Understanding foreign divestment: the impacts of economic and political friction

Past research on foreign divestment has recognized the impact of economic and political differences 1 . However, the prior findings remain equivocal. We adopt the Positive Organizational Scholarship perspective to provide more contextualized insights into the effects of economic and political differences on foreign divestment. Spe- cifically, we consider the juxtaposition of national differences and levels of firm interaction with the different contexts. Thus, we develop the concept of friction to assess levels of economic and political differences. We further argue that economic friction will have a curvilinear (U-shaped) effect on foreign divestment, whereas political friction will produce a monotonic (positive) effect. Moreover, we introduce ownership level as a moderator into the main hypotheses. Drawing on data from 2400 foreign subsidiaries of 310 Finnish multinational enterprises, from 1970 – 2010, we provide support for our main hypotheses, although the moderating effect of ownership levels is not supported. We further compare the effects of differences measured by friction with those measured by distance. Accordingly, our research highlights the importance of detecting specific conditions for the investigation of the impact of economic and political differences in the foreign divestment literature.


Introduction
Stora Enso, a Finnish MNE and leading global provider of renewable solutions, planned to sell its Sachsen Mill, located in Eilenburg, Germany, to Model Group, a Swiss company. Later, Stora Enso announced it was to close its paper factory in Sweden (Stora Enso, 2021). These two cases are among hundreds of instances of foreign divestment (FD) over the past few years, referring to the exit of an active foreign subsidiary of a multinational enterprise (MNE) from the host country (Boddewyn, 1979(Boddewyn, , 1983, via sell-off (first case) or closure (second case) (Konara & Ganotakis, 2020;Sartor & Beamish, 2020). Nevertheless, this striking business phenomenon has received scant scholarly attention (Arte & Larimo, 2019;Coudounaris, Orero-Blat & Rodríguez-García, 2020;Schmid & Morschett, 2020).
Economic and political differences have profound effects on the performance and survival rate of MNEs, that is, providing institutional arbitrage and conveying an important source of uncertainty for MNEs (Kostova & Zaheer, 1999;Gaur & Lu, 2007;Berry, Guillen & Zhou, 2010;Sartor & Beamish, 2020). Economic and political differences, which may be grouped as formal differences or regulatory differences (Scott, 1995;Gaur & Lu, 2007), refer to those between countries (Berry et al., 2010). The prior research has paid considerable attention to the influences of economic and political differences to provide knowledge on how MNEs should deal with the differences (Kostova & Zaheer, 1999;Kostova, Roth & Dacin, 2008;Kostova, Beugelsdijk, Scott, Kunst, Chua & Essen, 2020). Despite this, three striking issues have been raised.
Institutional differences, while making it hard for MNEs to compromise (Kostova & Zaheer, 1999;Kostova et al., 2008), could deliver several advantages that encourage MNEs to "stay abroad", including unique resources, arbitrage opportunities, and innovation capacity (Gaur & Lu, 2007;Edman, 2016). Hence, an inbuilt assumption that differences are detrimental can lead to erroneous conclusions (Clampit, Kedia, Fabian & Gaffney, 2015). Accordingly, it is important to broaden our mindset and provide a balanced treatment of the effect of differences (Stahl et al., 2016). The Positive Organizational Scholarship (POS) perspective reflects this idea in the IB literature, emphasizing the advantages of differences. However, this lens has received scant attention in the extant literature (Edman, 2016;Stahl et al., 2016).
Third, the prior research tends to focus only on national differences, excluding firms' specific interaction with the different context. This leads to oversimplification of the fact that the embeddedness of MNEs in national differences is not the same for all firms (Witt & Lewin, 2007;Graafland & Noorderhaven, 2020). For instance, it is likely that a Finnish MNE investing in China for the first time will be influenced differently by economic and political differences than a Finnish peer that already has ten subsidiaries in China. In this case, national differences refer to the differences between China and Finland, i.e., economic and political systems, while firms' interactions refer to specific situations where Finnish firms interact with these different contexts. Accordingly, although the national economic and political differences between China and Finland are similar in both firms, the levels of firm interaction in the first case are likely different from those in the second, due to different levels of involvement, that is, first-time entry vs. several previous entries, and sole subsidiary vs. ten subsidiaries. In other words, when a Finnish MNE has more interaction with China, it will be more deeply influenced by the host country environment than its home-country peers that have less interaction in the host country.
By the same logic, scholars argue that depending on the levels of equity ownership, that is, international joint ventures (IJVs) and wholly owned subsidiaries (WOSs), foreign firms may be involved and influenced dissimilarly by the different national context (Gaur & Lu, 2007;Dhanaraj & Beamish, 2009;Pattnaik & Lee, 2014). Nevertheless, Meyer, Li & Schotter (2020) recently reported scant interest in IB research exploring multi-level interaction to external influences; for instance, in home-host conditions, and inter-firm reaction, where parent firms and subsidiary units work together to respond to changes in their external environment. This has motivated our research interest to delve into the influences of firm-level interaction.
Motivated by these ideas, we examine the influences of economic and political differences on foreign divestments, and augment the extant literature in three ways. First, we apply the POS perspective, and reveal the multifaceted effects of economic and political differences, in order to understand whether or not differences could be benefits rather than always being disadvantageous (Clampit et al., 2015;Edman, 2016;Stahl et al., 2016). Accordingly, we eschew the conventional assumption regarding the negative influence of differences. Our tenet is that, at low levels of economic difference, foreign subsidiaries may enjoy several local resources and take advantage of economic arbitrage (Gaur & Lu, 2007;Jackson & Deeg, 2008;Mallon & Fainshmidt, 2017). However, these benefits are not always available, while adaptation costs associated with searching for the advantages increase significantly when economic differences are higher (Gaur & Lu, 2007;Shirodkar & Konara, 2017). Consequently, we propose that economic differences have both positive and negative effects that might emerge from cross-economic interactions. Thus, the effect on divestment probability is curvilinear.
In contrast, political differences are less likely to provide such arbitrage opportunities, including favorable political changes and unique support from local governments, due to the dominant powers of local political agencies, higher adaptation costs, and potential criticisms from external stakeholders (Dhanaraj & Beamish, 2009;Giambona et al., 2017;Lorenz et al., 2018;Cannizzaro, 2019;Cordero & Miller, 2019). Hence, we propose that political differences produce constraints and increase divestment propensity.
Second, we switch to an advanced approach, friction, proposed by Shenkar (2001Shenkar ( , 2012, instead of the common measure, "distance", to assess the differences. While previous studies theoretically proposed that friction may be a more useful lens than distance (Clampit et al., 2015;Lorenz et al., 2018), we are not merely invoking a friction lens when testing these differences, but actually measuring friction. Essentially, friction refers to the extent to which different entities resist or rub up against each other in interactions in multilevel analysis (Shenkar, 2001(Shenkar, , 2012Koch, Koch & Shenkar, 2016;Singh, Pattnaik, Lee & Gaur, 2019). In other words, friction reflects the combined differences perceived at different levels, that is, national and firm levels. We argue that, depending on how firms encounter the different contexts, levels of friction defined in different situations are different. In addition, while Shenkar and his colleagues (Shenkar, 2001;Luo & Shenkar, 2011;Li, Liu & Qian, 2019;Singh et al., 2019) focus solely on cultural friction, we focus on economic and political frictions. We further compare findings on the effect of differences measured by distance and by friction, in order to assess the advancement and validity of the friction metric.
Third, we contribute to the extant literature by uncovering the moderating effect of subsidiary ownership levels (WOSs vs. IJVs) on the economic and political friction -FD relationship. Past research reports that depending on different ownership levels, foreign subsidiaries will encounter different levels of interaction, leading to different propensity of divestments (Hennart, Kim & Zeng, 1998;Delios & Ensign, 2009). Luo & Shenkar (2011) echo this idea by theoretically proposing that foreign subsidiaries with different ownership levels encounter different levels of friction. Therefore, it is promising to empirically examine this proposal.

