CEO future focus and foreign market entry modes : and foreign market entry modes: and foreign market entry modes: and foreign market entry modes: and foreign market entry modes: and foreign market entry modes: and forei

(1)

Amsterdam Business School

CEO future focus and foreign market entry modes:

a content analysis of letters to shareholders

Master Thesis Business Administration:

track International Management

Jonathan Wiemer 6076149

University of Amsterdam Faculty of Economics and Business First supervisor: R.H. Kleinknecht

Second supervisor: N. Pisani

(2)

Abstract

Prior research on entry mode choice has associated full acquisitions with a long-term orientation. However, other authors suggest the opposite by stating that joint ventures have longer payback- and adjustment periods. Furthermore, JVs are considered important in achieving long-term objectives such as learning and new market discovery. A set of entry modes undertaken by Global Fortune 500 firms between 2012-2014 in combination with a content analysis of CEO letters to shareholders shows that future focused CEOs prefer joint ventures over full acquisitions when entering a new market. This study confirms the view that joint ventures are used as a mechanism to anticipate on future developments. Full

acquisitions, on the other hand, are characterized as a means to gain quick access to new markets.

(3)

1. Introduction

Globalization incentivizes firms to expand outside their home country, providing

opportunities to discover new markets, acquire new competencies and natural resources or increasing the efficiency of production. A main problem in international expansion is

determining the effective boundaries of the firm (Brouthers & Hennart, 2007). Various modes of entry are associated with different levels of control, commitment and risk. Entry modes range from low risk, non-equity entry modes such as exporting and licensing to high risk equity entry modes such as greenfield wholly owned subsidiaries (WOS) or full acquisitions (FA) (Pan & Tse, 2000). The importance of the entry mode decision is exemplified by its significant implications for firm performance (Brouthers, 2002; Brouthers, Brouthers & Werner, 2003). Moreover, the consequences of entry modes are suggested to persist for long periods of time, as the mode of entry is found to be hard to reverse (Pedersen, Petersen & Benito, 2002).

Various theoretical perspectives and empirical studies have emerged that identify antecedents of entry mode choice. Examples include firm- industry- and country level factors such as international experience, asset specificity and the uncertainty and risk corresponding to the institutional environments of home and host countries. Important contributions have been made by authors that point out the impact of cultural distance (Hennart & Larimo, 1998) and the efficiencies of organizing transactions in the market or through hierarchies

(Williamson, 1985).

More recently, the literature has started to address the relationship between corporate governance mechanisms and entry mode choice. Musteen, Datta and Herrmann (2009) find significant effects of equity ownership structure and CEO compensation types on the choice between shared- and full control entry modes. The implications of ownership and

(5)

horizons. For instance, Musteen et al. (2009), suggest that institutional shareholders are long-term oriented, which fosters a future focus of strategic decision makers. Decision makers, in turn, express their future focus by choosing more long-term oriented, full-control entry modes when expanding internationally.

Researchers like Musteen et al. (2009) infer the short-term or long-term character of entry modes from the level of commitment and risk associated with each. Hence, high risk, full control entry modes are by definition considered as long-term, while lower risk, shared-control entry modes are characterized as short-term oriented. However, Souder and Shaver (2010) show that the direct relationship between the risk and time horizon of an investment does not always hold. Moreover, many authors have provided arguments that suggest the opposite of the short-term (low risk) shared-control and long-term (high risk) full control characterization of entry modes. Woodcock, Beamish & Makino (1994) argue that joint ventures require longer adjustment periods than acquisitions. Moreover, Biggadike (1976) and Segelod (2000) suggest that acquisitions have shorter payback periods than joint ventures. Additionally, authors have associated joint ventures with learning and anticipating on future opportunities, which further supports their long-term character (Das & Teng, 1999; Kogut, 1991).

To summarize, Musteen et al. (2009) attach assumed time horizons to shared- and full control entry modes without providing support for their short- or long-term attributes.

However, no single study has examined the short-or long-term orientation of entry modes directly. Consequently, this study hypothesizes a relationship between the future focus of the CEO and preferences for joint ventures or full acquisitions when entering a new market.

The findings have implications for the assumptions made in the prior literature on antecedents of future focus and short-and long-term investment types. Second, by connecting the stream of inquiry on future focus with the international business literature, this study helps

(6)

researchers in building a broader comprehension of short- and long term corporate behavior. A greater future focus was generally exemplified by commitment to investments like training programs, R&D and capital (Souder & Bromiley, 2012). Yet, these consequences have

received criticism regarding their reliability and completeness (Jackson & Petraki, 2011). This study adds a new perspective by addressing whether the extent to which a firm aims for short- or long-term oriented goals is expressed in pursuance of international expansion. As far as can be ascertained, this study is the first to connect the fields of future focus and international management. Finally, this study provides an important methodological contribution by responding to the difficulties to operationalize, measure and compare the existence of future focus across companies (Brochet, Loumioti & Serafeim, 20120; Laverty, 1996; Marginson & McAulay, 2008). Building on Brochet et al. (2012) and Yadav, Prabhu and Chandy (2007), a proxy of future focus is developed at the CEO level by performing a content analysis of CEO letters to shareholders in annual reports of Global Fortune 500 companies.

The remainder of this paper is structured as follows. Chapter 2 addresses the literature on entry mode choice, the role of the CEO herein and the specifics of joint ventures versus full acquisitions. Next, it includes an introduction to the literature on future focus and its various measures. Chapter 3 discusses the short- and long-term characteristics of joint ventures and full acquisitions, and concludes with the research hypothesis. Subsequently, chapter 4 describes the sample, data collection procedure and the establishment of dependent and independent variables. Chapter 5 includes the results, and chapter 6 provides a discussion of the results and their interpretation in light of the extant literature. Finally, chapter 7 is the conclusion.

(7)

2. Literature review

2.1 Entry modes

Increasing globalization exposes multinationals to important strategic decisions such as where, when and how to enter new markets. International entry mode research deals with the issue of establishing the right boundaries of the firm when expanding abroad (Brouthers & Hennart, 2007). For each entry mode type, the MNE has a different level of commitment and control over its foreign operations. Extant research suggests that choosing an appropriate entry mode is important, as it has significant implications for firm performance (Brouthers, 2002; Brouthers, Brouthers & Werner, 2003). Once the entry mode is established, it often appears to be difficult to change or correct (Pedersen, Petersen & Benito, 2002). Hence, the entry mode choice may have performance consequences that hold for long periods of time.

