1 Introduction

Global diversification is an increasingly common corporate strategy which involves carrying out product and geographical diversification (Kim et al. 1989; Hashai and Delios 2012; Kistruck et al. 2013). The interactions between these two dimensions have recently started to gain attention from researchers. Thus, Kumar (2009), Kistruck et al. (2013) and Benito-Osorio et al. (2014) have recently identified the existence of a simultaneous relationship between both variables pointing towards the possibility that selecting one of these two strategies restricts the quantity of resources available for the carrying out of the other. These studies, however, examine the relation between these two variables on the underlying assumption of a certain degree of homogeneity among the countries into which MNEs expand as they internationalize (Yang et al. 2013). This study seeks to advance one step further in the study of the relations between these two components of global diversification by incorporating political risk, a factor which sharply differentiates some locations from others and which has a critical effect on the firm’s strategy.

Thus the objective of this paper is to examine whether there exists a simultaneous relations between political risk and product diversification. More specifically, we argue that those firms with an investment portfolio in countries with a higher level of political risk can take advantage of their experience interacting with host-country institutions to enter new markets with fewer restrictions. This broader range of potential locations favors a higher degree of relatedness in the product diversification strategy by increasing the possibilities of achieving and the impact of economies of scope.

In turn, we argue that the higher the degree of relatedness in the firm’s product diversification, the greater the incentives to enter into markets where both political risk and potential returns may be higher. This is due to the lower costs of absorbing knowledge and adapting to new environments when there exists a greater similarity between the activities and needs of the various strategic business units of the firm (Kumar 2009). By contrast, when the level of product diversification relatedness is low and, therefore, the products in the firm’s portfolio are not related to each other and have very different requirements, experience gained in negotiations and interactions with host-country authorities and institutions will be less useful.

In order to verify our hypotheses, we empirically examine a sample of 119 Spanish MNEs. We purposefully focus on companies from Spain given the demonstrated relevance and positive impact of political risk in the location of companies in regulated industries in Latin America (García-Canal and Guillén 2008), the scope of the international expansion of MNEs across industries (Jiménez 2010), and their performance (Jiménez and Delgado 2012). Notwithstanding, this proactive approach towards political risk, particularly salient in their international expansion in the last decade (Jiménez 2010), has also led to some negative outcomes. For example, some foreign governments have recently taken direct measures against MNEs from Spain located in their territories. Argentina recently expropriated the part of the YPF oil company owned by the Spanish firm Repsol and the government of Bolivia did likewise with the local subsidiary of Red Eléctrica de España. Direct measures such as nationalization and expropriation and more indirect ones, such as forced re-negotiation of previously agreed conditions between host nations and MNEs continue to represent a serious threat to the interests of MNEs (Jiménez et al. 2014). This highlights the current relevance of an appropriate planning of the firm’s “non-market strategy”, especially in relation to the political context, in order to achieve the firm’s goals (Hillman and Hitt 1999; Hillman et al. 2004; Bonardi et al. 2006; Holtbrügge et al. 2007; Oliver and Holzinger 2008; Doh et al. 2012).

As Doh et al. (2012, p. 23) have emphasized, non-market strategies are inextricably and inexorably linked with the traditional strategy approach focused on different aspects of the market. At the same time, many of the theories and perspectives used in the analysis of market strategies could be adopted and incorporated in order to understand non-market strategies, leading to a broader and more integral approach than that usually employed in the field of strategic management.

Thus, by combining the “New Institutional Economics” approach (Levy and Spiller 1994; DeFigueiredo 1997; Henisz 1998; Zelner 1999; Jensen 2003) with the resources and capabilities-based view (Wernerfelt 1984; Barney 1991) this present study contributes to the literature on the institutional environment and its impact on the management of the firm. It does so by showing that experience dealing with political risk represents a key resource for MNEs which plays a determining role in corporate product diversification strategy by encouraging a greater degree of product relatedness. Therefore, market and non-market strategies do not constitute two separate sets of decisions independent one from the other. Rather their joint study provides a new and integral conceptual framework which should produce a better understanding of the strategic decisions adopted by the firm.

A second contribution lies in the empirical evidence obtained showing that firms must consider political risk and the political environment as a source of potential opportunities—in this case to obtain economies of scope and create and sustain competitive advantages in markets where competitors without experience in managing risk cannot enter- and not just as a threat or a restriction (García-Canal and Guillén 2008; Oliver and Holzinger 2008; Holburn and Zelner 2010; Jiménez and Delgado 2012; Jiménez et al. 2014).

