1
The choice between joint ventures and acquisitions:
antecedents and their impact on performance
Student name: Marten de Bruin
Student numbers: S2222108* / B130026734**
Degree courses: MSc. International Business & Management*
MSc. Advanced International Business Management and Marketing**
Supervisors: Dr. Miriam Wilhelm* & Dr. Hanna Bahemia**
3
Preface
Dear reader,
By means of this preface, I would like to say thanks to several people that have helped me in the course of my project.
First of all, special thanks goes out to my supervisors Miriam Wilhelm and Hanna Bahemia who read through my pieces multiple times and provided me with insightful feedback that helped me improve this thesis. Not only was I lucky to have two supervisors who both were in good communication with me, but they were in good communication with each other as well.
Second, I would like to thank all those that helped setting up and maintaining the Double Degree/Dual Award program between the University of Groningen and Newcastle University Business School. I really enjoyed my time in the United Kingdom, and Newcastle in particular.
Third, there were quite a few people of whom I only know the name of some. These people, mainly my supervisors’ colleagues, or former professors of mine, helped check my methodology of data analysis. Though they may not read this thesis, I would still like to thank them for their assistance.
Fourth, there are the people around me that may not have read through my thesis, but who were simply there, and that’s enough. Thanks to all of them!
Hopefully this thesis provides some useful insights and you enjoy reading it!
5
Abstract
This thesis is aimed at extending the knowledge on antecedents that influence the choice between an international joint venture and an acquisition as a mode of entry, and its performance implications. It is argued that current ways of researching market entry modes are limited to universal scales in which market entry modes can’t be classified. Transaction cost theory is said to suffer from a same simplistic representation of reality with its ‘make’ or ‘buy’ decision. This research is aimed at overcoming this simplistic view of transaction cost theory and entry modes by comparing alternatives within these categories, rather than comparing categories themselves (make vs. buy). It is argued that most current research on entry modes provides managers with an indication whether to go with high or low equity modes, but doesn’t provide guidance on how to choose between the various alternatives available in such categories. The model was tested by means of binary logistic, and multiple regressions. High host country risk, high asset specificity, large relative size of the target firm, and high host-country specific experience were found to lead to the preference of acquisitions. Legal restrictions on foreign direct investment were found to lead to the preference of joint ventures. Furthermore, resource-based/manufacturing firms were found to prefer acquisitions over joint ventures, and the opposite was found to be true for non-resource-based/service firms. However, little results were found with respect to the performance implications. Joint venture performance was found to be increased with higher asset specificity, and when firms were from manufacturing industries, but the statistical evidence was only strong for industry type. Acquisition performance seemed to be increased with higher host country risk, international experience, and as the firm size decreased, but this was only supported by weak, but significant, evidence.
7
Table of contents
Section Title
Page
1 Introduction 11
2 Theoretical background 15
2.1 Literature review 15
2.2 Research questions 17
2.3 Transaction cost and institutional theory 17
3 Hypotheses 23
3.1 Variable selection and excluded variables 23
3.1.1 Cultural distance 23
3.1.2 Institutional distance 23
3.1.3 Industry distance 24
3.2 Country-level variables 25
3.2.1 Host country risk 25
3.2.2 Real effective exchange rate 26
3.2.3 Legal restrictions on foreign direct investment 27
3.3 Industry-level variables 27
3.3.1 Industry type: manufacturing vs. services industries 27
3.3.2 Asset specificity 29
3.4 Firm-level variables 30
3.4.1 Relative size of partner / target firm 30
3.4.2 General international experience 30
3.4.3 Host country specific experience 32
3.4.4 Equity stake foreign firm 33
4 Control variable 35
4.1 Firm size (SME/MNE) 35
5 Conceptual model 37
6 Methodology 39
6.1 Data collection and variable measurement 39
6.1.1 Preference of joint venture over acquisition 39
6.1.2 Performance 39
6.1.3 Host country risk 40
6.1.4 Real effective exchange rate 40
6.1.5 Legal restrictions on foreign direct investment 40
6.1.6 Industry type: manufacturing vs. service industries 41
6.1.7 Asset specificity 41
6.1.8 Relative size partner / target firm 42
6.1.9 General international experience 42
6.1.10 Host country-specific experience 42
8
6.1.12 Firm size (SME/MNE) 42
6.2 Sampling 43
6.2.1 Sample size 43
6.2.2 Sampling method 43
6.3 Data analysis 43
6.3.1 Step 1: Antecedents of market entry mode (binary logistic regression) 43 6.3.2 Step 2: The impact of the choice of entry mode under various conditions on
the performance of the market entry mode (multiple regression)
44 6.3.3 Evidence supporting the methodology of data analysis 45
6.3.4 Threshold 45
6.4 Testing the sample 45
6.4.1 A three-step testing of the sample 45
6.4.2 Step 1: Tests of representativeness of the population 46
6.4.3 Step 2: Test of performance measure 47
6.4.4 Step 3: Factor analysis and reliability analysis 48
7 Results 49
7.1 Setup of results 49
7.2 Summary of findings expressed in p-values and B’s 49
7.3 Antecedents of market entry mode 50