Pan and Tse (2000) classify entry modes in two categories, equity and non-equity entry modes. Non-equity entry modes include direct and indirect exporting and contractual agreements such as licensing and franchising. Within equity entry modes, researchers often distinguish between shared- and full-control entry modes. Shared-control entry modes are joint ventures (JV) and partial acquisitions (PA) (Brouthers & Hennart, 2007). A JV involves an investment in a new facility, in cooperation with a domestic partner. The new market entrant and the local partner share the equity and control rights over the newly created entity in the host country. In case of PA, the new market entrant acquires a share of the assets of an already existing local firm (Lopéz-Duarte & García-Canal, 2002). Full-control entry modes, on the other hand, can be full acquisitions (FA) or greenfield wholly owned subsidiaries (WOS) (Brouthers & Hennart, 2007). Scholars typically categorize acquisitions as full when 95% or more of the shares of an already existing local firm are acquired. A greenfield WOS is considered as the natural way of expanding internationally, where the new market entrant

(8)

establishes its own affiliate in the host country, using its own resources (Lopéz-Duarte & García-Canal, 2002).

2.1.1 Theoretical perspectives

The most commonly applied theoretical perspectives that explain entry mode choice are Transaction Cost Analysis (TCA), the Resource-Based View (RBV), institutional theory, and Dunning’s eclectic framework (Brouthers & Hennart, 2007). TCA considers three variables that are suggested to influence the entry mode decision: asset specificity, uncertainty and frequency (Williamson, 1985). TCA asserts that a higher specificity of assets leads to a preference for full-control entry modes, but researchers have found mixed and often insignificant results regarding its effect (Brouthers & Hennart, 2007). Within TCA,

uncertainty involves both uncertainty inside the organization as well as uncertainty related to the market that the firm is facing. Internal uncertainty is mainly exemplified by the experience of a firm in international expansion, whereas external uncertainty includes the stability of an economy, political hostility and corruption in the host country. The higher the uncertainty, the less likely it is that a firm will engage in high control, high risk entry modes like FA or greenfield WOS (Brouthers & Hennart, 2007). The final TCA variable, frequency, represents the frequency of transactions. According to Williamson (1985), an equity entry mode in a foreign country only fits when the fixed costs of the investment can be covered by a large volume of transactions.

Next, the resource-based view has a larger emphasis on the motives of firms in expanding internationally (e.g. Luo, 2002). Drivers for expansion may include the desire to exploit resources or to acquire new resources, which interacts with the chosen mode of entry. The prior literature has primarily related firm experience as a resource to entry mode choice (Brouthers & Hennart, 2007). It is argued that firms gain experience by initially expanding in

(9)

relatively close and similar markets with low risk entry modes, gradually gaining experience and engaging in higher commitment modes and more distant markets (Johanson & Vahlne, 1977). Several other studies have related the possession of other firm-specific resources to entry mode choice, showing that the possession of such resources increases the likelihood that a firm enters by equity investment (Brouthers & Hennart, 2007).

Institutional theory stresses that firms adjust their entry mode decision on the institutional environment of the country that the company considers to invest in, and the differences between that institutional environment and the environment of the firm’s home country. Similar to the uncertainty concept in TCA, it is argued that the level of development of a host country institutional environment affects entry mode considerations (Brouthers et al., 2003). Related to studies of the impact of institutional frameworks, researchers examine the effects of cultural distance between the host and home country on entry mode choice (Hennart & Larimo, 1998). Others have examined the impact of specific cultural traits. For instance, Makino and Neupert (2000) find that Japanese firms investing in the U.S. have a higher propensity to choose full control modes like FAs over JVs than U.S. firms investing in Japan. This effect is explained by drawing on the differences in power distance and uncertainty avoidance between the cultures of these two countries.

Finally, Dunning’s (1993) eclectic framework integrates insights encompassing TCA, the resource-based view and institutional theory (Brouthers & Hennart, 2007). Also referred to as OLI (ownership, location, internalization), it incorporates three concepts that result logically from the previously mentioned theories. Following Dunning’s (1993) tool, various authors have shown that firm-specific ownership advantages, country-specific location advantages and transaction cost-efficient internalization advantages all influence the decision on how to enter a foreign market (e.g. Agarwal & Ramaswami, 1992).

(10)

Additional to the four main theoretical perspectives, some studies contribute by starting to provide evidence of a relationship between corporate and business strategy and the accompanying motives of international expansion with entry mode decisions. A firm that is seeking for a market to sell its products may for instance choose another mode than a firm that is seeking resources abroad (Brouthers & Hennart, 2007). Similarly, authors explain entry mode choice by looking at the effects of different multinational strategy types, such as a global or multidomestic strategies (Domke-Damonte, 2000).

2.1.2 CEOs & entry mode choice

Studies that relate CEO characteristics with entry mode choice remain relatively sparse. Herrmann and Datta (2002; 2006) claim to be the first researchers that study the role of top management in market entry decisions. They reveal associations between CEO tenure, professional background and international experience with the likelihood of choosing full-control entry modes. By explaining how various CEO attributes affect preferences for greenfield WOS, acquisitions or joint ventures, Herrmann and Datta (2002; 2006) underline the role of the CEO in international strategic choice. More generally, Nielsen and Nielsen (2011) confirmed the impact of top management teams’ nationality diversity and international experience on entry mode choice. Nielsen and Nielsen (2011) find that greater diversity increases the likelihood of choosing for shared-control entry modes, while international experience stimulates to choose for full control.

Other work is done on the influence of corporate governance mechanisms on CEO strategic choice. The effects of corporate governance practices in international expansion strategies however remain relatively unexplored. Musteen et al. (2009) address the playing field of corporate governance mechanisms, the CEO and entry modes. More specifically, the authors address the influence of ownership structures and compensation systems on the choice

(11)

between shared- and full control entry modes. Institutional shareholder and inside director equity ownership are found to foster a preference for FAs and greenfield WOS, opposed to JVs and PAs. Similarly, CEO compensation in the form of stock options increases the likelihood that full-control entry modes are chosen. The relationships between these

mechanisms and entry mode choice are interpreted by referring to their association with the decision makers’ time orientation. Stock options, for instance, are stated to foster a future focus at the part of the CEO, which leads him to prefer an entry mode that is more advantageous in the long run.

2.1.3 Joint ventures vs. full acquisitions

Although the previously mentioned theoretical perspectives can be applied to study the full spectrum of non-equity and equity entry modes, it is the choice between JVs and FAs that is at the center of this study. This perspective is common in the existing literature, as these entry modes are two alternatives to access local firm-specific resources (Hennart & Reddy, 1997; Kogut & Singh, 1988). Moreover, Caves and Mehra (1986) show that the propensity to form a JV is negatively related to the choice for FA, which supports that these modes are substitutes in the decision on how to enter a market. Some researchers have placed PAs and JVs in the same entry mode category when studying the determinants of entry mode choice (Brouthers & Hennart, 2007), but other researchers state that PAs are legally and conceptually different from JVs (Kogut & Singh, 1988). The advantages of JVs include sharing risks, creating economies of scale and organizational learning. These arguments however do not fully explain the choice for PA when entering a new market (Chen & Hennart, 2004). PAs combine characteristics of JVs and FAs; they involve acquiring equity of an existing firm, but like JVs they imply sharing the control of the affiliate with a local partner (Lopéz-Duarte & García-Canal, 2002). With the aim of examining the effects of CEO future focus on entry mode

(12)

choice, PAs are excluded from this study. Hence, a clearer distinction can be made regarding the expected CEO future focus associated with different entry modes. In line with the focus on the distinction between JV and FA in this study, the prior literature that addresses the features of these modes specifically is discussed next.