The remainder of the paper is structured as follows: Sect. 2 reviews the literature on political risk and product diversification strategy and sets out the hypotheses. Section 3 describes the sample, model, dependent, independent and control variables. Section 4 covers the results and robustness tests and, finally, Sect. 5 sets out the main conclusions and limitations of the study and suggests future lines of research.

2 Literature Review and Hypotheses

2.1 Political Risk, Experience and Political Capabilities

Several authors have recently claimed that political risk is not only an exogenous variable but also, at least partially, it can be considered endogenous as MNEs can develop non-market strategies (i.e., strategies directed at factors external to the market) especially targeted to the political environment (Hillman and Hitt 1999; Hillman et al. 2004; Bonardi et al. 2006; Holtbrügge et al. 2007; Oliver and Holzinger 2008; Doh et al. 2012). This approach recognizes the possibility that MNE-host government relations involve opportunities for mutual benefit and are not just a zero-sum game (Luo 2001). This was precisely a limitation of what used to be the predominant perspective in the study of political risk; the bargaining power approach (Vernon 1971; Kobrin 1987). This approach builds on the idea that the existence of sunk costs reduces the firm’s negotiating power in comparison with that of the host government once the investment has been made. MNE’s bargaining power also decreases as local firms reduce their technological and management disadvantage. Eventually, a point can be reached where the host government may consider the benefits of appropriating the rents that the MNE generates by expropriating, nationalizing or making unilateral modifications to the business terms previously agreed (Kobrin 1987).

Recognizing that markets operate in a political, administrative, judicial, and social context, a new approach based on the analysis of political institutions has started to be employed to predict the level of risk as a function of the existence of constraints on the discretional behaviour of the authorities (Henisz 2000; Henisz and Zelner 2001). Building on the “New Institutional Economics” literature (Levy and Spiller 1994; DeFigueiredo 1997; Henisz 1998; Zelner 1999; Jensen 2003), this perspective acknowledges that the institutional context in which the firm operates affects its behaviour and that certain political regimes provide a better environment for investment. However, the impact of political hazards is not homogenous for all firms. Rather there is a notable degree of heterogeneity in terms of the political capabilities at the disposal of MNEs both at the subsidiary and the headquarters-level. Due to the different starting point, evolutionary paths and organizational attributes, the level of political capabilities and the ability to transfer them to different locations is unequal across firms (Eisenhardt and Martin 2000), which explains the propensity to invest in countries characterized by higher levels of political risk found in some firms (Lawton et al. 2013). This heterogeneity across firms may happen at the human capital, organizational structure and/or network relationships level (Lawton et al. 2013). Some MNEs can take advantage of organizational attributes, such as size or age, external relationships, groups of reference, available information, and their own previous experience in analogous situations in the home-country or in similar economies (Henisz and Zelner 2001; Henisz 2003). In fact, greater experience and exposure to political risk not only makes firms more conscious of possible threats, but it also incentivizes them to adopt proactive actions in order to take advantage of the situation such as developing and nurturing links with political actors to enhance their corporate political ties (Jiménez 2010; Sun et al. 2012).

Building on the resources and capabilities-based view (Wernerfelt 1984; Barney 1986, 1991; Peteraf 1993), political capabilities allow firms to interact with the authorities of the host country in more appropriate ways to achieve their strategic goals. They improve the firm’s ability to make a more accurate assessment of political risk, enhance negotiation, litigation and lobbying skills, choose better campaign contributions, form coalitions and political networks that facilitate access to information, resources and opportunities an even take advantage of corrupt systems (Boddewyn and Brewer 1994; Baron 1995; Hillman and Hitt 1999; Holburn 2001; McWilliams et al. 2002; Wan 2005; García-Canal and Guillén 2008; Jiménez 2010).

According to Lawton et al. (2013, p. 230), political capabilities can be defined as “organizational and strategic activities by which senior or acting representatives reconfigure, leverage and release political resources to achieve new resource configurations that enable the company to adapt to, anticipate or even shape changes in the corporate political environment”. They allow the firm to take advantage of changes in policies or regulations in order to generate unique income sources and, ultimately, to create and maintain value (Oliver and Holzinger 2008).