Theory and hypotheses development
Institutional theory discusses the influence of institutional differences on organizational survival and success (Kostova & Zaheer, 1999;Xu & Shenkar, 2002;Kostova et al., 2008). Since MNEs always experience institutional differences when operating abroad, institutional theory has become ubiquitous in the IB literature (Kostova et al., 2020). In Table 1, we review 16 novel studies on the effects of economic and political differences on FD, of which nine studies also include ownership levels in testing models. The table shows that while economic and political differences are measured using various indicators, the applied distance approach is similar in those studies. As noted above, IB researchers tend to assume economic and political differences have a positive effect on divestment (Pattnaik & Lee, 2014;Kang et al., 2017). However, these effects are also found to be negative (Tsang & Yip, 2007;Xia, 2011) or non-significant (Meschi & Riccio, 2008;Meschi, Phan & Wassmer, 2016). Notably, focusing on arbitrage opportunities and costs, Gaur & Lu (2007) reported a curvilinear effect of regulative differences. The equivocal findings have led to an inevitable debate: What is the influence of economic and political differences on foreign divestments? Cameron (2017) explains that because individuals often remember the consequences of negative events rather than those of positive events, negative bias exists, especially among studies on organizational outcomes and survival. Furthermore, researchers focus mainly on the strongest factors explaining the greatest variance, leading to unfair treatment of the positive effects (Cameron, 2017). Aiming to resolve the unbalanced treatment in the extant literature, we apply the POS perspective to highlight the multifaceted effects of economic and political differences on foreign divestment. POS is not a single theory per se, but represents a different view that focuses on both the positive and negative outcomes of a given phenomenon, and discusses the mechanisms of the outcomes (Cameron & Caza, 2004;Stahl & Tung, 2015;Edman, 2016;Stahl et al., 2016).

POS perspective in IB studies
One primary objective of POS is to challenge the traditional bias regarding the negative impact of differences, and emphasize the need to pay fair attention to positive influences (Cameron & Caza, 2004;Cameron, 2017). The POS perspective encourages researchers not to consider institutional differences not only as challenges, but also opportunities for arbitrage, complementarity, or creative diversity (Stahl & Tung, 2015;Zaheer, Schomaker & Nachum, 2012). For instance, elaborating on the POS lens, Edman (2016) admitted that while the liability of foreignness still mattered, being different could improve firm innovation, provide unique human capital, develop new market segments, and consumer preference. Accordingly, Stahl at el. (2016) encouraged researchers to allow for the multifaceted effects of institutional differences, rather than adhere to a monotonous hypothesis, when theorizing their proposal. On other words, researchers should study the mechanisms generating benefits or challenges stemming from the differences in specific situations. Applying the POS lens, we investigate the specific nature of each institutional environment, that is, economic and political institutions, and its influences on the chances of subsidiary survival. In so doing, we emphasize the benefits and disadvantages of the differences, and theorize that "being different" is not always disadvantageous. Instead, depending on the institution and situation, differences may create advantages and benefits that increase the probability of survival, or accelerate difficulties and increase hazards among foreign subsidiaries.