Authors that have attempted to explain the choice between JVs and FAs primarily draw on the TCA perspective. Hennart and Reddy (1997) identify four main reasons for choosing JVs over FAs. At first, the assets that a firm might want to acquire in a foreign market could be difficult to disentangle from the assets that are irrelevant to the firm, for example because they are part of a large and non-divisionalized company. The indigestibility of the target assets may lead a firm to prefer a JV over FA because the domestic parent firm can keep access to the assets that are unnecessary to the entering firm, which lowers the cost of the investment (Hennart & Reddy, 1997). A second reason for choosing JV over FA may lie in the differences between the cultures among employees of two firms, due to differences in industries or countries of origin. A JV minimizes the costs of managing different cultures because it stimulates the employees of both firms to maximize their contribution. Therefore, JVs are attractive for firms without experience in working with a foreign labor force (Kogut & Singh, 1988). Third, Balakrishnan and Koza (1993) suggest that JVs are an attractive mechanism to deal with the asymmetry between the information of the buyer and the seller about the value of the required complementary assets. This information asymmetry is higher once the two firms operate in different businesses. A form of shared ownership and control allows the firms to assess the individual contributions of different parties to the activity in the venture. This mechanism overcomes the problem of valuing the target assets in the case of FA (Balakrishnan & Koza, 1993). Finally, consistent with with institutional theory, researchers show that institutional differences such as legal restrictions restrain firms from doing FAs in some countries. Moreover, the integration problems resulting from the differences in cultures

(13)

of companies from different countries may form an institutional barrier that appears to be more easily addressed by a JV (Kogut & Singh, 1988).

Kogut (1991) uses the perspective of investment decisions as real options to analyze JVs. According to Kogut (1991), JVs can be perceived as a means for the firm to create a window on future opportunities. Namely, the investment decision regarding new product markets is often characterized by uncertain demand. Mostly, it is impossible to invest in all potential market opportunities. Sharing the investment with a partner may form a means to reduce the risk associated with the opportunity. Moreover, because two parties may bring complementary capabilities, the overall investment costs of the venture may decrease. When the new market is proven to have positive foresights, the option to acquire the JV is likely to be exercised. Hence, JVs are an attractive instrument to anticipate on possible future developments (Kogut, 1991). The options perspective on JVs is further discussed in section 3.

2.2 Future focus and economic short-termism

Various ways have emerged by which researchers refer to the balance between aiming for short- and long-term oriented goals (Laverty, 1996). Temporal orientation, temporal

perspective, corporate investment horizons, intertemporal choice, managerial myopia, short-termism and future focus all have the conceptually similar meaning, in that they refer to “the relative importance given in strategic choices to investments with differing distributions of costs and benefits over time” (Souder & Bromiley, 2012, p. 551). For consistency and

simplicity, this paper will adopt the terms ‘short-term focus’ and ‘future focus’, as these form the most accurate description of the measure that is developed and presented in this study. Although this section addresses papers that have used a wide array of terms, short-term focus and future focus are the words with which I refer to the concept of the importance of time horizons in strategic decision making. A short-term focus is reflected by a preference for

(14)

strategies with quick payoffs. On the contrary, a future focus is demonstrated by substantial commitment to investments with deferred payoffs, such as investments in new technologies, training programs, R&D and capital (Souder & Bromiley, 2012).

Future focus is a concept that arises at different organizational levels, and authors measure and explain its existence on these different levels. Examples include the CEO level (Yadav et al., 2007), top management team level (Demirag, 1998), firm level (Bushee, 1998), across an organization’s hierarchical levels (Marginson & McAulay, 2008) or at the part of shareholders (Samuel, 2000). Although the attention in my analysis and interpretation goes to the CEO level, the literature review includes findings of studies on future focus and related concepts regardless of their level of measurement.

The extent of future focus has been at the centre of the debate over economic

short-termism (Laverty, 1996). Many studies have stressed the concern that shareholders and

managers tend to be short-term focused. Ideally, the pursuit of short-term results extrapolates in positive long-term performance. However, “the course of action that is best in the short-term is often not the same as the course of action that is best in the long run” (Jackson & Petraki, 2011, p. 11). Therefore, economic short-termism is defined as an institutionalized preference for strategies that add less value but have an earlier payoff relative to strategies that would add more value but have a later payoff (Laverty, 1996; Marginson & McAulay, 2008; Jackson & Petraki, 2011). Consequently, short-termism endangers the long-term performance and viability of a firm. Porter (1991) asserts that the lack of future focus is a severe problem in the U.S., as exemplified by low investments in R&D, corporate training and supplier relationships relative to Germany and Japan, leading to a declining competitive position.

(15)

2.3 Measures of future focus

In an attempt to inquire the antecedents of short-term and future focused corporate behavior, the existing literature has developed numerous measures. However, one reliable and

ubiquitous measure has not yet emerged. Many researchers have operationalized future focus by using figures on R&D expenditures, because they are generally long-term in nature. R&D is considered a long-term activity because it creates chances for payoffs that lie far ahead. Moreover, R&D expenditures are not depreciated like other investments (Bange & de Bondt, 1998), which means the expenditures directly reduce current earnings. In addition, the future benefits of R&D are relatively uncertain and difficult to estimate (Jackson & Petraki, 2011).

Hansen and Hill (1991) use R&D expenditures in order to determine the impact of institutional investor ownership on future focus in four technology-driven industries. Bange and de Bondt (1998), in a sample of 100 U.S. corporations, show that firms may adjust their R&D spending based on an anticipated gap between analysts’ forecast and reported income. In other words, firms can increase their current earnings by reducing long-term investments in order to meet the financial market’s expectations. However, these effects are subject to

ownership types, stock price volatility and turnover. These results are supported by Bushee (1998), who addresses the issue in a broader sample and adds a distinction between the effects of dedicated and transient institutional shareholders on R&D expenditures. Specifically, the more institutional shareholders engage in momentum trading, the more likely it is that a firm reduces R&D to inflate profits (Bushee, 1998). Cheng, Subramanyam and Zhang (2005) demonstrate that firms which provide quarterly earnings updates invest less in R&D and have significantly lower long-term growth than firms that only provide annual performance

updates. The rationale is that issuing quarterly updates creates an over-emphasis on meeting short-term goals. According to Holden and Lundstrum (2009), the introduction of long-term LEAPS options augments a future focus, as measured by the R&D intensity. Johnson and Rao

(16)

(1997) employ R&D and capital expenditures in order to determine the effects of antitakeover amendments on long-term investment. In a similar vein, Meulbroek et al. (1990) investigate the threat of hostile takeovers on future focus by examining changes in R&D expenses surrounding antitakeover amendments. In a study on both shareholder and managerial short-term focus, Samuel (2000) shows that institutional ownership negatively affects R&D expenditures, but positively affects capital expenditures. Samuel (2000) finds no effects of shareholders’ short-term focus on advertising expenses. Another study by Wahal and

McConnel (2000) incorporates expenditures for property, plant and equipment (PP&E), next to R&D expenses. They find that higher levels of institutional investor ownership are

associated with both higher PP&E and R&D expenditures.