Together with ties to influential actors in a given jurisdiction (Henisz and Delios 2004; Faccio 2006; Siegel 2007), experience in risky countries has been one of the main sources identified in the literature to foster the development of political capabilities (Delios and Henisz 2003a, b). Firms learn from their own experience as repeated engagement in an activity allows firms to infer from previous outcomes and adjusts actions and routines (Cyert and March 1963; Levitt and March 1988). Markides and Williamson (1996) claim that the most important way to obtain new strategic resources that are costly to trade is to accumulate them through experience. As Zollo and Winter (2002) underline, experience accumulation is an important mechanism through which organizations develop capabilities, in this case of a political nature, making subsequent investments in politically risky countries easier to those firms which have already successfully invested in locations with a similar institutional configuration (Delios and Henisz 2003a). This has been the case, as various empirical studies have confirmed, of the electricity sector in the US (Holburn 2001; Holburn and Zelner 2010), the air transport sector in Europe (Lawton and Rajwani 2011; Lawton et al. 2013) the international expansion of Spanish MNEs (García-Canal and Guillén 2008; Jiménez 2010; Jiménez and Delgado 2012; Jiménez et al. 2014) and FDI flows from the South of Europe to North Africa and Eastern Europe (Jiménez 2011).

2.2 The Influence of Experience Dealing with Political Risk on the Degree of Relatedness in Product Diversification

Related product diversification consists of the incorporation of new businesses which are to some degree related to the firm’s existing portfolio (Ansoff 1965). Relatedness occur when a common skill, resource, market, or purpose applies to each business (Rumelt 1974, p. 29) and allow the firm the firm to achieve costs and/or differentiation advantages in one or more of its divisions (Markides and Williamson 1996). Various authors have suggested that related product diversification offer firms several advantages such as synergies (Pennings et al. 1994), easier exchange of knowledge between units (Zahra and George 2002), lower transaction costs and a more efficient strategic control (Chang and Wang 2007). In this paper we will use the term “economies of scope” to comprehensively refer to these potential benefits that firms may obtain when their businesses are related (Teece 1980, 1982; Panzar and Willig 1981; Tanriverdi and Venkatraman 2005).

As emphasized by the resources and capabilities-based view of the firm (Wernerfelt 1984; Barney 1986, 1991; Peteraf 1993), the key factor for obtaining sustainable competitive advantages lies in the possession of valuable, unique, non-imitable and non-substitutable resources and capabilities. According to this approach, those firms that focus their diversification efforts on the optimal utilization of their resources and capabilities will be able to build a sustainable competitive advantage and make more profit on a long-term basis (see e.g., Palich et al. 2000; Mayer and Whittington 2003; Park 2003; Tanriverdi and Venkatraman 2005; Miller 2006; Colpan 2008; Bausch and Pils 2009).

Precisely, political capabilities constitute a resource which can play a vital role in the firm’s strategy and the development of competitive advantages (McWilliams et al. 2002). The organization of political capabilities is mainly based on specific heuristic processes for interacting with political institutions and influencing the policy choices (Eisenhardt and Martin 2000). The content of these capabilities is unique to each particular firm and arises from the different starting point, evolutionary trajectory and organizational attributes throughout the company’s networks at home and abroad. Following Makadok (2001, p. 389) capabilities are “a special type of resource, specifically an organizationally embedded non-transferable firm-specific resource whose purpose is to improve the productivity of the other resources possessed by the firm.” Political capabilities are the result of deliberate investments in organizational structures and systems (Zollo and Winter 2002) that allow firms to reduce environmental uncertainty and transaction costs and to increase the amount of value generated over time and long-term sustainability (Hillman and Zarkhoodi 1999; Oliver and Holzinger 2008). They are characterized by a high level of tacitness and embeddedness that can be made use of in different jurisdictions (Holburn 2001; Henisz 2003). This high level of tacitness and embeddedness of political capabilities facilitates maintaining competitive advantages over time as they reduce the possibility of competitors to imitate them (Amit and Schoemaker 1993). However, it also hinders the existence of a market for internally-developed political capabilities, which in turn encourages the use of diversification as a means for the firm to leverage its networks and operational routines (Kistruck et al. 2013, p. 505) and to expand its stock of strategic resources and create new ones more rapidly and at lower cost than rivals (Markides and Williamson 1998).

Specifically, we argue that those MNEs that have develop greater political capabilities as a result of their experience in interacting with host-country authorities and institutions, will be able to make investments in more favorable conditions and in a wider range of countries than competitor firms that lack such capabilities (García-Canal and Guillén 2008; Jiménez 2010). As Boubakri et al. (2013) point out, firms with close relations with the political authorities can make less conservative investments and thus be less risk-averse in their international expansion. The higher multinational flexibility (Kogut and Kulatilaka 1994; Huchzermeier and Cohen 1996; Song 2013) means that the opportunities to benefit from economies of scope in foreign markets are increased as a greater number of potential markets are available, allowing the company to select the most suitable ones not only to obtain them, but also to maximize their positive impact. In other words, as firms are subject to fewer restrictions arising from high and non-tolerable levels of political risk, they are able to choose upon a wider range of potential location and select those where they can maximize the potential advantages of related diversification.