Measuring the differences: from distance to friction
Institutional theorists have coined the term institutional distance to measure institutional differences quantitatively, which is a convenient method to apply (Konara & Mohr, 2019). Distance is also applied to measure the differences pertaining to each dimensional institution, that is, economic and political institutions (Berry et al., 2010;Pattnaik & Lee, 2014;Kang et al., 2017). Nevertheless, given that previous findings on the influence of institutional distance are contradictory, researchers have questioned the reliability and validity of the distant concept (Shenkar, 2001;2012;Luo & Shenkar, 2011;Konara & Mohr, 2019).
One of the main criticisms of the distance concept is that distance reflects only national differences, even though firms may interact with the different context at different levels (see Shenkar, 2001Shenkar, , 2012Popli, Akbar, Kumar & Gaur, 2016;Konara & Mohr, 2019, for more details). Simply put, the distance concept assumes that all firms within the same pair of home-host countries are influenced equally by institutional differences, regardless of the different levels of the firms' interaction with the differences. However, IB scholars have emphasized the importance of considering contextual variations in examining the influence of institutional differences (Orr & Scott, 2008;Slangen & Hennart, 2008b;Singh et al., 2019). For instance, Orr & Scott (2008) highlighted that impacts of institutional differences vary in different situations firms encounter. Similarly, Singh et al. (2019) stated that in specific circumstances, foreign subsidiaries are influenced diversely by institutional differences, while Lorenz et al. (2018) found that institutional differences bring positive outcomes to innovation, due to the diversity. Moreover, Mondejar & Zhao (2013) encouraged researchers to determine levels of institutional differences based on the perceived entities, because the linkage between firms and the specific institutional environment remains underexplored. Schmid & Morschett (2020) added that a macro measure of institutional risk cannot assess the levels of risk that influence foreign divestments.
In addition, distance assumes a linear effect of institutional differences, since the concept does not consider the reverse impact, although foreign subsidiaries may gain experience over time or through changes during their operations (Luo & Shenkar, 2011;Zeng et al., 2013;Popli et al., 2016). Accordingly, distance could not reflect potentially changeable mechanisms that generate the influences of institutional differences. Overall, we conclude that distance may not be an appropriate measure by which to examine the influence of institutional differences, since the concept fails to reflect the multifaceted, multilevel, and contextual variation of institutional differences, resulting from the diverse industries from which the studies are often drawn. Therefore, in line with the POS lens that emphasizes the multifaceted effect of institutional differences, we employ friction, a metric proposed by Shenkar and his colleagues (Shenkar, 2001;2012;Luo & Shenkar, 2011), to evaluate the influences of economic and political differences.
Friction has garnered considerable attention as it considers not only the differences at the national level, but also the variation in organizational contexts when defining the differences (Koch et al., 2016;Li et al., 2019;Singh et al., 2019). In other words, levels of friction that foreign subsidiaries encounter will be specified depending on the national differences, and on the weighted domains of interaction points (Shenkar, 2011;2012;Luo & Shenkar, 2011).
IB scholars have employed friction through the cognitive-cultural aspect or language dimension (Orr & Scott, 2008;Luo & Shenkar, 2011;Joshi & Lahiri, 2015;Li et al., 2019;Singh et al., 2019). Yet, Zaheer et al. (2012) proposed that Shenkar's criticisms are applicable to other dimensions of institutional differences. In a similar vein, scholars have argued that the process of interaction and resistance leading to friction is equally validated in the context of other institutional dimensions (Orr & Scott, 2008;Popli et al., 2016). Hence, we expect friction may be more appropriate than distance to assess the differences in economic and political differences. Luo & Shenkar (2011) proposed several factors, that is, speed (V), sequence (G) of foreign expansion, and number of contact points or contact surfaces (N) to define levels of firm interaction. They argued that those factors reflect how foreign firms develop their operations in host countries, considering accumulated experiences and changes in expansion space, sequence or number of interaction points (Luo & Shenkar, 2011;Li et al., 2019). More precisely, N represents the degree of interaction that firms have in foreign countries. When N is higher, firm dependence on local resources is higher. Similarly, the level of friction is  (1) foreign divestment studies related to institutional differences, focusing more on economic and political differences or relevant indicators; (2) key studies by leading scholars in the field; (3) studies that show continuous research stream of foreign divestment, more up-todate papers. **For clarity, the conclusions presented in the 'Findings (adjusted)' column represent the effect of economic and political distance on foreign subsidiary divestment. For papers where the dependent variable was subsidiary survival, the original results have been adjusted accordingly. For example, if the relationship between economic distance and subsidiary survival in an article is reported as negative and significant, we report the relationship in this table as positive and significant.
also higher as a result of heterogeneous institutional requirements across locations. In addition, when MNEs adopt a lower speed of foreign expansion (V), they will better align their experiential knowledge with host-country risks and uncertainty. Consequently, the MNEs encounter lower levels of friction. Luo & Shenkar (2011) further argued that foreign subsidiaries face greater liabilities of foreignness at the initial stage of foreign expansion (G), and, thus, escalating levels of friction. Moreover, at this initial stage, it is difficult and costly for foreign subsidiaries to collect, analyze and interpret relevant information pertaining to host markets, given their lack of local experience. Collectively, the three factors reflect levels of firms' interaction with different contexts, playing a crucial role in defining levels of friction. It is also worth noting that economic and political friction values, as combinations of distance and firm interaction, are changeable, because levels of firm interaction (i.e., N, V, G) are changeable. Elaborating further on the concept, we explain that a simple combination of low levels of economic or political distance with low levels of firm interaction will generate low levels of economic and political friction, respectively. By contrast, high levels of economic and political distance combined with high levels of firm's interaction will generate high levels of economic and political friction. We further propose that while previous findings on effects of economic and political differences are essentially based on the distance metric, as opposed to the friction metric, mechanisms leading to economic or political distance' effects should broadly apply to friction. In arguing that the friction approach is superior to the distance approach, the former explicitly builds upon the latter. As elaborated above, the key distinctions between distance and friction are three firm-level factors: the speed, sequence and overall amount of international expansion.

Economic friction and foreign divestment decisions
Prior studies have confirmed the significant influence of economic differences on foreign divestments (Tsang & Yip, 2007;Pattnaik & Lee, 2014;Kang et al., 2017). Considering both the benefits and disadvantages of economic difference (i.e., Gaur & Lu, 2007;Malhotra et al., 2011;Mallon & Fainshmidt, 2017), we propose that economic friction is not always disruptive, and that friction may offer several benefits. Precisely, we theorize that economic friction influences FD probability as a U-shape for two critical reasons.
First, integrating insights from the institution-based view, resourcebased theory, and transaction cost economics, previous studies show that when operating in economically different countries, foreign subsidiaries gain various advantages, namely economic arbitrage (Gaur & Lu, 2007), and get access to new and more similar markets (Evans & Mavondo, 2002;Demirbag et al., 2011;Hutzschenreuter, Kleindienst & Lange, 2014;Liou & Rao-Nicholson, 2019). Further benefits may arise from ownership advantages (Dunning, 1988), and assets of foreignness, that is, unique resources, capabilities or opportunities only available to foreign firms (Edman, 2016;Mallon & Fainshmidt, 2017). We argue that, importantly, these advantages strengthen as level of economic friction increases. The advantages increase the benefits that foreign subsidiaries receive, and supply a motive for staying longer in distant countries. Those benefits accruing to MNEs for pursuing internationalization strategies or exploiting local resources would otherwise be lost (Song & Lee, 2017). It is worth noting that foreign subsidiaries always incur extra costs and take more risks to explore and exploit added benefits, while the advantages are not always available (Gaur & Lu, 2007;Popli et al., 2016). Accordingly, there would be a point at which increased costs were higher than achieved benefits, leading to a reverse net impact of economic differences (Gaur & Lu, 2007).
Second, as noted above, foreign subsidiaries may not gain benefits without committing to initial costs. Past research has highlighted that foreign subsidiaries incur organizational and transaction costs to set up business in a foreign country, and start exploiting or exploring local resources (Meschi et Kang et al. (2017) argued that since MNEs have less information about host markets, there will be start-up costs to acquire such information. Furthermore, when foreign subsidiaries want to exploit more resources, they are likely to have more interaction with the local context. Hence, they would incur higher ex-ante and ex-post costs and risks that create more difficulties (Malhotra et al., 2011;Popli et al., 2016). In sum, we propose a curvilinear relationship between economic friction and FD.
Precisely, at low levels of economic friction, foreign subsidiaries have to make an initial financial investment to set up their operation, as they encounter degrees of uncertainty in the host institutional environment. However, there will be several benefits to offset these costs and risks, and well-prepared strategic plans will further ameliorate the situation (Gaur & Lu, 2007;Mallon & Fainshmidt, 2017). But this relationship is not stable, and once level of economic friction rises above a certain threshold, which the achieved benefits do not offset the increased costs, the cost-benefit relationship is reversed.
Hence, once above that threshold, higher levels of economic friction create more difficulties for foreign subsidiaries to overcome, and they face more challenges to identify and satisfy customers' needs and preference (Berry et al., 2010;Kang et al., 2017). On encountering high levels of economic friction, subsidiaries also find it harder to apply accumulated experience from previous investments, due to greater contextual variation among previous experiences (Zeng et al., 2013;Popli et al., 2016). Furthermore, subsidiaries have to incur extra costs to leverage local resources, although the resources are narrower and not always available (Gaur & Lu, 2007). Hence, higher levels of economic friction escalate FD probability.
We integrate the aforementioned discussion in Fig. 1, following Haans, Pieters & He (2016), and contend that the combination of the benefits and costs that foreign subsidiaries are subject to in economically different countries will result in a U-shaped effect on foreign divestments. Precisely, at lower levels of economic friction, the benefits of exploiting and exploring local opportunities will increase, albeit at a decreasing rate, leading to a negative effect on foreign divestment. However, when levels of friction increase beyond a certain threshold, the relevant costs for foreign subsidiaries to access and diffuse the benefits grow at an increasing rate, thus, yielding a positive effect on divestment. Taken together, we propose the following: Hypothesis 1: Ceteris paribus, the relationship between economic friction and the likelihood of foreign divestment is a U-shape.

