint venture with Tractafric in the DRC (2004) and West Africa (Ghana) in 2012.
Barloworld not only focused on Africa, but also in 1998 expanded to Russia. It is currently operating in Mongolia, the Russian far east and Siberia. Although there were additional acquisitions in the US and Europe during the immediate post-Apartheid era, e.g. 1994 in Australia, 1999and 2003in the US, and in 2005 in Denmark and Sweden, Barloworld soon divested those assets. The only developed countries in which Barloworld retained a presence were the UK and a limited presence in Australia.
Decisions were not made only on the basis of location. Barloworld exited a number of markets, e.g. handling between 2014 and 2017. Thus location decisions are most comparable where the EMNE remained in an industry. For example, in 2005 Barloworld acquired Avis dealerships in Scandinavia, but by 2010 sold its European interests in the market, instead extending interests in wider Africa, e.g. Kenya and Tanzania. Barloworld showed a clear preference for emerging markets as its primary location once the institutional conditions in its home country had somewhat stabilized.

Nampak
Nampak is a South African packaging company that was listed on the JSE in 1969 after a merger between Amalgamated Packaging Industries (API) and National Packaging (Natpack). Its packaging operations cut across paper, ridges and metals.
From before the listing, API and Natpack were active across Africa: API from 1952 in Rhodesia (Zimbabwe), in Kenya from 1959 and Angola from 1966, and Natpack in Mozambique andZambia in 1963. In 1975, both Angola and Mozambique suddenly became independent, and soon after entered into civil wars, aggravated by the interference of the South African Apartheid government. By 1979, the changing political situation in Rhodesia (Zimbabwe), according to the official company history, "affected the ownership of Nampak's companies" (Hocking, 1987:214). In 1984 it was documented that "the Nampak group still had interests in other African countries, though it was becoming difficult to manage them effectively" (Hocking, 1987:257).
During the same period, as South Africa saw a series of intensifying periods of social instability, the holding family emigrated from the country. In the 1970s, the British paper manufacturer Reed International bought Nampak and rapidly sold it to Barlows, the predecessor of Barloworld. Given that sanctions made it hard to internationalize during Apartheid, during the period while Nampak was a wholly owned subsidiary of Barloworld, Nampak itself engaged in extensive domestic acquisitions, e.g. of metals and plastics packaging companies. As soon as the Apartheid-inspired limitations on international business were removed, the need to operate as a domestic conglomerate was removed and a number of firms unbundled. Nampak remained under the control of Barloworld until 1993 when it regained full control of operations.
When Nampak regained independence, it started making acquisitions in Europe, acquiring Plysu in the UK in 1994 and in 1996 Blow-Mocan, also UK-based but with a substantial presence in Europe. After consolidating the two as Nampak Plastics Europe in 1999, Nampak continued to acquire a number of additional packaging plants in the UK and Ireland.
As institutional conditions in the home country improved, Nampak started prioritizing countries in wider Africa from 2002 again, initially with the acquisition of the American MNE Crown Cork's African interests which included operations in Ethiopia, Kenya, Nigeria, Tanzania, Zambia and Zimbabwe. This acquisition started an acceleration of its expansion on the continent both in terms of further acquisitions such as in Nigeria (2007), Malawi (2013) and Kenya (2014), and in terms of greenfield investments, the most notable of which in Angola in 2010.
In 2019, Nampak decided to dispose of its European division (although not in the UK and Ireland) and subsequently branded itself as "Africa's largest packaging company" with 22 sites (contributing twothirds of revenue) in South Africa, and 17 sites (contributing one-third of revenue) in the rest of Africa. In South Africa it currently has an estimated market position in excess of 80% in the beverage cans sector with a strong position in packaging in two of Africa's largest economies, namely Nigeria and Angola.
However, Nampak has been struggling to repatriate earnings from Angola, Nigeria, Zimbabwe (the first two due to those economies' heavy reliance on oil, the latter due to economic distress). Moreover, the magnitude of its US dollar-denominated debt was of concern. In response, Nampak sold its Nigerian carton business to the Swedish AR Packaging Group.

Naspers
Naspers is a publishing group that was founded in 1915 as "Nasionale Pers", in a quest for empowerment by Afrikaners, the same people who would come to power with Apartheid policies in 1948. Over decades, Naspers grew organically with a stable of largely print-based offerings (newspapers, magazines, books) and from 1986, television through MNET (later named MultiChoice). In the first phase of Naspers's expansion (1915 until the early 1990s), it remained a South Africafocused media company (Robertson & Luiz, 2019), under the patronage of the Apartheid government with which it had a close relationship.