Conversely, when a firm lacks experience in dealing with political risk, its maximum tolerable level of political risk and flexibility will be lower. Consequently, the number of possible markets where it can invest will be reduced. This lesser variety of available locations makes it harder to find markets where economies of scope can be obtained. Instead, the high level of coordination and interdependence between activities involved in related diversification may make it easier that problems in one business unit also affect others within the firm (Jones and Hill 1988; Chang and Wang 2007; Benito-Osorio et al. 2014). Coordination costs can actually offset economies of scope and make related diversification unattractive (Rawley 2010; Zhou 2011). This disadvantage may raise the attractiveness of an unrelated product diversification strategy in which each subsidiary acts as an independent unit with the aim of reducing administrative bureaucracy (Porter 1985) and increasing autonomy so as to allow more rapid decision making when changes occur in the market (Chang and Wang 2007). These arguments lead us to propose the following hypothesis:

Hypothesis 1: The higher the exposure to political risk in the foreign direct investment portfolio of a MNE, the higher the level of relatedness there will be in its product diversification.

2.3 The Influence of the Level of Relatedness of Product Diversification on Political Risk

When carrying out a product diversification strategy, a firm needs to analyze the current resources and capabilities which it can rely on for its expansion as well as the key features of the market into which it intends to expand (Winter and Szulanski 2001; Kumar 2009). In addition to the characteristics of the customers, suppliers and competitors, some of these critical features of the target market are those relating to government, institutions and culture of the destination country (Kumar 2009). Thus, when product diversification is combined with internationalization, the firm needs to bear in mind these idiosyncratic features of the locations where it is investing because learning and adaptation to the local environment are essential for success (Bartlett and Ghoshal 1989; Mitchell et al. 1992; Zaheer 1995; Delios and Beamish 1999).

The firm’s ability to learn and adapt to new markets largely depends on its absorptive capabilities (Cohen and Levinthal 1990; Vermeulen and Barkema 2002). Thus, the ability to absorb, assimilate and use new knowledge in commercial activities functions as an incentive for the company to grow and exploit new business opportunities (Kumar 2009). Conversely, the lack or scant development of absorptive capabilities may constitute a barrier to or a restriction on diversification (Kumar and Seth 2001; Kumar 2009).

The absorptive capabilities of the firm depend on the cognitive abilities of its managers (Cohen and Levinthal 1990; Zahra and George 2002). To the extent that humans are subject to bounded rationality, thus restricting their absorptive capabilities, the company may be unable to completely understand and adapt to the characteristics of any given market. Similarly, absorptive capabilities are also influenced by the slack available in the company. A great level of slack may facilitate devoting resources to learning about new markets (Penrose 1959; Teece 1982; Mahoney and Pandian 1992). However, the amount of slack is not unlimited, restricting the amount of knowledge that the firm can absorb (Kumar 2009).

Several authors have pointed out that absorptive capabilities have a crucial impact on the firm’s product diversification, making unrelated diversification more costly (Ramanujam and Varadarajan 1989). As unrelated diversification is usually based on activities with quite different requirements, the greater costs related to knowledge absorption and adaption to a new business environment may offset the advantages and lead to a lower performance compared to related diversification (Kumar 2009).

As a higher degree of relatedness among activities reduces the costs of absorbing new knowledge and adapting to a new environment, we argue that firms with a higher degree of relatedness in their product diversification will have incentives to internationalize in a location portfolio with a higher level of political risk than MNEs following an unrelated diversification approach. As the activities and needs of the strategic business units are relatively more similar, the experience in previous investments becomes more useful in subsequent investments. The greater similarity in the requirements of the activities in related product diversification simplifies the task of identifying, analyzing, selecting and negotiating the entry into a new market, thereby reducing uncertainty and risks for the firm. Consequently, firms are able to invest in markets where, according to the classical financial and economic theory based on the risk aversion of investors (e.g., Brealey et al. 2005), there exist higher expected levels of returns, as a result of higher political risk. In fact, as Schuler (1996) and Hillman and Hitt (1999) point out, firms with a higher degree of related diversification can anticipate greater benefits from a closer interaction with the authorities, as their lobbying efforts can concentrate in a narrower policy domain that is likely to have synergies across business units.