Political friction and foreign divestment decisions
Past research has highlighted the significant effect of political differences, as the most impeding factor, on MNEs' IB strategies, and particularly on foreign divestments (Kang et al., 2017;Sartor & Beamish, 2020). We apply political friction as a combined measure of political distance and firm interaction with national political differences, in order to examine influence of political differences on foreign divestment.
Elaborating on the nature of political differences, we argue that political friction increases the likelihood of foreign divestments. Gaur & Lu (2007) stated that foreign subsidiaries threaten their own existence, due to the risks of governmental appropriation and local hostile pressure in host countries. Dhanaraj & Beamish (2009) also argued that political differences place foreign subsidiaries in a disadvantageous position, due to the complexity and lack of transparency of business regulations, or constraints relating to repatriating profits, obtaining local financing, and transfer pricing. Similarly, Pattnaik & Lee (2014) found that differences in political regimes and uncertainty in regulation adversely hampered the efficiency of MNEs' operations. The differences magnify agency costs exponentially, due to the opportunistic behavior of managers and other local collaborators, and outweigh the benefits that MNEs derive from internationalization (Chao & Kumar, 2010;Kang et al., 2017). Meschi & Riccio (2008) further showed that when political uncertainty increases, foreign subsidiaries are less likely to access local benefits to achieve initial entry objectives, since they have lower bargaining power.
Elaborating on the friction concept, we argue that if foreign subsidiaries have more political interaction with the host country, they are influenced more strongly as they need to comply with more political requirements for their operations, leading them to be more prone to divestment (Gaur & Lu, 2007). Similarly, if subsidiaries are first entrants, they need to deal with higher degrees of liabilities of foreignness and face greater pressure from local groups, yielding higher levels of political friction (Slangen & Hennart, 2008a). By contrast, subsidiaries receiving support from sister companies already present in the host country, may not be subject to high levels of friction (Kim, Lu & Rhee, 2012).
We further emphasize that, the effect of political friction on divestment is linear rather than curved, for three critical reasons. First, political differences constitute the greatest challenges for MNEs, while political advantages may be not available to foreign entrants (Chao & Kumar, 2010;Giambona, Graham & Harvey, 2017;Lorenz et al., 2018;Graafland & Noorderhaven, 2020). For instance, prior studies show that political differences may not produce arbitrage opportunities, due to the dominant powers of host political agents, and higher adaptation costs (Cannizzaro, 2019;Sartor & Beamish, 2020;Witte, Burger & Pennings, 2020). Similarly, Giambona et al. (2017) proposed that MNE managers tend to avoid, not simply reduce investments, and exit host countries with high political risks, since political risk is more severe than other forms of risk. Second, Dhanaraj & Beamish (2009) proposed that among the FD studies there is an implied assumption that when subsidiaries exist in foreign countries, they have already incurred start-up costs to overcome political hurdles, and any favorable change in those hurdles is unlikely to create new opportunities that will facilitate their survival.
Third, we propose that external stakeholders, i.e., customers, local interest groups and rivals, would criticize the MNEs and local government, were there to be negotiations between them relating to unique support or political changes available only to foreign firms, that is, lower local taxes, favored support for transfer pricing, and other favors. Hence, MNEs will be less likely to allow their foreign units to involve themselves in such negotiations. For instance, Sartor & Beamish (2020) showed that if MNEs have to deal with higher levels of political corruption, they will find it harder to manage and regulate subsidiaries' behaviors, leading to a higher probability of divestment. In sum, we argue that political friction creates difficulties and constraints for firms, leading to a higher propensity for divestment. Accordingly, we propose the following: Hypothesis 2: Ceteris paribus, the relationship between political friction and foreign divestment is positive.

Moderating effect of subsidiary ownership levels
The FD literature proposes that depending on different ownership levels, that is, IJVs or WOSs, foreign subsidiaries may encounter different levels of institutional differences, which may in turn bring additional uncertainty and complexity, leading to different propensities for foreign divestment (Makino & Beamish, 1998;Dhanaraj & Beamish, 2004;Lu & Hébert, 2005;Gaur & Lu, 2007;Meschi et al., 2016;Kim & Kim, 2018). Proposing the friction concept, Luo & Shenkar (2011) echoed the call to pay attention to the moderating effect of several factors that modify levels of friction. Ownership is one such moderator. However, this proposal lacks empirical evidence. In this respect, our study provides an ideal research setting to investigate the moderating effect of ownership levels on the economic and political frictionforeign divestment relationship.
The extant literature has examined the direct impact of ownership levels on divestment, yet the findings are equivocal. For instance, Lu & Hébert (2005) proposed that IJVs could take advantage of local partners, to alleviate differences in host countries, gain a better understanding of the external environment, and make better strategic decisions, leading to lower rates of divestment. Similarly, researchers argue that IJVs have higher chances of survival than WOSs, because IJVs can combine the strength of each partner, broaden product diversification, achieve economies of scale, enhance capacity to adapt to external uncertainty, and, reduce legitimacy asymmetry (Lu & Hébert, 2005;Meschi et al., 2016;Kim & Kim, 2018).
Contrariwise, Schmid & Morschett (2020) argued that with higher levels of ownership (i.e., WOSs), MNEs could promote their own executives to key positions in foreign subsidiaries to support the decisionmaking process. In the same fashion, WOSs have been confirmed to have lower rates of divestment, due to the avoidance of complications in the decision-making process, of declining efficiency, or of technological knowledge leaking to other firms (Leung, 1997;. Furthermore, because of higher equity ownership levels in WOSs, MNEs may input more attention and resources, thus decreasing the propensity of divestment (Gaur & Lu, 2007). Notably, a few researchers have highlighted that the fate of divestment is not likely related to ownership levels per se, but relate to entry motivations (Makino & Beamish, 1998;Lu & Hébert, 2005;Makino, Chan, Isobe & Beamish, 2007;Papyrina, 2007;Meschi et al., 2016). As noted, these previous findings pertain to a direct impact, rather than emphasizing the moderating impact of ownership levels on economic and political differences.
In their discourse on friction, Luo & Shenkar (2011) argued that IJVs involve higher levels of friction due to the greater scope for operational and managerial blending. When the relationship between IJVs' partners is complex, the interactions between entities become more frequent, multifaceted, and complicated. In line with this, we theorize that IJVs encounter higher levels of interaction with different contexts as there are two layers of interaction involved, that is, national and firm levels, while WOSs interact at only the national level. In other words, while IJVs need to handle home-host differences as a business system and with differences between partners, WOSs have to address only the differences between countries (Gaur & Lu, 2007;Hennart & Larimo, 1998;Makino & Neupert, 2000).
We acknowledge that with support from local partners, IJVs may involve lower levels of external interaction, while WOSs may incur initial costs in order to compromise on economic and political requirements in host countries. However, elaborating on the friction perspective, we argue that WOSs may manage their levels of friction at an acceptable rate, by gradually increasing levels of interaction with the external environment and accumulated experience (Zeng et al., 2013). In addition, they could maintain sustainable growth based on their own accumulated experience, and developed local networks with an external environment (Gaur & Lu, 2007). Further, in IJVs this interaction would likely increase when foreign entrants want more local business transactions, or at later stages of their development (Makino & Neupert, 2000;Gaur & Lu, 2007). In addition, WOSs may avoid social conflicts between partners due to their sole ownership structure or dominant powers over local partners (if any) (Dhanaraj & Beamish, 2004;Gaur & Lu, 2007;Schmid & Morschett, 2020). In contrast, IJVs may accelerate the negative outcome of "being different" due to higher levels of conflict in management teams, and complicated decision-making processes (Li, 1995;Luo & Shenkar, 2011). We summarize the key differences in interaction levels of IJVs and WOSs in Table 2. In sum, we argue that the curved effect of economic friction on divestment will be flatter, and the positive effect of political friction on divestment will be weaker among WOSs than among IJVs. Accordingly, we propose the following: Hypothesis 3: The U-shape relationship between economic friction and foreign divestment probability will become flatter in the case of WOSs compared to IJVs.
Hypothesis 4: The positive relationship between political friction and foreign divestment probability will be weaker in the case of WOSs compared to IJVs.