Although R&D spending is a widely used measure of future focus and economic short-termism, the accuracy of the measure has also been criticized. According to Jackson and Petraki (2011), a focus on R&D leads to an incomplete view, because other drivers of long-term performance, such as corporate reputation and employee skills, are ignored.

Pharmaceutical firms may not reduce their R&D expenses as a short-term measure, as it is their key activity. However, they may take other measures to increase current earnings at the cost of future benefits. Hence, the validity of R&D as a measure may differ across firms and industries.

Second, Laverty (1993) argues that R&D data does not necessarily represent long-term investment. A closer look shows that R&D figures are actually composed of investments in many short-term projects. Therefore, R&D may not be an appropriate indicator of future focus. This also explains why studies that employ R&D measures show somewhat

contradictory results (Laverty, 1996). Finally, there are even authors that argue that R&D does not create significant positive results in the long run (Erickson & Jacobson, 1992), which

(17)

further questions the validity of R&D as an indicator of future focus and the use of R&D in general.

A second stream of research has used self-reported perceptions of future focus, by making use of survey evidence. Demirag (1998) finds that, in the U.K., boards of directors that feel short-term performance pressures place more emphasis on short-term financial control measures in setting R&D budgets. Liljeblom and Vaihekoski (2009) study managerial perceptions of short-term pressures in Finland and make a link between the degree of pressure and the firm’s ownership structure. As expected, more short-term focused ownership is

associated with higher perceived short-term pressures by managers. Marginson and McAulay (2008), in their case study of a telecommunications firm, operationalize future focus with two sets of respectively two and five survey items, that are tested against individual and

organizational factors like role ambiguity and group members’ time horizons. Segelod (2000) uses a questionnaire and interviews to compare executives’ future focus between Sweden and the U.S. Amongst others, Segelod (2000) asked managers about the share of capital

expenditures with payback periods longer than 5 years and accepted time for break-even for new ventures in new areas.

An advantage of survey methods is the opportunity to analyse managers’ investment preferences more in-depth. For example, they can be asked about discount rates used, and a broader range of investment types can be examined. As a result, a more complete view of future focus is given relative to the studies that focus specifically on R&D expenditures. A limitation of survey instruments is that perceptions of the extent of future focus are self-reported. A respondent is probably unaware of the future focus in other firms, industries and countries. Hence, a comparison across respondents may yield unreliable results about relative short-term pressures.

(18)

Next, a stream of literature associates executive compensation schemes with the propensity of firms to engage in earnings management and earnings restatements (Jackson & Petraki, 2011). This literature largely asserts that a higher proportion of CEO compensation tied to stock performance results in earnings management and accounting restatements (Bergstresser & Philippon, 2006; Laux & Laux, 2009; Liu, 2006). The rationale is that accounting restatements represent cases where certain strategies have affected short-term payoffs substantially. Hence, they are put forward as evidence of a short-term focus (Jackson & Petraki, 2011). Jackson and Petraki (2011) state that this method is only suggestive and incomplete, as many short-term decisions will not lead to accounting restatements.

Next to the three previously identified arrays of research that employed R&D expenditures, survey methods and earnings management proxies, some extant studies have used alternative measures of future focus. Kochhar and David (1996) examine the rate of new product development under different levels of institutional ownership, and conclude that institutional investors foster a future focus by increasing innovation. Souder and Bromiley (2012) use data on depreciation expenses in order to indicate that good performance relative to aspirations leads firms to invest more in durable assets. Thus, a prosperous financial position seems to increase a firm’s future focus. Based on an analysis of cable television operators, Souder and Shaver (2010) examine the relationship between current cash flow and investment preferences, and find that greater cash flow is an incentive to make long-term investments. Souder and Shaver (2010) suggest that the time horizon of an investment is framed independently of the investment risk. This result is in contrast to many previous studies, which have used the time horizon of an investment as a measure of risk.

Finally, some researchers have introduced proxies of future focus using content analysis methodologies, which come close to the method that is presented in this paper. Kabanoff and Keegan (2009) perform a content analysis of annual reports of 1653 listed

(19)

Australian firms between 1992-2005. Using a machine learning approach, Kabanoff and Keegan (2009) measure the attention in the annual report for the long-term past, recent past, current, near future, distant future, and ongoing. Contrary to the arguments surrounding economic short-termism, Kabanoff and Keegan (2009) find a significant increase in future focus and decrease in current / past orientation over time. Similarly, Brochet et al. (2012) carried out an analysis on 70.042 transcripts of investor conference calls of U.S. firms. For all conference calls, Brochet et al. (2012) coded the number of words that refer to the short term and the words that refer to the long term. This results in a measure that reflects the balance between the number of short- and long-term words, which is subsequently related to different proxies of short-term corporate behaviour. Various tests show that the newly developed measure is significantly related to these proxies.In addition, short-term focused firms have a short-term oriented investor base, high stock price volatility and higher equity betas (Brochet et al., 2012). Yadav et al. (2007) measured CEO future focus based on letters to shareholders in annual reports, and find that a greater future focus has a positive and long-term impact on innovation. More recently, Ridge, Kern and White (2014) performed a content analysis of 100 letters to shareholders in order to examine the effects of future focus on firm strategy. Ridge et al. (2014) find that short-term focused firms are less likely to alter their strategy.

The biggest challenge in advancing the debate on short-term and future focus lies in the development of an appropriate approach in observing and measuring its existence. Current limitations and inconsistencies in research instruments form a barrier to address the issue holistically (Laverty, 1996). Marginson and McAulay (2008) point out that the lack of direct measures suggests that new measures are needed. This study addresses this research gap by introducing a new measure that builds on content analysis studies by Brochet et al. (2012) and Yadav et al. (2007). The added value of content analysis lies in its ability to provide a unique

(20)

look into the minds and cognition of CEOs, which is difficult to obtain through other methods (Yadav et al., 2007).