By contrast, unrelated-product-diversified firms are likely concerned with a wider diversity of policy domains, face more difficulties acquiring specialized knowledge both about relevant policy issues and policy-makers in charge of them, and find fewer synergies across business units from long-standing relationships with governments (Schuler 1996; Hillman and Hitt 1999). In addition, as the transfer of skills and knowledge among business units is more difficult (Palich et al. 2000, p. 168), adaptation to new environments will be more difficult and knowledge gained through experience will be less useful. In this case the negotiation and interaction with host country institutions have fewer similarities with those of previous investments, since each investment’s requirements vary greatly. This will cause firms with a higher degree of unrelated diversification to follow a more conservative foreign expansion and restrict foreign investment to a range of locations made up of countries with low political risk levels in order to reduce absorption, learning and adaptation costs.

These arguments lead us to formulate the following hypothesis:

Hypothesis 2: The greater the degree of relatedness in the product diversification of a MNE, the greater will be the exposure to political risk in its portfolio of foreign direct investments.

3 Methodology

3.1 Sample

The sample includes all MNEs with more than 250 employees registered at the Instituto de Comercio Exterior (ICEX) (Foreign Trade Institute) and the website (http://www.oficinascomerciales.es), managed by the Spanish Ministry of Industry, Tourism and Trade, excluding those with a holding company controlling at least 50.01 % of their share capital.

The sources from which the data were obtained were the SABI database [Sistema de Análisis de Balances Ibéricos (Iberian Balance Sheet Analysis System)] and the annual accounts of the firms themselves. Unfortunately, this database does not include companies in the banking industry so these companies had to be removed from the sample. Also, we excluded single-business firms to be able to build our product diversification relatedness index, which requires more than one Standard Industrial Classification (SIC). The final sample consisted of 119 Spanish MNEs.

3.2 Model

The statistical technique chosen to analyze the mutual influence between product diversification and political risk was the simultaneous equations model. Specifically, we employed the “3 stage least square” (3SLS) simultaneous equations estimation. By following this procedure it was possible to carry out a joint estimation of all the parameters of the model, instead of doing it equation by equation. 3SLS is a complete information method that, compared to limited information models (for example the “2 stage least square”, 2SLS model), does not lose efficiency when there is a correlation between the errors of different equations of the model (Cho 1998; Kim et al. 2007).

To justify the use of simultaneous equations the existence of endogeneity must be verified by a HausmanFootnote 1 test because in its absence the estimators thus obtained will be consistent but not efficient (Gujarati 1997). In this case, the Hausman test turned out to be significant (see Table 4) therefore confirming the existence of endogeneity and the appropriateness of using simultaneous equations. Furthermore, the conditions of order and rank were also met, which means that the model is sufficiently identified and that it was possible to estimate the parameters (Gujarati 1997).

The available information on the independent variables restricted us to running a cross-sectional analysis due to the lack of longitudinal data in the data sources. For this reason, and given this data limitation, we took 2007 as the base year to avoid any distortion of the results due to such decisions as abandoned or delayed investments caused by the current financial crisis, which has been particularly severe in Spain (Jiménez et al. 2014). Taking advantage of data availability, the arithmetic average between 2000 and 2005 (both inclusive) was calculated for the control variables. This allowed us to obtain more suitable and stable estimates than with a set of explanatory variables referring to just a single year (Wiggins and Ruefli 2005; Brouthers et al. 2008).

3.3 Dependent, Independent, and Control Variables

The first dependent variable is the degree of relatedness of the product diversification. This paper follows the literature that considers related and unrelated diversification as a matter of degree (Chatterjee and Wernerfelt 1991; Montgomery and Wernerfelt 1988; Zhou 2011). Therefore the first dependent variable is the degree of relatedness of the product diversification calculated as a continuous measure based on SIC codesFootnote 2 (Standard Industrial Classification). The underlying assumption is that there exists a higher level of relatedness between those business sectors which share digits of their SIC codes as they belong to the same area of activities (Palepu 1985; Varadarajan and Ramanujam 1987; Hoskisson et al. 1993; Lubatkin et al. 1993; Hitt et al. 1997; Chang and Wang 2007).