It remained a domestic company until the 1980s, when Naspers started exploring pay satellite television opportunities in Africa. The launch of MNET led to its international expansion into neighboring countries and then into the rest of sub-Saharan Africa. Naspers' first experiment with public listing was only in 1990, when it listed MNET on the JSE. After initially focusing on neighboring Namibia and Botswana, by 2002 MNET was broadcasting in twenty African countriesa number that grew to fifty over the next years. Buoyed by the success of that listing, Naspers was listed in 1994, the year of the first democratic elections.
As soon as the National Party lost power to Nelson Mandela's African National Congress, in 1994, Naspers started changing its portfolio of international locations. Its next forays internationally were into Europe for example its 1995 acquisition of FilmNet, a pay TV station that operated in northern Europe, followed by other acquisitions in the rest of Europe and the USA. By 2000 critical post-Apartheid reforms had been completed, and managers' confidence in the institutional framework of the country had been largely restored (Barnard & Luiz, 2018).
Naspers disposed of European investments to fund ventures into a range of similar emerging markets and turned itself into an emerging market media giant (Robertson & Luiz, 2019). Some of Naspers's notable investments during this third phase of its expansion included its investment in Tencent in China (2001), Mail.ru in 2006 which has become the largest Internet services company in Russia, and Editora Abril which is a major Brazilian publisher and printing company and one of the biggest media holdings in Latin America. In 2012 it completed its dominance of the BRICS countries with a 16% stake in Flipkart in India which it sold to Walmart for $2.2 billion in 2018 (netting a $1.6 billion profit from the sale of its stake).
The continued growth of Tencent was starting to overwhelm the overall Naspers portfolio of assets. In addition, the large-scale corruption of President Jacob Zuma and his cronies (from 2009 to 2018) led to widespread concern about the stability of South Africa, and Naspers was trading in South Africa at a substantial discount (Barnard & Luiz, 2018). Naspers started to again focus on the developed world. Its 2019 listing of its international shares in the Netherlands (as "Prosus") is perhaps the most prominent example, but it has also been investing in new educational technology companies with global ambitions in the USA.

BCx
The information technology firm BCX came into being as the result of a series of mergers and acquisitions. Its earliest precursors were Persetel and Q Holdings. Persetel was founded in 1979 as a result of IBM's exit from South Africa. At the time, IBM was represented in South Africa by Barlow Rand, and when IBM divested from South Africa, Barlow Rand acquired the firm, and later divested it. Whereas Persetel provided mainframes, Q Holdings, founded in 1986, offered software services. Both firms focused entirely on the South African market until they merged in 1997, when they almost immediately started internationalizing.
In 1998, as soon as the ANC-led government started implementing extensive domestic reforms, the merged company expanded its ownership of Comparex (since acquired by SoftwareOne) in Germany to 75%. As often happens with EMNEs (Barnard, 2014), the acquirer assumed the name of the target company. The stock exchange information news clip at the time highlighted that the firm had "strengthened its position as the leading independent pan-European [our emphasis] networking integrator" and that ensuring that all European operations reach their full potential was a strategic goal (Comparex -Interim Results, 1999).
Business Connexion was a black-founded and black-owned privately held company, founded in 1996. It merged with Comparex in 2004, at which point BCX was listed on the JSE. Given government's desire to encourage (and reward) economic transformation through black ownership (Tangri & Southall, 2008), the substantially black-owned BCX turned to exploiting the extensive opportunities in South Africa. BCX did however retain operations in the UK (where it had been operational since 1995), and in its closest and wealthiest neighbors, Botswana and Namibia (where there had been operations since 1994).
After dipping its toes into the Zambian market (1996), it took almost an additional decade before African operations became a priority, with entries into Tanzania (2000), Mozambique (2005) and Nigeria (2008). Telkom, South Africa's semi-privatized telecommunications provider, acquired BCX in 2015 in order to increase its presence in the information and telecommunications arena. On its website, BCX describes itself as an "end-to-end digital partner for corporates and governments across the African continent".