Sample
The empirical data for the study are based on Finnish MNEs' foreign direct investments (FDIs) in the manufacturing sector from 1970 to 2010, the status of which were checked at the end of 2016, itself used as the cut-off year to avoid the two-year honeymoon bias effects (Gaur & Lu, 2007;Wang & Larimo, 2020). We collected the Finnish MNEs' information from the Thompson and ORBIS databases, and performed a systematic analysis of the investing firms' annual reports, press releases, data gathered in FDI surveys, and direct contact, to identify divestment of foreign units.
Finnish MNEs constituted a particularly good research context for three critical reasons. First, Finland is among the most competitive of 140 ranked countries, despite its small size in the global arena (Global Table 2 Differences in interaction levels between WOS and IJV.

Definitions
Major equity (i.e. more than 95%) of firms are belonging to a parent from a home country.
A joint equity (<95%) shared between a foreign parent firm (or many) with a local parent firm (or many).

Managing legitimacy
Involving mainly external legitimacy because the WOS is invested by only a parent firm (do not involve internal legitimacy between partners) (Kim & Kim, 2018).
Involving in both external and internal legitimacy because the IJV needs to respond to external environment (at host country) and to local parents, as a part of the local institution (Lu & Xu, 2006) (Zeng et al., 2013). Hence, at later stage of operation, WOS could enjoy higher levels of benefits, while uncertainty of being foreign at initial stage could be offset by accumulated experiences. In addition, WOS could minimize transaction costs arising from coordination problems ( Gaur & Lu, 2007).
IJV may reduce levels of foreignness, increase levels of local legitimacy, enjoy structural legitimacy and enhance the external legitimacy because of its status as partly a child organization of a local firm and of a legitimacy process ( (Li, 1995); (Lu and Hébert, 2005)). IJV also shortens the time and decreases costs needed to establish legitimacy in the local environment, easily obtains financial and human resources in local markets, develops networks with local suppliers and buyers; and accesses local partners' distinctive and network-based resources and capabilities (Lu & Xu, 2006;Papyrina, 2007). However, IJV involves higher transaction costs to ensure that collaboration produces maximum synergies (Papyrina, 2007). Furthermore, local partners may not help much if IJV has higher levels of interaction with local environment, that is, exploring more local resources and dealing with local customers, suppliers, etc. Accordingly, IJV needs to depend more on local support, leading to lower levels of bargaining power in decision-making process ( Luo & Shenkar, 2011). Interacting with internal environment (local partners) WOS may have faster and more efficient decision making processess and less scope of conflict because higher levels of equity ownership provide a parent firm with a greater degree of control over the systems, methods and decisions of its subsidiary (Dhanaraj & Beamish, 2004; There are more conflict in making organizational decisions as both foreign and local partners may fight for their preferences . Furthermore, managerial blending also leads to complicated decisionmaking processes (Gaur & Lu, 2007;Luo & Shenkar, 2011). IJV may also be (continued on next page) Competitiveness Report, 2018). Also, along with other Nordic countries, it accounts for a significant amount of outward FDI. Second, although our paper is not about culture, it is still important to consider cultural differences, as they reflect how people think, believe and behave, which later influences their strategic decisions (Hofstede, 1980;Koch et al., 2016). As Finnish national culture, based on the Hofstede and GLOBE frameworks, differs from those of the United States, Japan, and other non-Nordic countries, it makes our sample an excellent venue for investigating the influences of institutional differences on FD decisions. Third, the extant studies on foreign divestment focus mainly on US, Japanese, Korean or Chinese MNEs, while Western firms have received scant attention (Koch et al., 2016). Hence, as a good representative of the Nordic region, Finland could help fill this gap, and provide good knowledge on how firms in the region develop IB strategies.
In assessing the influences of economic and political friction, we attempt to compensate for the use of single-country data by including FDIs with diverse levels of firm interaction, that is, sequence and speed of internationalization, and number of subsidiaries. In addition, our data include subsidiaries with different ownership levels, in different industries, and with various years of international experiences. In sum, the diverse levels of interaction lead to different levels of friction, although Finland constitutes the only home country.
In total, we identified 2548 investments, with 1190 cases divested during the period. However, 75 divested cases were excluded because they were the consequence of corporate divestment, referring to closure or sell-off of whole MNEs. We further excluded 73 cases due to missing information concerning parent firm size or divestment years. Thus, our final sample comprises 2400 investments made by 310 MNEs in 65 different host countries, with 1042 cases divested during the period.