7.3.1 Antecedents choice between joint venture and acquisition 50 7.3.2 Model fit antecedents choice between joint venture and acquisition 51 7.4 Impact of antecedents on joint venture performance 51 7.4.1 Model test antecedents on joint venture performance 51 7.4.2 Model fit antecedents on joint venture performance 51
7.5 Impact of antecedents on acquisition performance 51
7.5.1 Model test impact antecedents on acquisition performance 51
7.5.2 Model fit antecedents on acquisition performance 52
8 Discussion of findings 53
9 Conclusion 59
10 Limitations and future research 61
References 63
Literature 63
Websites 74
Appendix 75
Appendix 1: Literature Tables per Variable 77
Appendix 2: Screenshots Orbis and Zephyr Database 89 Appendix 3: Frequency Tables Home and Host Countries, and Factor Analysis Results
95 Appendix 4: Regression Performance Measures Growth and Return on
Assets
9
11
1. Introduction
Each time a firm enters a new host country various factors are, either consciously or unconsciously, taken into consideration when choosing a way of entering this new host
country. Some main considerations are related to the firm, and the environment it comes from and is planning on entering. Estrin, Baghdasaryan and Meyer (2009, p. 1171) acknowledge this need for adjustment to this ‘multifaceted business environment of each host country’.
Examples of firm-related considerations are: a firm’s resources, its experience in the process of internationalizing, and its experience with the specific host country (or similar countries) it is planning on entering. Firm-related considerations can enable or constrain a firm in choosing an appropriate way of entering a new host country. Whereas familiarity and experience with the host country, and a firm’s resources enable a firm to enter in the most optimal way, the lack of these can force firms to choose sub-optimal ways. This entering of a host country is referred to as market entry (Erramilli, 1991), and the way of doing so is most often referred to as market entry mode (Czinkota et al., 2009), or simply, entry mode (Zhao, Luo & Suh, 2004), entry strategy (Meyer et al., 2009), or ownership strategy (Eden & Miller, 2004).
With increased globalization in the past couple of decades, scholars have increasingly been researching these various market entry modes.
12
of a learning race in which the joint venture’s interests are considered inferior to the partners’ own interests. Somewhat similar to joint ventures are acquisitions which are also a quick access to a new market. Compared to alternatives the fastest, and often also the largest, initial
international expansion may be an acquisition (Jung, 2004). The acquiring firm has total (operational) control, which avoids possible conflicts of interests between partners. However, obtaining good information for negotiating an agreement and avoiding risks also brings about some costs (Jung, 2004).
Joint ventures and acquisitions are similar in the sense that both: encounter cultural issues (be it company or country cultures), require direct investment (Jung, 2004), and involve two or more firms. How closely these two modes of entry are related is shown in the article by Brouthers and Hennart (2007) who compare the two modes of entry. In some articles, a distinction is even made between full acquisitions and partial acquisitions, which makes the choice of an acquisition vis-à-vis a joint venture even harder (Duarte & García-Canal, 2004; Brouthers & Hennart, 2007). A nice illustration is the example in which a foreign firm has a majority equity stake of 90%: is a joint venture so different from an acquisition then?
However, there are also some important differences between joint ventures and acquisitions which make the choice between them an important one. Joint ventures on the one hand: involve the start-up of a new firm/entity, have shared ownership, pose problems of goal conflict, but have a relatively low financial risk (Newburry & Zeira, 1997). Acquisitions on the other hand: do not involve the start-up of a new firm/entity, (usually) have full ownership held by the acquiring firm, don’t pose problems of goal conflict, but have relatively high financial risk (Newburry & Zeira, 1997).
13
years later Kogut (1991) claims that joint ventures could be considered a real option for firms to be exercised by means of an acquisition if performance is high.
15
2. Theoretical background
2.1 Literature review
Having discussed the practical relevance to managers, it seems appropriate to discuss the theoretical relevance. Kogut (1988) notes that entry modes in general have been researched frequently, but he adds that research comparing two specific entry modes is still an important topic for research. One of the topics that Kogut (1988) proposes is on the antecedents of joint venture as an entry mode. Twenty years later, scholars have slowly started to research
individual entry modes and compare the antecedents of closely related entry modes. However, entry mode research often is still limited to the antecedents of market entry modes in general. Hence, despite the fact that market entry modes have been extensively researched by scholars, there is still some work left to be done (Canabal & White III, 2008; Kogut, 1988). Topics such as full ownership versus partnership, and equity levels have been commonly researched (Chen & Hu, 2002; Anderson & Gatignon, 1986; Erramilli & Rao, 1990; Morschett, Schramm-Klein & Swoboda, 2010; Jung, 2004). Of course an overall view of entry modes in general is very useful, but it has some limitations. Entry modes are often categorized, for example, as direct,
cooperative, and wholly owned entry modes (Canabal & White III, 2008; Jung, 2004).