(21)

3. Future focus in joint ventures and acquisitions

The internationalization literature advocates that entry into a new international market involves a learning period over which entering firms establish themselves (Johanson and Vahlne, 1977). In this start-up stage, the performance of the venture is often poor and unstable because it requires time to penetrate the market and adjust to the competitive and institutional environment. Woodcock et al. (1994) find empirical evidence that supports the existence of the initial adjustment period, and show that this period lasts longer for JVs than for FAs. This effect is explained by the novelty and vulnerability of a newly created entity.

“An entry into a new market requires not only investments with long maturity but also often investments which need a long time to reach break-even” (Segelod, 2000, p. 250). Anticipating on the growth of the Chinese economy, multinationals like Procter & Gamble, Coca-Cola and Walmart took 8-11 years after entry to become profitable (Barton & Wiseman, 2014). In order to meet pressure from the impatient stock market and competing business units, businesses could try to reach the break-even point faster by acquiring existing companies, market shares and knowledge, instead of developing them over time (Segelod, 2000). Thus, payback times of investments in new markets are, inter alia, determined by the chosen mode of entry. The required payback time, in turn, is subject to the decision maker’s future focus. Consequently, greater future focused CEOs may pick different entry modes when compared to short-term focused CEOs.

In an inquiry on the performance of new ventures by established companies, Biggadike (1976, p. 110) states that “large scale entries might require less managerial

patience”. Large-scale entrants report the least negative results in their early years. Moreover, achieving higher relative production and market scale in the early years, through higher resource commitment, will lead to shorter payback periods and quicker net positive financial performance. If this finding is translated to the level of commitment in entry modes, it can be

(22)

suggested that the choice for a FA, with higher levels of commitment and control, has a shorter break-even time relative to JVs. In short, FAs can be considered as a means to gain quick access to a new market that suits the needs of firms that are in pursuit of rapid international expansion.

According to Laverty (1996), the decision to establish a JV can be classified as an important long-term strategic decision, similar to R&D programs, workforce training and production outsourcing. Das and Teng (1999) also stress that JVs are time-consuming projects because partner firms need time to learn to work together smoothly and efficiently. Similarly, Park and Ungson (1997) point out that a future focus is required in accomplishing the goals of a JV. Authors that studied the motivations behind JV formation often emphasize the learning opportunities that JVs create for their partners (Kogut, 1988; Inkpen, 2000). Recognizing opportunities for gaining knowledge requires a future focus by the investing partners, because organizational learning processes are long-term in nature. Future focused CEOs may therefore favor JVs over FAs, because they attach more importance to learning opportunities that arise from the differences in partner skills and knowledge.

Myers (1984) argued that, next to discounted cash flow, investment opportunities should be evaluated in a framework of options. An options framework recognizes that the outcome of an investment is not only the direct cash flow, it also includes a new set of opportunities that would not have been possible without the initial investment. Options analysis is particularly useful when investments are made under a high degree of uncertainty and long time horizons. With such investments, traditional discounted cash flow analysis tends to underestimate their strategic value, because cash flows are severely discounted (Nichols, 1994). Options analysis provides a framework that appropriately treats long-term sequential investment related to cumulative and path-dependent learning (Laverty, 1996, p. 854). As stated in section 2, Kogut (1991) uses the real options perspective to analyse

(23)

investment behavior in joint ventures. Kogut (1991) asserts that joint ventures are created in order to anticipate on future technological and market developments. Companies may not have sufficient resources to be active in all potential future growth areas, a JV allows them to have a stake in them and learn about the associated business segment, technology or market. Once an unexpected growth in the market occurs, the likelihood that a firm will acquire full ownership in the venture increases. On the other hand, unfavorable developments do not increase the likelihood of dissolution of the venture. Kogut (1991) concludes that JVs may serve as a platform for anticipating on possible future developments.

Folta (1998) supports these findings by stressing that JVs may constitute a means to discover and learn about domains in which the investing company is not yet active. Especially in markets with high technological uncertainty, JVs are useful because they reduce the firm’s exposure and dependency on one specific technology. In addition, they provide the investing partners flexibility to adapt once new information about the technology is revealed. For instance, the firm may increase its ownership level in the venture when it perceives the value of the assets to be increasing. Similarly, it may reduce its stake in the venture if it perceives the technology to be overvalued (Folta, 1998).

Based on these perspectives, this study hypothesizes that a preference for JVs results from a greater CEO future focus. This is explained by the finding that a future focus is accompanied with an interest in markets of which the attractiveness is uncertain and profitability lies far ahead. A JVs is an appropriate mechanism to explore those future opportunities. Furthermore, the length of the initial adjustment period, the longer payback periods and the importance of learning demonstrate the long-term character of JVs. FAs, on the other hand, can be perceived as a means to gain quick market access with short payback periods.

(24)

Hypothesis: The greater the future focus of the CEO, the more likely he or she is to choose a

(25)

4. Research design and method

4.1 Sample

Any investment outside the home country by a MNE represents an entry mode, for which the focal company has to make a decision on how to do it. As discussed in the literature review, it is the choice between JVs and FAs that is at the center of this study, as these entry modes are two alternatives to access local firm-specific resources (Hennart & Reddy, 1997; Kogut & Singh, 1988), while PAs combine characteristics of JVs and FAs (Lopéz-Duarte & García-Canal, 2002). By excluding PAs from the analysis, a clearer distinction can be made regarding the expected CEO future focus associated with JVs and FAs.

The sample consists of a set of entry modes undertaken by 2013 Global Fortune 500 companies, the world’s 500 largest companies in sales over the fiscal year 2012. The

companies on the list are diverse in terms of industries and geographical areas. The entry modes of these companies were collected via the Zephyr database. This database contains comprehensive information on deals like mergers, acquisitions, IPOs and joint ventures. The BvD codes of the 2013 Global Fortune 500 firms were entered in the search function of the database. Under search criteria, their role in the deal is specified as acquirer, rather than target or vendor. Joint ventures and acquisitions were selected as the deal types of interest. Similar to previous studies (Gaur & Lu, 2007; Gomes-Casseres, 1990), a transaction is defined as a FA when the acquirer acquires a stake of 95% or more in the target firm. Entry modes that involve an acquisition of less than 95% ownership by the investing company are considered as PAs, which is why these cases are excluded from the sample. In case the acquired

ownership stake of the deal was not specified in the database, the deal was excluded from the analysis. In addition, all cases where the acquirer already holds a share in the target company are excluded. For example, when a firm increases its share in a subsidiary from 50% to 100%, this does not represent a case of new market entry.

(26)

Only the deals that were announced and / or completed after January 1st_{2012 are} included in the dataset. This is done because the independent variable is based on the 2012 letter to shareholders, meaning that the data on the extent of CEO future focus reflects the strategic choices made in 2012 and outlook towards the year(s) coming. Because the entry mode data collection was performed on October 18th 2014, deals that were announced after this date are not picked up.