Shoprite
Shoprite was established in 1979 and operated domestically in the then-turbulent South Africa during the first decade of its existence. Its first international foray was in 1990 (the year the dismantling of Apartheid started with Nelson Mandela's release from prison) to Namibia, a stable neighboring country with a close history to South Africa. For the next decade, Shoprite focused on neighbors and other geographically close Southern African countries: Zambia, Swaziland, Lesotho, Mozambique, Botswana and Zimbabwe.
Only in 2000 did Shoprite start venturing further afield in Africa. In the next decade Shoprite opened up, in sequence, in Uganda, Malawi, Madagascar, Mauritius, Tanzania; Ghana, Angola, Nigeria, the Democratic Republic of Congo and Kenya. Apart from an acquisition in India in 2006, disposed of in 2010, Shoprite has operated entirely on the African continent.
In recent years, however, as instability in the home country increased, Shoprite has been exiting various African countries. It very quickly reversed its late entry into Kenya and then announced the end of its fifteen-year presence in Nigeria. After the July 2021 riots in South Africa, Shoprite also decided to exit other African markets where it has a smaller footprint such as Ghana and Madagascar, instead focusing on rebuilding (most of) the hundreds of Shoprite stores that had been damaged or destroyed inside South Africa.

Cross-case comparison
In making sense of firms' location choices in response to instability, we summarize each firms' location choices in a map where the four periods are laid out longitudinally -with all the early internationalizers in one figure, and the more recent internationalizers in another. (See Figs. 5 and 6.) Whereas we visually represented our overall argument in terms of stability or not (Figs. 1 and 2), in presenting our empirical evidence in Figs. 5 and 6, we categorize the host locations as developed and developing regions, rather than in terms of (perceived) stability and instability, as those could be challenged as subjective.
We expect a lag between when the level of home country (in)stability changed and EMNEs' actions. For firms to have confidence that the intense violence of the 1980s had indeed ceased in response to Mandela's release, or that the community unrests observed in the mid-2000s were in fact the start of an intensifying trend takes some time. In terms of making sense of EMNEs' internationalization decisions, we therefore interpret each period as having ended four years after the actual social changes (shift to stability or instability) took place.
We use the visual summary of the EMNEs' foreign locations to guide our theory building about the level of home country (in)stability and EMNEs' locational portfolio construction. As we reviewed the evidence, the location choices of the first two periods (i.e. of the early internationalizers) seemed consistent with prior work. During the first period (see Fig. 5), a time of relative stability, an incremental Uppsala-type process (Johanson & Vahlne, 1977) was observed: EMNEs located in geographically and institutionally proximate countries. The overwhelming preference of EMNEs to locate in the developed world during the second period of extreme instability was in turn consistent with escape-driven FDI (Cuervo-Cazurra, Narula & Un, 2015).
But much of what we observed in the two later periods was not well explained by extant research. We therefore iterated between extant theory and our evidence to develop an explanation for what we were observing.

Theory development
Although it is known that the bulk of EMNEs' operations are in their home country or at least home region (Banalieva & Santoro, 2009), extant scholarship has not considered that EMNEs' exposure to less stable home locations may systematically affect their location choices.
Our evidence suggests an incremental Uppsala-type process (Johanson & Vahlne, 1977) during the initial period, of tenuous stability. All the early internationalizers located in countries that were immediate neighbors with only Barloworld also exploring options in the developed world. In this regard it is important to note Barloworld was managed by social progressives who expressed concerns about the implications of South Africa's Apartheid policies while economic growth was still in excess of 5% a year and before the country started suffering international censure (Lelyveld, 1982). There was stability but it was tenuous; it seems that Barloworld took the incidents of social turmoil seriously.
As the period of extreme instability unfolded, Barloworld engaged in escape FDI (Barnard & Luiz, 2018). Given very strict exchange controls at the time (Havemann, 2014), Barloworld used its foreign listings and the proceeds of the sales of its African interests to fund a shift into the developed world, specifically the USA, UK and Europe. But Barloworld was not the only EMNE seeking to escape to more stable developed markets.
Even before the post-Apartheid government was sworn in, as soon as financial controls were relaxed, Nampak engaged in transactions in the UK and USA. Unless it had previously been forced to leave a country, SAB had held on to existing assets in wider Africa with the explicit purpose of generating funds to support international expansion, and for the same purpose introduced a unit to drive exports on the continent. As soon as exchange controls were lifted, SAB embarked on a series of acquisitions in Central Europe.