16
scholars, according to Canabal and White III (2008). This research offers some valuable insights, such as a distinction between full acquisitions and partial acquisitions (Lopez Duarte & García-Canal, 2004), and various variables that should be taken into consideration as factors influence the choice between joint ventures and acquisitions (Hennart & Reddy, 1997). However, this research that has been performed has various limitations, such as: 1) often the variables taken into consideration are very limited and different units of analysis at either firm-level, industry-level, and/or country-level have been used (Jung, 2004; Lopez Duarte & García-Canal, 2004; Montoro-Sanchez & Ortiz-de-Urbina-Criado, 2009), and 2) specific industries (Jung, 2004) and countries (Hennart & Reddy, 1997) have been researched, rather than industry and cross-country. Furthermore, within the research on this topic, there is some contradiction on the main factors that influence the choice of either an acquisition or a joint venture: whereas Hennart (1988) claims it to be the ‘digestibility’ of the partner firm, Balakrishnan and Koza (1993) argued that information asymmetry is the main factor.
Furthermore, Canabal and White III (2008) note that very few studies have investigated the influence of entry decisions on performance of these entries. This is also claimed by Brouthers, Brouthers and Werner (2003), and Chen and Hu (2002). A reason for this could be the fact that so many scholars have researched entry modes in general, which makes it hard to distinguish and measure performance of one entry mode compared to another. In case of a manufacturing firm, for example, a wholly owned subsidiary may yield superior performance compared to direct exporting, but this requires (relatively) high investment into assets compared to (close to) no investments. However, since joint ventures and acquisitions are much closer related than, for example, wholly owned versus a sales agent, it is easier to make performance comparisons. This leads the author to believe that there is research to be done on the antecedents of the choice between joint ventures and acquisitions and their impact on performance. A more holistic approach is needed to integrate the factors proposed by various authors. This may also be the reason for their sometimes contradicting nature, as discussed above. The choice
17
2.2 Research questions
As a result of the aforementioned, the research questions have been formulated as follows:
Research question 1: “What are antecedents to the choice between joint venture and acquisition as a mode of market entry?”
Research question 2: “What are the performance implications of these antecedents when a joint venture or acquisition is chosen as the mode of market entry?”
2.3 Transaction cost and institutional theory
A literature review on the antecedents of the choice between higher control modes/
acquisitions, and lower control modes/joint ventures has been performed to see the theories and variables other scholars have applied. Various variables have been found which might have an influence on the dependent variables that are to be measured (market entry mode and performance). For each variable a table has been made with information on the articles applying them. These tables can be found in appendix 1. Within this section, the column, ‘theory applied’, is of particular interest to the author. The tables and their information will be elaborated on in the next section.
18
The main premises of transaction cost economics is the existence of so-called transaction costs. According to Child, Faulkner, and Tallman (2005, p.4) “transaction costs are those that are incurred in arranging, managing, and monitoring transactions across markets, such as the costs of negotiation, drawing up contracts, managing the necessary logistics, and monitoring
accounts receivable”. This theory predicts that firms will likely benefit from internalizing the transaction if 1) there are few partners to choose from, 2) the conditions of the market are uncertain, and 3) if one or more parties has more or better information that is relevant to the transaction. Child, Faulkner, and Tallman (2005) mention five concepts that are central to this theory: 1) opportunism (the risk that a partner firm will act solely in its own interest), 2)
bounded rationality (the boundaries of a manager’s knowledge required for decision making in a transaction), 3) small numbers (few partners to choose from), 4) uncertainty (the uncertainty of the environment makes it hard to predict the outcome of the transaction), and 5)
information impactedness (knowledge differences between partner firms within a transaction). Basically, transaction cost theory states that the choice between internalizing a transaction or not depends on the degree to which these aforementioned costs are present. For example, if a firm’s management has extensive market knowledge, then the risks will be reduced.
Transaction costs that result from the need to reduce the risk by means of contracts, for example, will equally be reduced. Similarly, cultural distance between two partners can increase the uncertainty, which would then mean that a firm is less certain of the outcome of its internationalization effort. Therefore, it would be more favourable to internalize the transaction. In doing so, the risks of investment, and the transaction costs resulting from safeguarding these risks, are reduced.
Initially, transaction cost theory was proposed to be a theory for the choice of ‘make or buy’ (Coles & Hesterley, 1998; Leiblein, Reuer & Dalsace, 2002; Lyons, 1995). However, this thinking has been extended with the introduction of “make, buy, or ally” (White, 2000; Gulati &
19
with others firms. Relating this to the research question, joint ventures are most closely related to ‘ally’, and acquisitions are most closely related to ‘make’. However, like discussed above, joint ventures and acquisitions have some overlap. Especially when categorizing entry modes, it is important to note that not all entry modes within one category, such as ‘make’, are equal. Acquisitions do not have the problem of conflicts of interest to the same extent as joint ventures, but despite the fact that acquisitions and greenfield subsidiaries are categorized as ‘make’ there are distinct differences. It is important to acknowledge the subtle difference between acquisitions and greenfield subsidiaries as a ‘make’ strategy, because both suit
different organizational strategies, and different circumstances (Hennart & Park, 1993; Slangen & Hennart, 2007; Harzing, 2002; Brouthers & Brouthers, 2000).