The search in the Zephyr database, taking into consideration the criteria mentioned before, results in a large number of transactions, of which the vast majority involves irrelevant domestic deals. Furthermore, some cases represent companies that establish a joint venture with a foreign company in their home country. After deleting these and the domestic deals manually, 537 cross-border JVs and FAs remain. These deals correspond to 187 different Fortune 500 companies. No cross-border FAs and JVs are found for the other 313 companies belonging to the 2013 Global Fortune 500 between January 2012 and October 2014.

4.2 Dependent variable

The dependent variable is a categorical variable that indicates a value of 0 when a firm enters a foreign market by joint venture and 1 when a firm enters a foreign market by full

acquisition.

4.3 Independent variable of interest: CEO future focus

The independent variable that is hypothesized to explain entry mode choice is CEO future focus. One of the contributions of this study is that it uses an alternative, newly developed measure of future focus. This measure is based on a content analysis of CEO letters to shareholders in annual reports. This section will explain why content analysis is an appropriate approach in this context. Next, it provides a description of the data collection

(27)

procedure regarding the independent variable, the establishment of this variable and its validation.

4.3.1 Content analysis

Content analysis is defined as “any methodological measurement applied to text for social science purposes” (Shapiro & Markoff, 1997, p. 14; in Duriau, Reger & Pfarrer, 2007, p. 6). A content analysis allows the researcher to perform quantitative analyses on qualitative data, which is one of the main strengths of the method. Content analysis is founded on the assumption that text analysis is an instrument to get insight in the cognitive schemas of the subject of study (Duriau et al., 2007). Studies have considered the frequency with which someone uses specific words as an indicator of the importance that person attaches to the underlying themes. The method is specifically suited to get an understanding of deeper collective structures like values and attitudes (Huff, 1990). Kabanoff and Keegan (2009) state that content analysis is effective when studying future focus, as the frequency with which people use words that refer to specific temporal perspectives give insight in the relative importance given to those temporal perspectives.

In management research, content analysis is applied in studies on many strategy topics, such as strategic groups, corporate reputation and corporate social responsibility. Those studies have primarily used documents like annual reports, trade magazines and mission statements as sources of data that serve as inputs to content analysis (Duriau et al., 2007). In the existing literature, annual reports are considered relatively reliable and valid sources of data. Some argue that senior executives spend substantial time on adjusting the structure and content of the report according to their preferences (Bowman, 1984). Others have criticized the use of annual reports, because they can be composed by public relations specialists, without involvement of top management (Abrahamson & Hambrick, 1997).

(28)

Kabanoff and Keegan (2009) respond to this criticism by stating that it is very unlikely that the annual report is produced by outsiders alone, because executives would expose

themselves to serious reputational risk. Indeed, Cho and Hambrick (2006) show that the annual report is always reviewed and refined by all top executives before being published.

The analysis performed in this study is based on the content of letters to shareholders, which by default are published in the annual report. An advantage of the use of letters to shareholders is that they are consistently available over time and across a wide range of companies (Yadav et al., 2007). Moreover, previous studies have found evidence that letters to shareholders form a proper indicator of the orientation and focus of CEOs at the time they are issued (Abrahamson & Hambrick, 1997; Kabanoff, 1997). According to Yadav et al. (2007, p. 89) “the letter to shareholder is often the output of a top management team that includes more than the CEO alone, CEOs nevertheless have primary fiduciary responsibility for the

statements made in the letters”. Moreover, many researchers show that the themes that are reflected in the letter to shareholders are associated with theoretically expected actions (Yadav et al., 2007). For example, Fiol (1995) shows that there is conformity in the themes that are expressed in letters to shareholders and the themes arising in internal planning documents in the same period.

Reviewing the evidence in the literature, it is reasonable to conclude that it is valid to examine CEO future focus using a content analysis method (Kabanoff & Keegan, 2009). Second, letters to shareholders are a valid source of data when performing a content analysis (Yadav et al., 2009; Abrahamson & Hambrick, 1997; Kabanoff, 1997). Following the key assumption behind content analysis methods, this study is based on the assumption that the content of the letter to shareholders reflects the actual focus of the CEO, not the effect of corporate communications or impression management. Because Yadav et al. (2007) point out that CEOs have the primary responsibility for the statements in the letters, the future focus

(29)

that is measured is defined as exhibited at the level of the CEO. It can be assumed that the focus of the CEO has a significant role in the entry mode decision, as some studies have already found associations between CEO characteristics and entry mode preferences (e.g. Herrmann & Datta, 2002; Herrmann & Datta, 2006).

4.3.2 Measure of CEO future focus

As is mentioned in section 4.1, my sample consist of entry modes undertaken by 2013 Global Fortune 500 firms. As the 2013 list is based on the fiscal year 2012, this study makes use of annual reports written with reference to the year 2012. First, the 2012 annual reports of the Global Fortune 500 firms were collected. 397 letters to shareholders were actually found in the reports, this list of companies can be found in Appendix A. The letters were extracted from the annual report using the PDF split function of the software Adobe Acrobat XI Pro. Companies that do not publish an annual report in English were excluded from the sample (primarily Chinese and Japanese companies). Some companies only have an annual report in 10-K form (primarily American companies), which does not include a letter to shareholders. In several cases, the letter was provided independently as an attachment to the annual report, or it was solely available in the sustainability report. In addition, some companies provide two letters to shareholders, one by the CEO and one by the Chairman. In these cases, the CEO letters were taken, because these are more closely connected to the daily operations of the company, and hence their orientation is of higher relevance. Occasionally, the company does not publish a letter to shareholders, but instead, a report of an interview with the CEO. These interviews are addressed to shareholders and hold the same information, in a different design. Therefore, these interviews were included when no letter was available.

In order to examine the CEO future focus, data on the content of the letter in the 2012 annual report is collected with regard to 83 search items, divided in 4 categories. There were

(30)

(1) financial terms such as ‘shareholder’, ‘profit’, ‘capital’ or ‘dividend’. A second category consisted of (2) strategic terms, for example ‘strategy’, ‘vision’ and ‘mission’. Third, terms that refer directly to the (3) short term were included, like ‘currently’, ‘quarter’ and ‘year’. Similarly, terms referring to the (4) long term were selected, including ‘decade’, ‘looking ahead’ and ‘outlook’. These terms were selected based on a previous analysis of conference calls transcripts by Brochet et al. (2012), and a quick scan of frequently occurring terms in CEO letters to shareholders. Table 3 in Appendix B provides an overview of the four

dimensions and their underlying search terms and table 4 displays their descriptive statistics and correlations. Financial and strategic terms were deliberately separated from the terms that directly refer to the short- or long term, because previous studies are inconclusive about their relations. For example, Brochet et al. (2012) only use terms that directly refer to the time horizon, while Ridge et al. (2014) use the number of statements referring to the financial statements. By simply counting words, as opposed to interpreting texts by the researcher (e.g. Ridge et al., 2014), the subjectivity of the measure is minimalized. In addition, the context in which a word occurs is checked, which assures that the word is a valid representation of its overarching category.