Even Naspers and BCX, two more recent internationalizers, entered Europe as soon as exchange controls were lifted. Indeed, the early and the more recent internationalizers responded in the same way to the instability of the 1980s: they all seemed to seek escape to stable, developed economies.
The only exception to this trend is Shoprite, which entered neighboring countries (rather than the developed world) from 1990 onwards and before it was deemed wise (Cohen, 2010). In the 1995 annual report, the chairman and managing director of this low-cost retailer explained Shoprite's position (1995:4): We are optimistic, not only about South Africa, but also about the African continent, which is awakening as a force in the world economy, in which we intend to play an active part. Our investment in neighbouring countries will assist in stimulating their economies and ultimately help reduce illegal immigration across our borders, which places pressures on our own country, its economy and resources. We regard this venture as an excellent investment opportunity, offering great benefits in terms of exports and revenues, both for our organisation and for the country.
Thus whereas the behavior of EMNEs during the initial period of tenuous stability was consistent with an incremental model of internationalization, and their actions during the period of extreme instability were consistent with escape-explanations, extant theory did not explain what took place during the following period of comprehensive stability. For both the early and more recent internationalizers, by far the most internationalization activity occurred during this period, after decades of violent resistance also a period of sustained economic growth.
But while South Africa was stable, growing in prosperity and characterized by general optimism about the path of the new government, the EMNEs sold almost all of their assets in the developed world. Moreover, these assets were all sold soon after being acquired. The evidence challenges the presumed "stickiness" (Rugman & Verbeke, 2001) of locations.
Nampak and BCX each retained assets in the UK, partly to serve as a second home location, similar to that of another South African EMNE, Discovery (Barnard, 2021). SAB was increasingly positioning itself as a global player, and acquired Miller in the USA, as well as operations also in Italy and the Netherlands during this period.
But by far the most important host region during the period of comprehensive stability was wider Africa. BCX lost strategic momentum after its acquisition by the South African state-owned telecommunications company, but retained a presence in the African countries where it had operated before. Nampak (after some experimentation elsewhere) and Shoprite (from the outset) developed explicit Africa-focused strategies. Barloworld positioned itself in two main regions, wider Africa and Russia. Naspers, a global consumer internet company, positioned itself as an emerging market provider, and acquired firms in emerging markets across the globe. This disengagement with the developed world was unexpected. An EMNE like Barloworld had more than forty years' experience in the developed world; most of the others had about fifteen years' experience. Those EMNEs had presumably developed capabilities to deal with such locations, but the appetite for expansion into the developed world seemed limited. Why did EMNEs seem willing to abandon not only their interests in developed host locations, but apparently also the learnings and capabilities they had developed there?
There was precedent for what we found: Thomas, Eden, Hitt and Miller (2007) found that many EMNEs did not return to developed markets after exiting them, even though the survival rate after re-entry was quite high. Their explanation was that EMNEs were suffering from a cognitive bias, but we noted that their sample consisted of firms from Brazil, Chile, Colombia and Mexico and over a turbulent period for these Fig. 6. Summary of location decisions of more recent internationalizers. Note: Grey shading represents the areas where we argue EMNEs will internationalize, depending on the level of host country stability. Underlined locations represent exits. Bold represents first entry during a period, bold italics subsequent, and italics the latest entry into a location during the period.
countries (1991 to 2000). We became increasingly convinced that the instability of EMNEs' home countries did not simply trigger "escape", but that another process was at work.
The changes during the fifteen years of the final period, where South Africa was characterized by growing instability were far more limited. In fact, in spite of the continued growth of outward FDI (as per Fig. 3), the EMNEs made much smaller changes to their locational portfolios than had been seen in earlier periods. As the home country became increasingly unstable, the EMNEs sought to strengthen their positions. Barloworld redoubled its focus on key partnerships like with Caterpillar, Shoprite divested from less profitable markets in wider Africa but did not expand to new regions, and Nampak's sought to reduce its debt, using divestment to do so. All these changes placed the EMNEs in a better position to weather instability, but the changes were consistent with how the EMNEs defined themselves. The largest two EMNEs started defining themselves as global players. SAB acquired Foster's in Australia (before being acquired by Anheuser-Busch InBev), and Naspers, buoyed by its 31% share in the tremendously successful Tencent, stopped positioning itself as an emerging market player, and started making acquisitions in the developed world.