The assumption underlying transaction cost theory is that managers have a simple choice between ‘make’ or ‘buy’. The addition of ‘ally’ is the first step in acknowledging that such a view of a manager’s decision is too simplistic. Not only can entry modes not simply be categorized into ‘make’ or ‘buy’, nor can they even be really accurately categorized when adding ‘ally’ to the equation. When transaction cost theory predicts a ‘make’ is best, how should a firm ‘make’ the product? In collaboration with another firm or alone? If in collaboration with another firm: with equity or a new venture involved, or simply by means of an agreement without equity and a separate venture? If alone: by way of an acquisition or a greenfield? Despite its remaining shortcomings, ‘ally’ as an option in between ‘make’ and ‘buy’, is a significant extension to the earlier model that only featured ‘make’ and ‘buy’. With the introduction of ‘ally’, firms are provided with another option in case concepts (central to transaction cost theory) such as uncertainty and/or bounded rationality neither directly point to ‘make’ nor to ‘buy’. Moreover, as the understanding of factors influencing the choice of market entry increases, models incorporate increasingly more variables. Often some variables may point at ‘make’, whilst equally many point at ‘buy’. Hence, ‘ally’ is of particular interest.
20
of equity? With respect to equity, all three aforementioned entry modes seem to involve none (or barely any). So if transaction cost theory were to point a manager in the direction of low to no equity, how is the manager supposed to know which entry mode to go with?
Perhaps entry modes are better off being treated separately rather than trying to put them all on a single scale, because many entry modes have such subtle differences that cannot simply be revealed by applying such a universal scale in terms of, for example, equity or investment risk. This assumption leads to believe that a comparison of joint ventures vis-à-vis acquisitions is interesting to make.
Above, transaction cost theory was criticized for being based on some assumptions that do not necessarily hold. It has even been claimed that transaction cost economics is wrong (Ghoshal & Moran, 1995). Another important assumption underpinning transaction cost theory is the notion of efficient markets (North, 1990a). This means that markets are efficient and that firms have the freedom to choose the market entry mode that they think suits their needs in the most optimal way. However, often this is not the case. First, markets are not fully efficient, because governments impose tariffs, taxes, and other measures that favour one group of firms over another. Hence, “efficient market depends on supporting institutions that can provide, in North’s (1990b) terminology, the formal and informal rules of the game of a market economy” (Meyer, 2001, p. 358). Second, firms do not always have the freedom to make the most optimal choice. There are countries with restrictions on ownership. Hence, even though, for example, an acquisition would seem to be the most optimal way, if the host country government limits foreign ownership to 50%, then a joint venture could be the chosen entry mode as second best option. This second assumption leads to believe that including market regulations in the
21
The relevance of institutional theory is emphasized by the fact that research incorporating the variable ‘institutional distance’, which will be elaborated on later on, seems to be dominated with the application of institutional theory (refer to appendix 1, table 4.2). Furthermore,
research taking into account the host government restrictions on foreign direct investment has, more often than not, used institutional theory either as substitute, or as complement to
transaction cost economics (refer to appendix 1, table 4.6).
Hence, institutional theory is often either used to complement transaction cost economics (Brouthers, 2002; Lu, 2002; Delios & Beamish, 1999; Brouthers & Hennart, 2007), or used separately (Eden & Miller, 2004; Gaur & Lu, 2007; Meyer et al., 2009; Xu & Shenkar, 2002; Xu, Pan & Beamish, 2004; Ionascu, Meyer & Erstin, 2004).
Interesting to see is that so many scholars have focused on predicting market entry modes by means of transaction cost theory, but that so few have looked at the performance implications of these choices (Rasheed, 2005; Brouthers, 2002). This lack of measurement of performance implications was acknowledged by Brouthers (2002, p.203).
23
3. Hypotheses
3.1 Variable selection and excluded variables
The literature has been scanned to determine which variables, in general, were prevalent in the literature on market entry mode choice. Refer to appendix 1 for the results of this scan. Of the variables found, not all have been included into the model for various reasons. These variables will be discussed in this section.
3.1.1 Cultural distance
One of the most often cited articles within entry mode research is the one by Kogut and Singh (1988) who propose a formula to measure the cultural distance between two firms. The formula is based on the work by Hofstede (1980) who (initially) proposed four dimensions by which culture could be categorized.