All 397 shareholder letters and interviews were put in one folder. 26 files were

unsearchable and were converted to a searchable document with ABBYY Finereader. With the use of Adobe Acrobat XI Pro, an advanced search enables to search a complete folder for a specific word or combination of words. The software displays all cases where a specific term is found. Data on the counts of the 83 search terms were manually entered in Excel, per company / letter. This file was imported in SPSS Statistics, where each data point was divided by the total word count of the corresponding shareholder letter. The result is a relative

measure of the frequency of occurrence of a specific term. Next, singular and plural terms referring to the same concept were merged (summed up) into one variable, for instance

(31)

‘dividend’ and ‘dividends’. Finally, some variables were deleted, because their occurrence in the total dataset was extremely low (below 10).

A large correlation matrix was printed and all significant correlations with r < -.10 and r > .10 were marked. Terms that can be considered as synonyms, such as ‘profit’ and

‘earnings’, often showed negative correlations, confirming that a letter typically contains one of these words, not both. Two variables were merged (summed up) when their qualitative meanings were similar. In addition, their correlations with all other variables should not conflict. In that way, both a qualitative and quantitative way of reasoning led to the reduction of the total number of variables in the dataset.

Initially, this resulted in dimensions that consisted of similar words. For example, the profit dimension that consisted of profit-related terms, such as ‘profit’, ‘earnings’ and ‘return’. Several other dimensions were created, and with a new correlation matrix, these dimensions were aggregated again when they are qualitatively similar and show no quantitatively ambiguous results. In a few steps, this resulted in the four mutually exclusive dimensions, each consisting of between 7 and 23 terms. Not all 83 initial terms were included in the final dimensions, as some terms had an extremely low count and others did not correspond statistically to one of the dimensions. In addition, sometimes a closer look at how a specific search term occurred in the data led to the conclusion that it did not fit the meaning of one of the overarching dimensions. For instance, it was expected that ‘invest’ would fit under the strategy dimension, but the statistical analysis and a closer look at the appearance of the term in the data led to an exclusion of this variable.

By using the newly created dimensions, the final independent variable is built that indicates the future focus of the CEO. This is done by dividing the long term dimension by the short term dimension. The reason behind this approach is that all companies communicate about the short term in their shareholder letter, as the letter reports about the events that

(32)

occurred in the past year. Hence, a CEO will distinguish itself when it uses relatively more terms referring to the long term than its peers. In that way, it may demonstrate a greater future focus than its counterparts. The same approach holds for the strategic- and financial

dimensions, where the strategic is divided by the financial dimension. The higher the score on this variable, the more the CEO is oriented towards strategy as opposed to financial results.

4.3.3 Validity test of CEO future focus

A validity test is performed to assure the validity of the instrument that measures CEO future focus in our sample. This validity test builds on insights from Yadav et al. (2007) and Brochet et al. (2012). At first, Yadav et al. (2007) operationalize CEO future focus by counting the occurrence of the term ‘will’ in letters to shareholders. In order to be able to assess the strength and additional explanatory power of the variables built in this study, the approach of Yadav et al. (2007) is copied and applied to our sample. Hence, I divide the frequency of occurrence of the term ‘will’ in each shareholder letter by the total number of words in that letter for comparison purposes. Second, Brochet et al. (2012) have performed a content analysis of conference call transcripts to measure executives’ future focus. Next to the investor base and various earnings management proxies, Brochet et al. (2012) examined the effects of their measure on firm risk, measured by the equity beta. It is found that firms with executives with greater future focus have lower equity betas. Based on this finding, the validity of the measures that are developed in this study will be tested. The higher the significance and explanatory power of an independent variable on the equity beta, the more accurate it is considered.

An equity beta is a measure of the volatility and risk of the share of a firm compared to a reference group. Because the measure of CEO future focus that is presented in this study is based on data from shareholder letters about the year 2012, I make use of the equity beta over

(33)

2012 from Bloomberg, based on weekly data points. Furthermore, the sample of Global Fortune 500 firms involves a group of international firms from different countries. Hence, the reference group MSCI All Country World Index (MSCI ACWI) is used, as defined in

Bloomberg. In this way, the volatility of all shares in the sample are compared to the same worldwide index. 66 out of 397 companies were unlisted and were not taken into account for the validity test as they do not have an equity beta.

The association between CEO future focus and the equity beta is theoretically

explained by the fact that short-term focused companies are less likely to invest in assets that impose a high current cost, but have the potential to generate great future profits. Short-term focused firms rather pursue actions that have quick payoffs, such as cost cutting, and often choose similar strategies as their industry competitors. Strategies based on such mimetic behavior are likely to result in cash flows that correspond to average market-level cash flows (Brochet et al., 2012; Ridge et al., 2014). In addition, short-term focused firms are more disposed to shift resource allocations towards activities that meet short-term goals, which creates stakeholders’ uncertainty over the firms’ investment strategy (Polk and Sapienza, 2009).

Moreover, a lack of information can lead to complications in assessing a firm’s future focus by shareholders, as they cannot easily distinguish short-term and future focused firms. Their estimation of future focus and relative firm risk could be based on communications in public disclosures, which would lead to a positive association between a short-term focus and the equity beta (Brochet et al., 2012). It would be the direct influence of the attention in the shareholder letter that is taken into consideration in the evaluation of the firm by the

shareholder. Some studies focus on this impact of communications on shareholder perceptions (e.g. Matsumoto, Pronk & Roelofsen, 2011), but this study supposes that it is the actual focus of the CEO and its consequences that are measured.

(34)

The results of the validity test are reported in Appendix B. In short, the results indicate that the use of a variable that is based on the frequency of occurrence of ‘will’ only does not explain variance in the equity beta. Dividing ‘will’ by the short term dimension accounts for the attention of the CEO to the short term and past and appears a more accurate, valid

measure. Dividing the long term dimension, that consists of a full set of words referring to the long term, by the short term dimension however appears to be even more accurate in

explaining variance in the equity beta. Finally, the variable that is composed of the balance between a focus on strategy versus financial results does not provide significant results. I conclude that the variable ‘long term / short term’ is the most accurate and valid instrument. Hence, this variable is leading in testing the research hypothesis on entry mode choice.

4.4 Control variables

In order to control for external effects that have shown to affect the entry mode decision, a few control variables are included in the analysis. One firm-level variable, one industry variable and one host country variable are applied.