To build the measure, we analysed the SICs code of each company. Specifically, we compared the first three digits of each possible pair of SIC codes in the company product portfolio. When both codes were similar we assigned a value of “1” as an indication of a related-product diversification strategy. Otherwise a value of “0” was assigned when the codes were different and showed an unrelated-product diversification strategy. After comparing all possible pairs, we added the scores and weighted the result by the number of pairs. By following this procedure, the value of the measure can be considered as an indicator of the degree of relatedness of the MNE’s product portfolio (Montgomery 1982), ranging from a maximum score of 1 when all business in the company belong to the same sector and 0 when none of the business do (Benito-Osorio et al. 2014).

The second dependent variable measures the level of political risk faced by a MNE in its internationalization strategy. The POLCONV index (Henisz 1998) was used for this purpose. It analyzes the degree of government discretionality in relation to the presence of and the role played by other political powers established in the host country. It takes account of the number of independent powers with veto authority in each country and the possible alignments between the various powers that might affect the constraints to which the government is subject. POLCONV has frequently been used in the literature as a proxy of political risk because it measures a very important determinant of foreign direct investment: how easily the government can change the rules arbitrarily and the credibility of its commitments to maintain the policies unchanged (Henisz and Zelner 2001, 2002; Delios and Henisz 2003a; Holburn and Zelner 2010). Specifically, as we are interested in the impact of experience dealing with political risk that the firm has accumulated in its international expansion, we calculated the variable as the average of the scores of the firm’s FDI location portfolio.

In addition to these dependent variables, other alternatives measures were also used as robustness tests. First, we used the Corruption Perceptions Index developed by Transparency International (http://www.transparency.org),Footnote 3 and the Economic Freedom Index developed by the Heritage Foundation (http://www.heritage.org)Footnote 4 as alternative variables of political risk. This way we take into account the criticisms that have been made of the various attempts to calculate such a complex phenomenon as political risk in a single measure (Kobrin 1982; De la Torre and Neckar 1988; Henisz 2002) and offer an accurate picture of the diverse elements which influence the global institutional quality of a country (Slangen and Tulder 2009; Jiménez and Delgado 2012). Furthermore, we also employed the variance of the scores in each MNE’s location portfolio as an alternative variable of political risk, as a large degree of diversity in the portfolio of investments of the firm is also a risk factors, regardless of whether the average risk level is low or high (Jiménez 2010).Footnote 5 Finally, following the methodology proposed by Chang and Wang (2007), we also run the models including an estimation of the degree of relatedness of the product diversification based on two rather than three digits of the SIC as an alternative dependent variable.

Along with these variables, which were introduced as a dependent variable in one equation and an independent one in the other, the simultaneous equations model was completed by a set of predetermined control variables which included various characteristics of the MNE such as its size, profitability, age, whether or not it is listed on any stock market sales growth, solvency, number of products and in the company’s portfolio and the business sector to which the company belongs. This last variable distinguishes between four dichotomous categories: manufacturing,Footnote 6 food, construction and the regulated sector—sectors traditionally regulated by the government but recently subject to a greater deregulation, such as air traffic, telecommunications, energy and water (OECD 1993). A fifth category “Other sectors”Footnote 7 was excluded as a reference omitted category to avoid problems of multicollinearity.

Given that a few of these business characteristics could have a role as explanatory variables in both equations, we decided to use alternative measures of size (resources and income) and profitability (ROA and ROC) as well as the most appropriate count (number of products or countries) in each equation to ensure that the model was sufficiently identified.Footnote 8 Thus, in the equation in which the degree of product diversification relatedness is the dependent variable and the experience dealing with political risk in the FDI portfolio is the independent variable, the following measures were introduced as control variables: number of products, ROA, resources (subject to a logarithmic transformation) stock market quotation, age, sales growth, solvency and sector.

$$ PRODUCT \, DIVERSIFICATION \, RELATEDNESS = {\gamma_{0}} + {\gamma_{1}} NUMBER \, OF \, PRODUCTS + {\gamma_{2}} ROA + {\gamma_{3}} RESOURCES + {\gamma_{4}} STOCK \, MARKET + {\gamma_{5}} AGE + {\gamma_{6}} SALES \, GROWTH + {\gamma_{7}} SOLVENCY + {\gamma_{8}} MANUFACTURING + {\gamma_{9}} FOOD + {\gamma_{10}} CONSTRUCTION + {\gamma_{11}} REGULATED + {\gamma_{12}} POLITICAL \, RISK + {\varepsilon_{ij}}$$

In the equation in which the political risk of the FDI portfolio is the dependent variable and the degree of relatedness of the product diversification is the dependent variable, the following measures were introduced as control variables: number of countries, ROA, income (subject to a logarithmic transformation) stock market quotation, age, sales growth, solvency and sector.