Dependent variable
The dependent variable in this study is the probability of foreign divestment, and is operationalized as a binary variable, coded as 1 if the investment is divested, and 0 otherwise. In addition, a subsidiary surviving at the end of the observation period would be treated as a rightcensored case (

Independent variables
The key predictors in our study are economic and political friction. Following Luo & Shenkar (2011), we construct economic friction (EF) and political friction (PF) at national and firm levels. Accordingly, we first follow Berry et al. (2010) and apply Mahalanobis distance to compute economic distance (ED) and political distance (PD) between countries (national level). Berry et al.'s (2010) scale is applied in this study because it measures ED and PD as a set of multi-dimensional indicators, which are confirmed to significantly influence MNE operations. Hence, the scale could reflect multifaceted aspects of different contexts. IB researchers have applied this scale in subsidiary divestment analysis and yielded significant findings (Kang et al., 2017;Pattnaik & Lee, 2014). Essentially, Berry et al. (2010) construct ED to reflect the differences in economic development and macroeconomic characteristics, including national income, inflation, export and import, while PD measures the differences in political stability, democracy, and trade bloc membership, measured by policy-making uncertainty, democratic character, size of the state, WTO membership, and regional trade agreements.
Then, at firm level, we evaluate firms' internationalization speed (V), sequence (G), and contact surface (N). Precisely, V is measured as the increase in the number of active foreign investments held by the parent firm in the corresponding year. G = [0;1] represents the sequence of internationalization of MNEs. G is computed such that the first investment a parent firm makes in a specific country is coded as 0. Subsequent investments in that country are coded as the ratio between the order of the investment and the maximum number of entries by any Finnish firm into that specific country, yielding a maximum value of 1. We also computed N as the sum of all the active foreign investments held by the parent firm in the corresponding year. Hence, friction is calculated as follows: where, e is constant and equal to 2.7183. To test our Hypotheses 3 and 4, we coded the subsidiary ownership levels (WOS) as a dummy variable, 1 for foreign subsidiaries with over 95 percent of equity, 0 otherwise Delios & Ensign, 2009).

Control variables
Our study controls for several variables pertaining to multilevel analysis, which have been confirmed in previous studies to significantly influence FD probability. More precisely, we control for several factors at host country levels, including: population, birth rate, literacy rate, corporate income tax rate, EU membership, and cultural friction. Those variables are popularly controlled in the previous IB studies to reflect MNEs' entry and exit decisions. At the parent level, we control for size, product diversification, research and development (R&D) intensity, number of foreign countries, and years of international experiences. At the subsidiary level, we control for the age and relatedness of the unit. Table 3 describes the definitions, measurements, and references of all the variables included in our models.

Analytical strategies
The Cox proportional hazards model (Cox & Oakes, 1984) is ubiquitous in the FD literature, as the model offers several advantages for analyzing hazard rate (Pattnaik & Lee, 2014;Kang et al., 2017;Song & Lee, 2017). One of the advantages is suitability for modeling different forms of event history data, since the model does not need an assumption of any functional form for the underlying hazard function, relative to parametric models (Song, 2014). As such, the hazard rate can be presented as log-linear functions of the various firm-and subsidiarylevel covariates (Kang et al., 2017). However, instead of using the basic Cox model, which assumes no unobserved heterogeneity or event dependence, we apply a frailty Cox proportional hazard model to test the likelihood of foreign divestment (Berry, 2013;Lee, Chung & Beamish, 2019). This frailty model accounts for cluster-specific homogeneities, the inherent nature that the subsidiary is nested in its parent companies (Austin, 2017;Lee et al., 2019). The frailty models also consider whether the same firm may suffer the hazard more than once as a result of unmeasured causes (Berry, 2013).

Results
The descriptive Pearson correlation in Table 4 displays a few high correlations among the variables, so the variance inflation factor (VIF) test was conducted to diagnose multicollinearity among the variables. The result shows that multicollinearity is not a problem among our variables (because the highest was 1.63 for MNEs' size). troubled not only by cultural differences between partners, but also by difficulties in sharing proprietary assets (Li, 1995).
We report the survival analysis results for the hypotheses testing in Table 5, including 12 models. Model 1 includes only control variables, Models 2-5 test the linear and non-linear effects of EF and PF on FD probability. Models 6-12 test the moderating effects of subsidiary ownership levels on the friction-divestment probability relationship. In general, our models are significant at high levels, and adding the moderating variables increases the significant value (p-value < 0.001).
Among the control variables, our results show that parent and subsidiary levelsfactors are significantly associated with foreign divestment rate, except for R&D intensity and number of foreign countries in which MNEs are operating. We also found that cultural friction, birth rate, literacy rate, and corporate income tax rate significantly influence FD probability, while population and EU membership are not relevant to divestment decisions.
In Hypothesis 1, we predicted that EF influences FD probability following a U-shape, such that low levels of friction decrease FD probability until a turning point, after which, adding friction increases the probability. As noted above, we follow Haans et al.'s (2016) proposal to examine the U-shaped relationship. Accordingly, we first checked the direction of linear and square-coefficients. Second, we examined whether the slopes on both sides are significant, and, third, we examined whether the turning point is within the data range. Our analysis in Model 3 shows that the EF coefficient is negative (β = − 2.097, p-value < 0.001), while the squared term is positive (β = 1.334, p-value < 0.01). Next, we checked the significance of the negative and positive slopes of the U shape, using the following formula: β 1 + 2*β 2 *X L and β 1 + 2*β 2 *X H where β 1 and β 2 are the estimated coefficients of EF and its squared term, respectively, while X L and X H represent the lowest and highest EF values in the data range, respectively. In the current data, the minimum value of EF is 0, while the maximum value is 1.49. We found that at the low end, the slope is negative and significant (− 2.097, p-value < 0.05), and at the high end, the slope is positive and significant (1.878, p-value < 0.05). We then estimated the turning point of the EF impact (asβ 1 /2* β 2 ), and confirmed that the turning point (0.786) is well within the data range. Hence, we can reasonably confirm the existence of the U-shaped relationship between EF and foreign divestment probability, supporting Hypothesis 1.
To test the relationship between PF and FD probability (Hypothesis 2), we included PF in Model 4. Our result shows that the PF coefficient is significantly positive (β = 1.043, p-value < 0.001). We further added the PF squared term in Model 5 to test a potential U-shaped relationship between PF and FD probability. The empirical results do not support the curved shape. Hence, Hypothesis 2 is supported, meaning that PF is positively associated with FD probability. We further plotted the relationships of EF and PF on foreign divestment probability in Figs. 2 and 3. In Fig. 2, we show that EF influences the divestment probability, following the U-shaped form, while Fig. 3 indicates that PF increases FD probability.
We proposed in Hypothesis 3 that WOSs flatten the U-shaped relationship between EF and divestment probability. Therefore, we continued incrementally adding variables (WOS and the interacting measure of WOS with EF, linear and squared values) in Models 6-8. The likelihood value was significant at p < 0.001, showing that the model was a good fit. Precisely, our results show that WOS is negative and significant. Hence, we confirmed that WOSs are less likely to be divested Accessing the differences in income, inflation, export and import turnover between countries by applying friction ( Luo & Shenkar, 2011) (Östermark, 1998Dharmapala, 2014) 20. Political friction Accessing the differences in policy-making uncertainty,