Despite its prevalence in literature, cultural distance has been excluded as a variable, because both the theories and the results of research are very scattered (refer to appendix 1, table 4.1). Besides, its application has not been limited to either independent and/or controlling variable, but has also extended to moderator. Hence, this lack of clarity on the theory that should
accompany the application of this variable, the research results, and its application as a variable leads to believe that despite its prevalence in literature, cultural distance is best not applied in this research.
3.1.2 Institutional distance
24
that Brouthers and Hennart (2007, p.407) acknowledged that more work needs to be done on institutional distance.
Despite the fact that institutional distance has become a commonly applied variable in recent research, it was not incorporated in this research. The main reason for this is the lack of connection with the theoretical perspective: transaction cost theory. The lack of application of transaction cost theory with respect to institutional distance leads to believe that this variable is outside the scope of the research. Besides, both hypothesized relations and relations found, are contradictory despite the fact that mostly one theory has been applied (refer to appendix 1, table 4.2).
3.1.3 Industry distance
Padmanabhan and Cho (1996) hypothesize that investing firms are less likely to choose market entry modes that provide full ownership when a firm expands outside of its core business. In line with their thinking are the articles by Balakrishnan and Koza (1993) and Hennart and Reddy (1997). Other articles such as the one by Brouthers & Hennart (2007), make a similar
proposition when they hypothesize that firms will choose different entry modes in case of firms that are vertical and firms that are horizontal to the entering firm.
Industry distance is the last variable that was found to be often present within literature, but was not applied in this research. The main reason for this being that there seems to be no consistent construct of industry distance. Various terms and measurements have been used such as product experience (Bell, 1996), diversification (Larimo, 1993), product differentiation (Mutinelli & Piscitello, 1998), diversified entry (Chang & Rosenzweig, 2001), industry experience (Gomes-Casseres, 1989), same industry (Hennart & Reddy, 1997), etc. Furthermore, results are often either not significant, positively significant, or negatively significant. There seems to be no consensus in terms of common results.
25
3.2 Country-level variables
3.2.1 Host country risk
Jung (2004) hypothesizes that joint ventures are preferred over acquisitions when the host country risk is higher. This is hypothesized, because joint ventures are said to be a way of reducing transaction costs related to (safeguarding against) political and economic
complications and at the same time diversifying against the risks of expropriation (Jung, 2004). “The extent to which a country’s political, legal, and economic environment threatens the stability of a business operation is generally understood (Davidson, 1982)” (Jung, 2004, p. 39). Elections and even wars can bring about political changes in the ruling party, and this can ultimately result in great policy changes. Other factors that can affect the investments of multinational firms are factors such as inflation, high interest rates, recession, and hard currency shortage, which are all factors of macroeconomic instability (Jung, 2004). Previous studies provide findings that suggest that country risk has a negative relation with the degree of control (Pan, 1996; Brouthers, 2002; Bell, 1996; Luo, 2001; Jung, 2004). Hence, multinational firms seem to prefer entry modes that offer low control when entering high-risk countries (Gatignon & Anderson, 1988; Goodnow & Hanz, 1972; Luo, 2001; Jung, 2004). In addition to Jung (2004) and Davidson (1982), many other scholars have used the concept of country risk to explain the choice of entry mode (Gatignon & Anderson, 1988; Goodnow & Hanz, 1972; Luo, 2001; Pan, 1996; Mutinelli & Piscitello, 1998; Delios & Beamish, 1999). In the majority of the articles (applying transaction cost theory), host country risk was hypothesized (and found) to negatively relate to higher equity modes (Palenzuela & Bobillo, 1999; Delios & Beamish, 1999; Chen, 2008; Gatignon & Anderson, 1988; Luo, 2001; Jung, 2004). Therefore, the following is hypothesized:
26 3.2.2 Real effective exchange rate
Exchange rate is another commonly researched variable. However, various measures relating to exchange rates and market entry modes and foreign direct investment (FDI) have been
researched: exchange rate now compared several years ago (Chang & Rosenzweig, 2001), the (real) exchange rate (Bell & Campa, 1997; Kouvelis, Axarloglou & Sinha, 2001; Baek & Kwok, 2002; Banerjee, Oetzel & Ranganathan, 2006; Buckley et al., 2007; Liu et al., 1997, Cleeve, 2008; Lecraw, 1993), the volatility of the exchange rate (Kouvelis, Axarloglou & Sinha, 2001;
Maradiaga, Zapata & Pujula, 2012; Herring, 1983; Clark et al., 2004; Huchet-Bourdon & Korinek, 2011; Chowdhury, 1993). Frequently, either the (real effective) exchange rate is used, or the volatility of the exchange rate. However, according to a recent article by Huchet-Bourdon and Korinek (2011, p.16), up till now there has not been consensus on a measure for exchange rate volatility. Hence, the real exchange rate is used, because various authors have used the World Bank’s development indicators for real (effective) exchange rates (Banerjee, Oetzel &