$$POLITICAL \, RISK = {\gamma_{0}} + {\gamma_{1}} NUMBER \, OF \, COUNTRIES + {\gamma_{2}} ROC + {\gamma_{3}} INCOME + {\gamma_{4}} STOCK \, MARKET + {\gamma_{5}} AGE + {\gamma_{6}} SALES \, GROWTH + \gamma_{7} SOLVENCY + \gamma_{8} MANUFACTURING + \gamma_{9} FOOD + \gamma_{10} CONSTRUCTION + {\gamma_{11}} REGULATED + {\gamma_{12}} PRODUCT \, DIVERSIFICATION \, RELATEDNESS + {\varepsilon_{ij}}$$

4 Results and Discussion

Table 1 shows the descriptive statistics for all the variables used in these equations.

Table 1 Descriptive statistics
Full size table

Tables 2 and 3 show the correlation matrices and variance inflation factors (VIFs) for each of the equations. They show that the correlations are low and all the VIF values are below both the limit of 10 recommended by Neter et al. (1985), Kennedy (1992) and Studenmund (1992) and the stricter limit of 5.3 proposed by Hair et al. (1999). It can thus be affirmed that there are no significant multicollinearity problems.

Table 2 Matrix of correlations and variance inflation factors (VIFs) of the product diversification equation
Full size table
Table 3 Matrix of correlations and variance inflation factors (VIFs) of the political risk equation
Full size table

Table 4 shows the results obtained from the models and tests of robustness. Model 1 shows the coefficients of both equations, taking the degree of relatedness of the product diversification and the political risk of the firm’s FDI portfolio as dependent variables. In the first equation, in which the dependent variable is the degree of relatedness of the product diversification, POLCONV shows a negative and significant coefficient.

Table 4 Results table
Full size table

Since these indices are constructed so that higher scores reflect countries with lower levels of risk, this result verifies the Hypothesis 1 and confirms the existence of a positive influence of experience dealing with political risk in host countries on the degree of relatedness product diversification. In other words, a higher level of political risk in the FDI portfolio of the MNE has a positive effect on a simultaneous related product diversification strategy. The greater experience of these firms in the political and institutional context allows them to increase their multinational flexibility by considering investing in a wider range of countries which, in turn, permits that they select the ones in which they can obtain economies of scope to a greater extent as a benefit from their related diversification strategy.

By contrast, the second equation, in which POLCONV is the dependent variable, shows that the degree of relatedness in product diversification is not significant. This result means that it is not possible to either confirm or reject Hypothesis 2 as the degree of relatedness of product diversification has no significant effect on the level of political risk the firm decides to face in its internationalization strategy. This could be due to the fact that firms following an unrelated product diversification may also invest in a location portfolio characterized by a high level of political risk and try to minimize this risk with a variety of instruments such as insurance, hiring former politicians or professional lobbying agencies, the use of local workers and suppliers or diverse actions directed to gain legitimacy such as patronage, charity events, etc. It may also be the case that firms following a related product diversification strategy may believe that, despite enjoying a higher degree of similarity among activities and a lower uncertainty in their interactions with host governments, their political experience may still not be enough to make them able to compete in markets where the level of risk is simply too high or where their global competitors have better political skills.

To check the robustness of the results, a number of alternative dependent variables were used. Firstly, in models 2 and 3 alternative measures other than POLCONV were used. The Transparency International’s Perception of Corruption Index was used in model 2 and the Heritage Foundation’s Index of Economic Freedom was used in model 3. The results, however, show no significant changes and confirm again Hypothesis 1 but not Hypothesis 2. In model 4, following the recommendation of Chang and Wang (2007), we used a product diversification relatedness variable using two instead of three common digits. Once again no significant changes were found as the results show that only experience dealing with political risk has a significant influence on the degree of product relatedness. Finally, the variance of political risk scores was used in models 5 and 6 as a large disparity in the levels of political risk faced by the firm in different locations also represents a source of risk for the company (Jiménez 2010). Thus model 5 includes the variance of the POLCONV index and the average of the degree of relatedness at the three digit level as dependent variables, whereas in model 6 the two digit average was used. Once again the results showed no significant differences and confirmed Hypothesis 1. In model 5, nevertheless, the degree of product diversification relatedness showed a slightly positive result (at the 10 % confidence level).Footnote 9

5 Conclusions

In this study we examine the interrelation between market and non-market corporate strategies. In particular, we focus on the possible simultaneous relationship between two of the fundamental components of the firm’s global diversification strategy: political risk and product diversification relatedness. Building on the institutional approach and the resources and capabilities-based view, we have analyzed the influence of experience managing high levels of political risk on the degree of product diversification relatedness and vice versa, in a sample of the main Spanish MNEs.