Variables Definitions and measures References
democratic character, size of the state, member of WTO and regional trade agreement between countries by applying friction (Luo & Shenkar, 2011). (Walgrave & Nuytemans, 2009;Xu, Xu, & Yuan, 2013) than IJVs. This finding is consistent with previous studies in the FD literature (Gaur & Lu, 2007;Tsang & Yip, 2007;Demirbag et al., 2011;Song & Lee, 2017). Nevertheless, interacting effects between EF and IJV are not significant in Models 7 and 8. Hence, Hypothesis 3 is not supported. We followed the same path to examine Hypothesis 4, adding the interaction terms between WOS and PF in Models 9-12. The terms remain statistically non-significant in those models. Therefore, Hypothesis 4 is not supported. In sum, our results show that while WOS decreases foreign divestment probability, it does not moderate the influence of EF and PF on foreign divestment propensity. We propose two critical explanations for the non-significant effect of ownership levels on the friction-foreign divestment relationship. First, Gaur & Lu (2007) confirmed that compared with normative differences (i.e., informal, cultural), regulative differences (i.e., economic, political) are more clearly stated and easier to comprehend, regardless of ownership levels. Hence, we suggest that MNEs do not need to set a specific mode to enter countries with larger differences in regulative institutions. Second, Luo & Shenkar (2011) proposed that levels of cultural friction may be different with or without considering organizational culture, which differs for WOSs and IJVs. Nevertheless, this is less likely to involve EF and PF, since organizations seem not to have their own organizational economic or political institutions. Accordingly, it is less likely that WOSs and IJVs involve different levels of EF and PF.

Post-hoc test
We conducted several robustness tests to consolidate our findings. First, looking to compare the distance and friction concepts when measuring ED and PD levels, we replicated our main models using the Mahalanobis distance concept in place of friction. As such, in each instance, we used the standard ED and PD metrics in place of the EF and PF metrics. The results are presented in Models 13-16 shown in Table 6. We found that ED has a significant negative effect on foreign divestment (Model 13), while the squared term of this variable is not statistically significant (Model 14). Hence, in line with Tsang & Yip (2007), and Demirbag et al. (2011), we confirm that ED decreases divestment probability. Further, we plotted the relationship between ED and divestment probability in Fig. 4. Similarly, we replaced PF with PD in Models 15-16. Notably, PD is non-significant in both models. This interesting result shows that when operating in countries with higher levels of national political difference (i.e., PD) without engaging in any political interaction, foreign firms are less likely influenced by political differences. While this finding is consistent with Meschi & Riccio (2008), and (Liu & Li, 2020), we highlight that friction is superior to distance in evaluating ED and PD effects, because it reflects the influence of firm interaction even when national differences (i.e., PD) are not relevant.
Moreover, we plotted the relationship between EF and the hazard ratio of divestment with the baseline of the survival model (h0), to describe in Fig. 5 the effect of EF on divestment probability, accounting for time perspective. As such, Fig. 5 includes three dimensionseconomic friction, subsidiary hazard ratio, subsidiary ageand depicts a relationship that is initially negative, then positive as EF increases. The U-curve is consistent with differing levels of divestment probability over time.
Next, we applied discrete time logit models which are used in the FD literature (Delios & Beamish, 2004). We find that the results, reported in Table 7, are largely consistent. Furthermore, as WOS is categorized at 95% level in the main tests, we followed Dhanaraj & Beamish (2004), and Papyrina (2007), and recategorized ownership levels at 80%. Precisely, we recoded the WOS variable to show any investment that has ownership levels higher than or equal to 80% as 1, and 0 otherwise. Then, we re-ran Models 6-12 to test Hypotheses 3 and 4. The results, reported in Table 8, are still robust, meaning that WOS, while it decreases the likelihood of divestment, does not moderate the effect of EF and PF on foreign divestment probability. Finally, we examined the divestment rate between subsamples with different economic Table 4 Descriptive and Pearson correlation.
Correlation is significant at the 0.01 level (2-tailed).

Table 5
Survival analysis of the foreign divestment probability for main hypotheses.  Robust standard errors shown in parentheses, *** p-value < 0.001; ** p-value < 0.01; * p-value < 0.05; + p-value < 0.1, fixed-effect for parent firms, years and industry. development levels based on the Organization for Economic Cooperation and Development (OECD) categories. The results are robust in the subsamples. For brevity, we do not report these results.

Discussion and conclusion
IB scholars recognize the influence of economic and political differences on FD probability. Nevertheless, the extensive discussion on the negative effects of the differences, combined with the oversimplification of distance-based measurement, has led to insufficient knowledge on the multifaceted effects of economic and political differences (Gaur & Lu, 2007;Jackson and Deeg, 2008) Aguilera & Grøgaard, 2019;Graafland & Noorderhaven, 2020). This research broadens our mindset regarding the effect of EF and PF, by elaborating on the POS lens. We further examine the role of ownership levels in modifying the friction-divestment relationship. Based on previous reviews on foreign divestment (Arte & Larimo, 2019; Coudounaris et al., 2020; Schmid & Morschett, 2020), our model also controlled for other known effects regularly included in FDIs and FD analysis.