Throughout, the type of dramatic shifts seen in both the period of extreme instability and the period of comprehensive stability were not in evidence. We think it is relevant that all the EMNEs had by now developed experience in a range of host countries abroad. To the extent that they had become less reliant on their home country, they were less vulnerable to home country instability and therefore experienced less of a need for a dramatic transformation of the locational portfolio.
Comparing this with what we observed in the initial periods, we propose: Proposition 1: The greater the proportion of overall operations of an EMNE in its home country, the greater the propensity of the EMNE to change the composition of its locational portfolio in response to changes in home country conditions.
Most EMNEs have substantial operations in and exposure to their home country. Their locational decisions are sometimes made against the backdrop of an unstable home country and home region. When instability permeates an EMNE's home country context, the EMNE finds itself in a Catch-22 situation. Much as the EMNE may wish to leave its unstable home, it is hard to reduce its exposure to its home country. The instability reduces EMNEs' assets, absorbs a lot of managerial attention (which thus cannot be directed at improving the offering of the firm) and because the instability also makes the location less attractive for foreign MNEs, the firm has reduced opportunities for learning from stronger competitors (Wu & Chen, 2014). However, severe instability also places the very existence of the firm at high riskespecially if and to the extent that an EMNE has a large proportion of its operations in its home country.
Having to make location choices that can safeguard the existence of the firm during periods of extreme instability explains many of the choices made by EMNEs. First, it explains the willingness of EMNEs to internationalize even though the process involves navigating extensive restrictions and institutional hazards. (For example, Havemann (2014) reports that during the period of extreme instability in South Africa, there were limits on even how much gold jewelry could be taken out of the country.) Second, it explains why EMNEs so strongly prefer to internationalize to the more stable developed marketseven if those markets present more challenges and have less potential for growth (Witt & Lewin, 2007). Finally, EMNEs' concern about ensuring their survival explains why they seem less concerned about whether or not they have the capabilities to succeed in a host location. Instead, they seem to conceptualize their internationalization as the construction of a more balanced locational portfolio. This gives rise to our second proposition: Proposition 2: The greater the proportion of overall operations of an EMNE in its home country, the greater the propensity of the EMNE to internationalize to stable (unstable) host countries when home country conditions are unstable (stable).

Discussion
EMNEs' home countries can experience changes that are not only negative but also positive; in their paper on Latin American MNEs, Hermans and Reyes (2020:166) speak of "periods of economic bonanza". But in both cases, there is a violation of an important boundary condition to extant research, namely the presumed relative stability of home country conditions. The longitudinal nature of our evidence allows us to examine what happens during periods when that is true, but also when there is instability.
Much as capability development is an important motive for EMNEs, when an EMNE is particularly exposed to a home country undergoing sudden and dramatic challenges, other considerations may become more important. The presence of an unstable home county challenges extant theory on EMNEs by suggesting that scholars' strong emphasis on learning and capability upgrading when EMNEs locate in developed countries is perhaps not always justified. Cavusgil's (2021) observation that scholars have had somewhat of a tendency to "glorify" emerging markets is worth highlighting, as is the fact that extreme instability in a home country can threaten EMNEs' very survival. If EMNEs regard internationalization to developed countries as a way to secure a "safe haven", they are thinking differently about locations. Indeed, the search for stability rather than for knowledge gives rise to a range of subtle but important differences in what EMNEs value. Table 3 summarizes some of the central differences.
Whatever the motive that had led EMNEs to locate in developed countries during the period of extreme instability in our study, at least some of the EMNEs can be expected to have accumulated valuable experience and capabilities when they were located there. It would not be unreasonable to expect those firms to continue operating in the same locations in the developed world or to expand to similar locations once there was comprehensive stability at home. Yet we did not observe that. Instead, almost all the EMNEs exited the developed regions. Most left to focus purely on African expansion; some also expanded to other developing regions.
Scholars sometimes seem to suggest that the competitiveness of EMNEs in the advanced economies represents a pinnacle in an evolutionary path of capability upgrading (e.g. Kotabe & Kothari, 2016). However, our evidence suggests that such an evolutionary path cannot be assumed. Instead, EMNEs seem to construct a portfolio of locations. Moreover, to the extent that their home country is stable rather than unstable, the attractiveness of emerging markets (even given the associated risk of their greater instability) increases. In the era of Table 3 EMNEs internationalizing to developed host locations in search of stability versus capability upgrading. comprehensive stability, and given the continued large proportion of an EMNE's operations in the home country, it makes sense to adjust the portfolio to include locations with a greater risk of instability, but also the promise of greater returns. Cantwell (2009:35) conceptualized the MNE as "a synergistic locational portfolio of complementary sources of knowledge". EMNEs also seem to construct a locational portfolio. However, the EMNE locational portfolio seems to be more strongly influenced by the instability and risks of the locations that they seek to avoid or exploit than about knowledge. This shift in emphasis matters, because it suggests the need for a different range of explanations.