The results showed that the influence of the degree of product diversification relatedness on political risk was not significant. This is probably due to the available tools and strategies which allow firms, including those pursuing an unrelated product diversification strategy, to invest in countries with a high level of political risk. By contrast, we find empirical evidence that proves that experience in risky environments constitutes a relevant resource for MNEs. Our results confirm that a higher level of political risk in the FDI portfolio of the MNEs has a positive impact on the degree of product diversification relatedness. This is due to the fact that firms, having faced high levels of political risk before, can draw on this experience later when they come to make new foreign investments. This increases the maximum level of risk they can tolerate and thus the number of possible countries in which they can invest. Consequently, there is a positive effect on the degree of product diversification relatedness as the higher multinational flexibility increases the chances of finding a suitable environment where the firm can achieve economies of scope to a greater extent.

In this paper we contribute to the literature on the role of institutions on corporate strategy (Levy and Spiller 1994; DeFigueiredo 1997; Henisz 1998; Zelner 1999; Jensen 2003) by analyzing the role of experience dealing with political risk on the degree of product diversification relatedness. Despite there are certain political regimes which provide a better environment for investment, we show that experience interacting with authorities in risky locations can be useful to establish operations in subsequent investments. Secondly, this study also contributes to the growing body of literature focused on global diversification, by showing that although there is evidence of a simultaneous relationship between its two dimensions, product and geographical diversification (Kumar 2009; Benito-Osorio et al. 2014), not all the components of these dimensions exercise a bidirectional influence on each other. Finally, our paper contributes to the study of non-market strategy in general and political risk and political capabilities in particular. It does so by demonstrating that greater experience of interaction and negotiation with the authorities and institutions of host countries allows firms to enter markets where the level of political risk is non-tolerable for competitors, which facilitates the achievement of economies of scope between related activities.

As a managerial contribution, this paper emphasizes that political risk can provide advantages in the future for the MNE, as it allows learning opportunities that transform into experience on which managers can rely in the future. This means that political risk may constitute a source of opportunities that MNEs can make use of if an appropriately strategy for the political environment is designed, as previous literature on political risk has emphasized (García-Canal and Guillén 2008; Oliver and Holzinger 2008; Holburn and Zelner 2010; Jiménez and Delgado 2012).

This study is subject to some limitations. The use of SIC codes allows a significant reduction in classification errors caused by subjective researcher opinion (Chatterjee and Blocher 1992) and combines the objectivity, replicability and simplicity of the SIC codes (Hall and St. John 1994) with the richness of a continuous measure that allows the classification of firms (Rumelt 1974). However, this procedure is also subject to some criticisms as they are based solely on similarities in terms of production or technological connections (Markides and Williamson 1996, 1998). Additional criticisms are related to the equidistance between SIC categories (Robins and Wiersema 1995). A further limitation of this study is that the sample used is made up entirely of Spanish MNEs, given the particularly relevant role of a proactive management of political risk in their internationalization strategy (García-Canal and Guillén 2008; Jiménez 2010; Jiménez and Delgado 2012; Jiménez et al. 2014). Consequently, generalizations of the results to other contexts must be done with caution. Also the results only apply to non-financial companies, as we had to exclude financial firms due to data unavailability. These companies tend to be single-business or follow a very idiosyncratic product diversification strategy, quite different from non-financial firms. Finally, and also due to data unavailability we are unable to assess the relative weight of each product or location in the firm’s portfolio.Footnote 10

As a guideline for future research, it would be interesting to find out whether a higher exposure or dependency to a particular product or market exercises a clear bias on the experience and the benefits the company is able to get from it and, if so, whether it is subject to diminishing returns or presents an inverted U-shaped relationship. In addition, but beyond the scope of this paper, future studies could focus on the dynamic nature of political capabilities (Lawton and Rajwani 2011) and analyze the creation, evolution and repercussions of them over time. Also, subsequent studies could verify whether the conclusions of this study are also applicable to small and medium sized firms, to MNEs from other geographical contexts and employ other alternative measures of product diversification relatedness other than SIC codes (i.e. “strategic relatedness measures” Markides and Williamson 1996). Finally, developing a direct and objective methodology to measure how firms develop and leverage political capabilities represent, in our opinion, one of the most promising areas of study to better grasp the complex interrelations between non-market environment and corporate strategy.