Firm size. The analysis controls for firm size as arguments in the literature suggest that

entry modes with higher levels of commitment and control are preferred by larger firms. This is due to the fact that international expansion requires a strong resource base; for entering into competition with domestic firms, enforcing contracts and establishing economies of scale (Agarwal & Ramaswami, 1992). The bigger the size of the firm, the lower is the relative pain of these costs of entering and the more likely it is to engage in foreign direct investment (Buckley & Casson, 1976). Firm size was operationalized as the natural logarithm of the company’s total sales in 2012, because this gives an accurate indication of the size of the firm and the availability of these data is complete. Namely, the data is included in the list of 2013 Global Fortune 500 firms. Other measures like the number of employees or market

(35)

capitalization have limited availability in databases such as Orbis. The natural logarithm of total sales was taken in order to correct for the wide distribution of the data.

Industry. Plenty of research shows that firms in different industries tend to have

different entry mode preferences. Examples include evidence from Brouthers and Brouthers (2003), who show that firms in the manufacturing industries show different entry mode preferences than firms in service industries. Next, Caves and Mehra (1986) relate the durability of goods across industries to entry mode choice, and Kogut and Singh (1988) examine R&D and advertising intensity across industries with entry mode choice.

Next to the arguments surrounding the relations between industries and entry mode preferences, Segelod (2000) suggests that there are differences in the required payback

periods by firms from different industries and that these payback periods apply similarly when entering a new market. For instance, service companies want quick payoffs, while companies in the forest- and aircraft sectors are used to making more long term investments and hence are willing to wait longer when entering a new market. The effects of industries on the choice for a JV or FA can thus be explained by the industry’s association with the known

characteristics of JVs and FAs, as well as the extent of future focus across industries.

The NACE rev. 2 code, as available in the Orbis database, is used in order to classify the firms in different industries. The NACE rev. 2 code consists of 6 digits, but this study classifies based on the first two digits of the code. The code stands for the industry in which the firm has its primary activities. The NACE rev 2. code originally classifies companies into 21 sectors, but only 16 sectors were represented in the sample of 397 Global Fortune 500 companies. 10 of these sectors contained only few companies (< 4% of the sample), and hence were added up in one category named ‘Other industries’, which includes a total of 40 companies (10.1% of the sample) and constitutes the reference group. The other industries are (1) Manufacturing, (2) Mining and quarrying, (3) Electricity, gas, steam and air conditioning

(36)

supply, (4) Wholesale and retail trade, repair of motor vehicles and motorcycles, (5) Information and communication and (6) Financial and insurance activities.

Host country risk. The extant literature has largely shown that host country level

factors affect the chosen mode of entry by the investing firm (Pan & Tse, 2000). The host country’s location characteristics are said to influence the risk a company faces when entering that foreign market, and the firm adapts its investment strategy according to the risk to which it will be exposed (Dunning, 1988). Higher levels of host country risk are associated with lower levels of control and commitment (Pan & Tse, 2000). A JV involves lower levels of control and commitment than FA. Within the context of this study, it could thus be expected that firm’s investing in high risk countries tend to prefer JVs above FAs. The extant literature has used many measures that relate to host country risk, including cultural, economic,

geographic and administrative dimensions (Ghemawat, 2001). This study incorporates one variable that represents the host country risk, which is the Human Development Index (HDI) as established by the United Nations. The fact that the HDI is composed of a wide range of indicators about education, living standards and health, makes it an all-encompassing measure that indicates the level of development in a country. The level of development, in its turn, relates to the risk associated with economic activity in a country (Wang & Schaan, 2008).

4.5 Statistical technique: GEE

Taking into consideration all variables presented, a set of models aims at testing the

hypothesis presented in section 3 of this study. This means that CEO future focus instruments and control variables are used to predict the preference of a CEO for either a JV or a FA. Because the dependent variable is binary (0 = JV, 1 = FA), binomial logistic regression is the appropriate statistical method. The use of binomial logistic regression is common in the entry mode literature, as researchers generally distinguish between two entry mode categories

(37)

(Musteen et al., 2009). However, one key assumption in regular binomial logistic regression is that observations are independent and uncorrelated (Field, 2009). The sample of this study includes companies that did multiple JVs or FAs in the period of study, which means that the cases are not independent.

Generalized Estimating Equations (GEE) is an approach that takes into account the dependence of observations by allowing the researcher to specify subject variables and subject variables (Ziegler, 2011). Therefore, GEE is adopted to control for the within-subject dependence of the CEO future focus measure and the size and industry variables. In order to define subjects and within-subject observations, the rank of the investing firm in the 2013 Global Fortune 500 is attached to each entry mode case. The different entry modes undertaken by that firm in the period of interest are the within-subject cases. GEE accounts for the correlations between within-subject cases in the model by incorporating the correlation structure of repeatedly measured subjects (Ziegler, 2011). A specification of the assumed correlation structure is required to model these effects, which can be an independent, exchangeable or unstructured correlation structure. Choosing an appropriate assumed correlation structure is done by trying the different options, and observing the change in the Quasi Likelihood under Independence Model criterion (QIC). The lower its value, the better is the fit of the model. In this study, an independent correlation structure yielded the best fit, which is why an independent working correlation structure is selected in the GEE models. Although it is important to specify the structure with the best fit, in the case of inappropriate correlation structure specification the model’s coefficients in GEE are generally robust and independent of its effect (Ziegler, 2011).

(38)

5. Results

This section discusses the results of the research hypothesis, starting with descriptive statistics and following with the GEE models on entry mode choice.

5.1 Descriptive statistics

The descriptive statistics and correlations of the sample of JVs and FAs and the associated firms are reported in table 1. 55% of entry modes are undertaken by firms in the

manufacturing industry. The sample consists of a total of 537 entry modes of 187 different firms, of which 194 are JVs (36%) and 343 FAs (64%). The HDI is included for 528 entry modes, as the index was not available for a few host countries, such as the Cayman Islands and Bermuda. Although the equity beta is not included in the GEE models on entry mode choice, the variable is included in the matrix to examine its correlations with other variables. As can be seen, the equity beta does not significantly correlate with the choice for JV or FA (r = -.02, p > .05). Therefore, it is assumed that the validity test and the GEE models on entry mode choice yield independent results on the predictive ability of the measures due to the independence of equity betas and entry modes. The control variable HDI shows a relatively strong significant correlation with entry mode choice (r = .48, p < .01). As expected, this means that the higher the development of the host country, the more likely it is that entry occurs by FA rather than JV. Firm size correlates significantly with entry mode choice (r = -.09, p < .05), indicating that larger firms are associated with a preference for full acquisitions. Among the industry dummies, only the Manufacturing and the Wholesale & retail dummy variables are significantly correlated with entry mode choice (respectively r = .15 and r = -.27, p < .01). Finally, two different proxies of CEO future focus (‘will / short term’ and ‘long term / short term’) are significantly correlated with entry mode choice, with ‘will / short term’ being the most strongly correlated (r = -.25, p < .01).

CEO future focus and foreign market entry modes : and foreign market entry modes: and foreign market entry modes: and foreign market entry modes: and foreign market entry modes: and foreign market entry modes: and forei

Amsterdam Business School