Ranganathan, 2006; Buckley et al., 2007; Cleeve, 2008). Furthermore, several authors found evidence that (real effective) exchange rates are positively related to the choice of acquisitions and direct investment (Baek & Kwok, 2002; Banerjee, Oetzel & Ranganathan, 2006; Chang & Rosenzweig, 2001). Like mentioned, lack of consensus on applied theory leads us to apply transaction cost theory. In line with literature, the article by Chang and Rosenzweig (2001), which applies also transaction cost theory, hypothesizes and finds evidence for a negative relation between (real effective) exchange rates and the preference for joint ventures. This seems sensible, because a higher exchange rate (a stronger home currency vis-à-vis the foreign currency) reduces the value of investment that is at risk. The lower the value of investment, the lower the risk of opportunism, because the incentive to act opportunistically is increasingly outweighed by the payoff for cooperating, as explained in the prisoners dilemma (Hill, 1990). As a result of this, the transaction costs involved with the monitoring and safeguarding of the partner’s behaviour, as is the case in a joint venture, decrease. Therefore:
27 3.2.3 Legal restrictions on foreign direct investment
Legal restrictions on foreign direct investment (FDI) has sometimes been used as control variable (Brouthers & Hennart, 2007; Gomes-Casseres, 1989; Nakos & Brouthers, 2002; Lu, 2002; Dhanaraj, 2000), but to a greater extent as independent variable (Brouthers, 2002; Brouthers, Brouthers & Werner, 2003; Gomes-Casseres, 1990; Padmanabhan & Cho, 1996; Delios & Beamish, 1999; Mutinelli & Piscitello, 1998). Legal restrictions are of crucial importance in the choice of market entry mode (Brouthers & Hennart, 2007, p.417).
Furthermore, joint ventures have been researched as a response to government regulations (Tomlinson, 1970; Friedman and Kalmanoff, 1961). Despite a ‘make’ strategy is possibly the most optimal transaction cost strategy, if foreign direct investment ownership is constraint by legal restrictions, an ‘ally’ is the second best option. This is where transaction cost theory no longer holds, because it is based on the assumption of free markets, which is not always the case, like discussed earlier. Here, institutional theory provides guidance by stating that
government regulations shape the behaviour of firms. In this case, this means that firms obey the legal restrictions, and go with joint ventures when acquisitions are forbidden. Therefore, in accordance with the literature (Gomes-Casseres, 1989; Gatignon & Anderson, 1988; Brouthers, 2002; Delios & Beamish, 1999; Luo, 2001; Padmanabhan & Cho, 1996; Shane, 1993; Chen, 2008), legal restrictions are expected to lead to preference for joint ventures over acquisitions. As a result of the aforementioned (which is in line with literature on this topic applying
transaction cost theory), the following is hypothesized:
Hypothesis 3: Legal restrictions on foreign direct investment is expected to have a positive effect on the preference of joint ventures as a market entry mode over acquisitions.
3.3 Industry-level variables
3.3.1 Industry type: manufacturing vs. service industries
28
manufacturing firms (Brouthers, 2002; Kogut & Singh, 1988; Ekeledo & Sivakumar, 1998; Nakos & Brouthers, 2002; Brouthers & Hennart, 2007; Brouthers, Brouthers & Werner, 2003), or resource-based and non-resource-based industries (Mutinelli & Piscitello, 1998; Gomes Casseres, 1989, 1990; Hennart, 1991). Service firms are said to tend towards different market entry modes compared to their manufacturing counterparts (Ekeledo & Sivakumar, 1998). Even though the direction of the relationship seems not to be entirely clear amongst scholars in general, and a slight tendency seems towards a negative relationship, scholars applying transaction cost theory seem to predict and find a positive relation between
manufacturing/resource-based vs. service/non-resource-based industries and the preference for joint ventures (Gomes-Casseres, 1989; Gomes-Casseres, 1990; Hennart, 1991). According to these authors, one of the main reasons for joint ventures is gaining access to resources which local firms control (Hennart, 1991; Gomes-Casseres, 1989). They argue that a higher joint venture propensity is expected in resource-based/manufacturing industries, because in such industries first movers (which are often local firms) “benefit from differential rents, while government policies discourage or prohibit full ownership by foreigners” (Hennart, 1991, p.485; Gomes-Casseres, 1989). In other words, when comparing manufacturing or resource-based firms to service or non-resource-based firms, manufacturing or resource-based firms are expected to have a tendency towards joint ventures whilst service or non-resource-based are expected to tend towards acquisitions. Despite the fact that this may seem to be a control variable, its common application as an independent variable amongst other transaction cost theory applying articles, led industry type to be applied as an independent variable (refer to appendix 1, table 4.7). As a result of the discussion above, the following hypothesis is drawn:
29 3.3.2 Industry-related asset specificity
Research on asset specificity has rapidly increased over the last couple of decades. Canabal and White III (2008) recognised the importance of asset specificity when they proposed that future research could look at its influence on the choice between joint ventures and wholly owned subsidiaries (Canabal & White III, 2008, p. 278). Brouthers (2002, p.213) found no evidence indicating that asset specificity influences the choice of market entry mode. In addition to this, Brouthers and Hennart (2007, p.402) claim asset specificity and its influence on the choice of market entry mode to be less applicable to target firms that are horizontal to the entering firm, these type of firms are claimed to be more influenced by information asymmetry. Although Brouthers and Hennart (2007, p.419) question whether asset specificity has been correctly used in transaction cost-based theory, many scholars claim asset specificity to have a significant influence on the choice of market entry mode (Tsai & Cheng, 2004; Erramilli & Rao, 1993; Brouthers & Brouthers, 2003, Anderson & Coughlan, 1987; Teece, 1986; Mauri & Michaels, 1998). As a result of asset specificity’s dominant presence in market entry mode literature and its application with transaction cost theory, the variable has been included.