Theoretical contributions
Exploring the influence of economic and political differences on foreign divestment, our study contributes to the extant literature in several ways. First, using the POS lens, our study reshapes the traditional perspective of institutional theory concerning the negative outcomes of institutional differences on MNE internationalization and foreign divestment. Our tenet is that each institution imposes different constraints on foreign subsidiaries and provides them with different resources, depending on the unique nature of the specific institutional environment, and the interaction levels that the subsidiaries encounter with the different context. Precisely, our study found that economic and political friction have different impacts on FD probability.
On the one hand, we confirmed that economic friction has a curvilinear effect on foreign divestment probability. This finding is consistent with the previous studies on the impacts of economic differences (Gaur & Lu, 2007;Wu, 2013;Fortwengel, 2017). More precisely, we find that at lower levels, economic friction is negatively associated with divestment probability. Economic arbitrage is a benefit that outweighs initial costs (Evans & Mavondo, 2002;Demirbag et al., 2011). Foreign   Robust standard errors shown in parentheses, *** p-value < 0.001; ** p-value < 0.01; * p-value < 0.05; + p-value < 0.1, fixed-effect for parent firms, years and industry.
subsidiaries could also overcome low levels of economic friction by generating well-prepared strategic plans, learning from accumulated experience, and exploiting local economic resources (Gaur & Lu, 2007). However, the positive outcome of economic friction is not stable, and once a certain threshold has been reached, the relationship is reversed. We argue that at higher levels of friction, economic arbitrage is narrower (Gaur & Lu, 2007). In addition, prepared plans may not be efficient, due to the lack of suitability and sufficiency of practical experience and knowledge about similar situations, while using stereotypes could lead foreign subsidiaries to failures in compromising on local legitimacy requirements (Zeng et al., 2013;Popli et al., 2016). Higher levels of economic interaction could also trigger more conflicts, and increase ex-ante and ex-post costs and risks (Malhotra et al., 2011). Hence, once a certain threshold has been reached, higher EF levels increase the divestment probability.
On the other hand, our empirical results confirmed the positive relationship between political friction and FD probability. In contrast to other institutions, political friction introduces serious impediments to foreign subsidiaries to comprehend and compromise (Dhanaraj & Beamish, 2009;Berry et al., 2010;Chao & Kumar, 2010). Political friction also increases conflict, leading foreign subsidiaries to fail in achieving legitimacy (Pattnaik & Lee, 2014;Kang et al., 2017). In addition, political friction results in more constraints on foreign subsidiaries, since host governments often hold greater powers (Cordero & Miller, 2019;Sartor & Beamish, 2020). Furthermore, MNEs and their foreign subsidiaries are more cautious about political benefits, especially when they need to deal with local governments to gain such benefits (Sartor & Beamish, 2020;Witte et al., 2020). Importantly, our findings confirm an implied assumption in the FD literature that for extant foreign subsidiaries, political differences and any favorable change in the political system are unlikely to create new opportunities to enhance their survival (Dhanaraj & Beamish, 2009).
Second, our study extends friction's application, and proves the validity of the concept in assessing the effect of economic and political differences. Elaborating on criticisms of the distance concept in terms of measuring differences (Shenkar, 2001;Luo & Shenkar, 2011), we apply friction to measure the economic and political differences. Our post hoc results further confirm that compared to the friction metric, using distance to measure the differences delivers less meaningful findings. Hence, we highlight the importance of considering firms' specific conditions in evaluating the influence of economic and political differences.  Third, we explore the role of ownership levels, and show that higher levels (WOS) decrease FD probability. Notably, we found that ownership levels do not significantly change the effects of economic and political friction on foreign divestment. This finding is interesting, albeit inconsistent with our expectations. We argue that compared to normative and cognitive (i.e., informal, cultural) differences, regulative (i.e., economic and political) differences are stated more clearly and, thus, foreign subsidiaries find them easier to overcome due to the availability of secondary information (Gaur & Lu, 2007). We therefore encourage IB scholars to delve deeper into the effect of ownership levels.

Managerial implications
The managerial implications of our findings are straightforward. First, by elaborating on the different natures of economic and political differences, we stress that MNE managers should be aware of the differences between institutional environments. This awareness would help MNEs generate better strategies to deal with specific institutional differences. Second, we encourage managers to carefully consider the combined effect of national differences and specific firms' interactions, that is, friction rather than distance. Furthermore, we urge MNE managers to nurture moderate levels of economic friction, since our findings indicate this friction has a U-shaped effect on FD probability. We also encourage managers to maintain low political friction, because it is positively associated with divestment probability.
Since we find that ownership levels do not moderate the frictiondivestment relationship, we encourage MNEs to build on other strategies focused on organizational prescription. That is, hiring experienced expatriates, managing levels of communication between headquarters and subsidiaries, and among subsidiaries, as well as educating personnel about local knowledge and social norms, in order to modify the effects of economic and political friction (Luo & Shenkar, 2011). These strategies may provide sufficient tools to alleviate friction (Luo & Shenkar, 2011;Sartor & Beamish, 2020). Robust standard errors shown in parentheses, *** p-value < 0.001; ** p-value < 0.01; * p-value < 0.05; + p-value < 0.1, fixed-effect for parent firms, years and industry.

Limitations and future research
Our study is not without its limitations. First, it focuses solely on the effects of economic and political friction. This limited range of dimensions could be addressed by examining the effects of other dimensions, namely, knowledge, financial, administrative, or demonstrative institutions, since MNEs communicate different aspects of the institutional environments in host countries (i.e., Pattnaik & Lee, 2014;Kang et al., 2017). Considering these factors is also important because our study confirms that different institutional environments have different effects on foreign divestments. Furthermore, with respect to institutional friction, we encourage future research to examine its influences on escape-based FDI (Witt & Lewin, 2007). This could develop our nuanced understanding of why firms decide to exit their home country.
Second, in line with Zimmerman & Zeitz (2002), and Jackson & Deeg (2008), we encourage future researchers to examine a specific mix of relevant institutional forces, as well as the interactions among institutional dimensions. For instance, future studies could examine the combined effect of cultural, economic and political friction, among others, on firms' internationalization, and particularly on foreign divestments. This could be extremely fruitful, as foreign subsidiaries often encounter friction with several institutions at divergent degrees simultaneously (Graafland & Noorderhaven, 2020).
Third, Luo & Shenkar (2011) proposed using friction to measure the influence of institutional differences in multilevel analysis, that is, at national, firm and individual levels. Our study examines the effect of economic and political friction defined at the national and firm levels. We acknowledge that with different levels of interaction at the individual level, that is, chief executives, top management, and expatriates (Sartor & Beamish, 2020), foreign subsidiaries may perceive and be influenced differently by institutional differences. As such, we encourage future research to focus on the effect of institutional differences at the individual level. IB researchers have further confirmed that MNEs follow regional expansion, rather than national borders, that is, economic cluster, cultural cluster, and geographic cluster (Arregle, Beamish & Hébert, 2009;Arregle, Miller, Hitt & Beamish, 2013;Demirbag, Glaister & Sengupta, 2020). Hence, we urge researchers to develop the friction concept at cluster levels, thus constituting the generalization of this concept in the IB literature. Furthermore, as there are several ways to access levels of firm interactions (i.e., Shenkar, Tallman, Wang & Wu, 2020), we encourage future studies to widely examine new measures of firm interaction, as an interesting way to further develop the friction metrics.
Our findings may also have limited generalizability, since they are based on a sample of Finnish MNEs and their foreign subsidiaries. While we claim that our friction concept, combining both national distance and firm interaction, could compensate for the single-country related issue, we urge future research to draw on diverse settings, including multiple home and host countries (Brouthers et al., 2016), when investigating the influences of economic and political friction. Furthermore, although our research does not provide support for the moderating effect of equity ownership levels, we acknowledge the significant effect of ownership strategies on different levels of communication, power structures, and staff assignments (Luo & Shenkar, 2011). This organizational prescription, a so-called "black box", has seldom been examined (Koch et al., 2016;Singh et al., 2019). However, due to the unavailability of data on staff or managerial communication, we could not examine the effects. In closing, due to a lack of information on subsidiary profitability or performance, and exit barriers, among others (Arte & Larimo, 2019; Schmid & Morschett, 2020), we could not control for these variables in our models. Therefore, we propose that future research should.  (  Robust standard errors shown in parentheses, *** p-value < 0.001; ** p-value < 0.01; * p-value < 0.05; + p-value < 0.1, fixed-effect for parent firms, years and industry.