Conclusion
In this paper, we provide evidence that EMNEs use a portfolio logic in making location choices, opting for 'safe', stable host locations when the home country is unstable, and for riskier, less stable host locations when the home country is stable. We develop propositions that acknowledge the outsized influence of the home country, and in which the desire of the EMNE to construct a balanced locational portfolio has an important influence on where they locate.
Research within international business on Africa is still in its infancy and Africa is probably its least researched continent (Adeleye et al., 2020;Boso et al., 2019). If we are to seriously understand EMNEs, then Africa has to be part of that conversation. We therefore believe that our empirical focus is useful, although it is a limitation of our work that the South African context is so extreme in the frequency and extent of changes in (in)stability experienced. However, as Barnard, Cuervo-Cazurra and Manning (2017) point out, the extreme conditions in Africa are often useful to highlight the boundaries of extant theory. Moreover, the South African case is not unique: The fallout from Argentina's dollarization experiment and subsequent debt default in 2001; Mexico's peso (so-called tequila crisis) of 1994; the South-east Asian financial crisis of 1997; recent hyperinflation in Venezuela, and the Russia-Ukraine conflict are but some other examples of home country instability in emerging markets.
Instability in our case was pervasive; the period saw (quite rapid) institutional change, societal transformation, political contestation, economic distress, and violent protest. Future work is needed to consider cases that are less extreme, specifically to hone in on specific sources of instability and their effects. For example, Banalieva, Cuervo-Cazurra and Sarathy (2018) identify the asymmetrical effects of either intensifying or fading pro-market reforms versus reversals. The relative effects of the dynamismhow fast things changeversus the content of changes (e.g. violent protest versus new formal institutions) are likely to be different, even though both are likely to affect whether and where firms choose to internationalize. Understanding how is an important topic for future research.
The question must also be asked whether EMNEs 'abandoned' capabilities they had developed in the more stable high-income countries when they exited those countries to go to lower income countries with potentially greater growthor whether they in some other way benefited from the experience in that location. In other words, additional research is needed to better understand whether (and if so, to what extent, and in which ways) the capabilities developed in one type of context resemble or differ from capabilities developed in a very different context.
Our findings are tentative and need to be quantitatively verified. Much work can be done in this area. A fruitful area is in better understanding the locational portfolio construction of EMNEs. For example, responsiveness to home country instability for EMNEs with varying proportions of operations inside their home country should be tested. The effects of instability in a particularly important host region are also worth examining. For example, Barloworld and Naspers both had exposure to Russia at the start of the Russia-Ukraine war, but the relative proportion of assets Barloworld held in Russia was greater than that held by Naspers. How does instability differentially affect those firms?
Risk diversification is argued to involve limited active adjustment of portfolios, whereas real options theory is argued to involve the active and ongoing exercising (or not) of options. But as the Russia-Ukraine war started, that distinction was challenged for both Naspers and Barloworld, both of which realized handsome returns in Russia and adjacent regions, and both which saw their investment there as a stable commitment. It is critical to remember that emerging market are, for a range of reasons, more prone to instability than the developed world. This means it is quite likely that many EMNEs will at some point experience instability in either their home or in developing host locations. Practically, our work suggests that EMNEs need to be able to shift quite rapidly from thinking about locations in terms of risk diversification into real options thinking. Hernandez and Guillén (2018) urge scholars to consider what is theoretically distinct about EMNEs. We show how EMNEs, far more than MNEs from the developed world, must chart a path in the face of considerable home country instability and how this simple fact challenges many well-established theories and motives of internationalization for EMNEs: These conditions affect where EMNEs internationalize, not only during periods of instability, but also once stability returns. Future theorizing of EMNEs will do well to understand how such turbulenceboth during and across periods of home country instabilityaffect the strategic choices of EMNEs.

CRediT authorship contribution statement
John M. Luiz: conceptualisation, methodology, investigation, analysis, writing, review process and final editing. Helena Barnard: conceptualisation, methodology, investigation, analysis, writing, review process and final editing.