Transaction-cost predicts that “firms integrate when asset specificity is high, because the higher costs of vertical integration are more than offset by the benefits flowing from such an
arrangement” (Erramilli & Rao, 1993, p.21). These higher transaction costs arise from factors such as risk for opportunism (Gatignon & Anderson, 1988; Hennart, 1991; Erramilli & Rao, 1993; Delios & Beamish, 1999).
The results that are found with respect to the influence of asset specificity on the choice of market entry mode are in line with that what is predicted by the transaction costs economics theory: firms making high asset specific investments tend to prefer wholly owned entry modes over joint ventures, because the costs of integration are outweighed by the risks of
opportunism (Brouthers & Brouthers, 2003; Tsai & Cheng, 2004; Anderson & Coughlan, 1987; Teece, 1986; Lu, 2002). The following hypothesis is the result of the aforementioned:
30
3.4 Firm-level variables
3.4.1 Relative size of the partner / target firm
Relating back to the Kogut (1988), the size of the target firm relative to the entering firm is also of importance. According to Hennart and Reddy (1997), firms will favour acquisitions over joint ventures when the targeted assets are ‘digestible’, which means that as the size of the target firm relative to the entering firm increases, the likelihood of an acquisition diminishes.
They are not the first to propose this influence of the relative size of the partner on the choice of market entry mode. Padmanabhan and Cho (1996) hypothesized a similar proposition when they proposed that “the larger a foreign affiliate (in relation to the size of the investing firm), the more likely the foreign affiliate will be jointly owned” (Padmanabhan and Cho, 1996, p.50). In addition to the aforementioned articles, Mutinelli and Piscitello (1998) mention the same ‘indigestibility’ and predict it to be an impediment to the choice of acquisition and hence would lead to the preference of joint ventures over acquisitions. A larger target firm will increase the risk a firm is faced with, which it will then decrease by means of a cooperatively owned venture. As a result of the consistent predictions and findings in the literature (refer to appendix 1, table 4.9), and the prediction by transaction cost theory, the following is hypothesized:
Hypothesis 6: The relative size of the partner / target firm has a positive effect on the preference of joint ventures as a market entry mode over acquisitions.
3.4.2 General international experience
31
scattered. Whilst it is clear that most scholars hypothesized a negative relationship with the degree of equity, many have not found the evidence to support this hypothesis. However, looking at appendix 1, table 4.10, one thing that becomes clear is that general international experience has often been used as independent variable as well as control variable. It is striking to see that the insignificant relationships that were found with respect to general international experience may be attributable to the use of general international experience as control variable (Xu, Pan & Beamish, 2004; Davis, Desai & Francis, 2000; Palenzuela & Bobillo, 1999; Hennart & Larimo, 1998; Lu, 2002; Brouthers, 2002). When applied as independent variable rather than control variable, general international experience seems to give much better results (Gatignon & Anderson, 1988; Chang & Rosenzweig, 2001; Hennart, 1991; Delios & Beamish, 1999; Tsai & Cheng, 2004; Zhao, Luo & Suh, 2004). Scholars applying transaction cost theory have mostly applied general international experience as an independent variable, which is why it will be used as such a variable within this research as well. The fact that the majority of the literature on this topic seems to have agreed that general international experience leads to preference of wholly owned entry modes/acquisitions is in line with the way international experience reduces the perceived transaction costs (Gatignon & Anderson, 1988; Delios & Beamish, 1999; Agarwal & Ramaswami, 1992; Mutinelli & Piscitello, 1998; Chang & Rosenzweig, 2001; Hennart, 1991; Lu, 2002; Meyer, 2001). Even though Erramilli (1991) agrees with this proposition that international experience is somehow positively related to the choice of wholly owned entry modes over jointly owned, he argues the relationship to be U-shaped. Whereas firms with both low and high experience prefer wholly owned entry modes, firms with medium international experience prefer jointly owned as an entry mode. The mixed results have been discussed by Brouthers and Hennart (2007) who conclude that meta-analysis by Zhao, Luo & Suh (2004) has given conclusive results: international experience is positively related to the preference of wholly owned over jointly owned. As a result of the discussion above, the following is hypothesized:
32 3.4.3 Host country-specific experience
In their article, Chang and Rosenzweig (2001, p.750-751) claim that the (transaction cost) predictions no longer apply when a new entry into a specific country is part of sequential foreign direct investment in that specific country. They state that as firms gather more
experience in the specific host country, the transaction costs with respect to an acquisition are significantly reduced. With host-country specific experience firms could be much more able to estimate another firm’s value and the factors involved in an acquisition in the host country. Whereas the costs of opportunism outweighed the costs of integration, the costs of integration are then claimed to outweigh the costs of opportunism. Hence, more experienced firms, the ones that have previously established foreign operations in the host country, are likely to favour full ownership as a market entry in the host country. Padmanabhan & Cho (1996), agree on the fact that previously established subsidiaries influence the market entry choice, and they predict firms to tend towards higher equity modes. They support this proposition with the evidence they find in their research. Besides, they also make a distinction between
international business experience in general, and host country familiarity (experience). Kogut and Singh (1988) used this very distinction to distinguish between these two types of
experience. Although host country experience may seem similar to general international experience, there is an important difference. A firm may have much international experience, but if it is only limited to a certain region, and the firm is planning on entering a host country in an entirely new region, than this general international experience is only applicable to some extent. Delios and Henisz (2003) call the type of experience that is useful for the choice of entry mode in a specific host country “relevant experience”. Johanson and Vahlne (1977) make a distinction in knowledge that is quite similar: there is some knowledge that can only be acquired through really entering the host country, and other experience (more general experience) can be taught by, for example, previous experiences in countries dissimilar to the host country. Dikova and Witteloostuijn (2007) and Meyer (2001) also emphasize the
33
international experience, this variable is aimed at measuring country-specific experience. Besides the fact that transaction cost theory, again, is by far the most applied theory with respect to this factor, the predictions and results of these scholars seem equally shared: host-country specific experience has a negative effect on the preference for joint ventures (Meyer, 2001; Hennart & Reddy, 1997; Gomes-Casseres, 1989; Delios & Beamish, 1999; Luo, 2001; Padmanabhan & Cho, 1996). As a result of the discussion above, the following hypothesis is drawn:
Hypothesis 8: Host country-specific experience has a negative effect on the preference of joint ventures as a market entry mode over acquisitions.
3.4.4 Equity stake foreign firm
34
tend towards increased performance of joint ventures when the foreign firm has a higher equity stake. Hence, the following hypothesis is drawn:
35
4. Control variable
The variables mentioned in the previous section and their hypothesized relation with choice of entry mode or performance have been selected as independent variables due to the fact that that is how they have been applied in literature. However, there is one final variable that has most often been applied as control variable, which is firm size. This control variable will be discussed in the following section.
4.1 Firm size (SME/MNE)
Firm size is a very commonly used variable. Looking at appendix 1 table 4.13, it becomes clear that firm size has both been used as independent variable (Brouthers, Brouthers & Werner, 2003; Czinkota et al., 2009; Nakos & Brouthers, 2002), but more often as controlling variable (Maekelburger, Schwens & Kabst, 2012; Brouthers, 2002; Meyer, 2001; Kogut & Singh, 1988; Lu, 2002; Chang & Rosenzweig, 2001). With respect to the relationships that are hypothesized above, firm size has been found to be a controlling variable for the relationship between asset specificity and market entry mode (Maekelburger, Schwens & Kabst, 2012; Lu, 2002),
international experience and market entry mode (Nakos & Brouthers, 2002), cultural distance and market entry mode (Chen, 2008), and industry distance and market entry mode (Lu, 2002). Furthermore, firm size (measured by employee count) was found to be highly interdependent with international experience (Agarwal & Ramaswami, 1992).
36
Third, the risk of multicollinearity exists when using firm size as an independent variable
alongside ‘international experience’. Fourth, Nakos & Brouthers (2002) claim firm size not to be a proper independent variable for predicting entry mode choice in SMEs.
37
5. Conceptual model
Like mentioned in the literature review, the antecedents of the entry mode choice between joint venture and acquisition will be researched and the performance of these entries tested. Having discussed the hypotheses, the conceptual model that is deducted from this (and the literature that underpins this) graphically depicts the variables and the predicted relationships as is shown below (see figure 1).
39
6. Methodology
6.1 Data collection and variable measurement
The way of measuring the various variables will be discussed in the following section.
6.1.1 Preference of joint venture over acquisition
The preference of joint venture over acquisition will be measured using a variable indicating 1 if joint venture has been chosen as an entry mode, and 2 if an acquisition has been chosen as an entry mode. This information will be obtained from the Zephyr database, which will be
discussed later on.
6.1.2 Performance
40 6.1.3 Host country risk
The host country risk is measured by the country credit risk rating as used by Pan (1996), for example. The data was obtained from the OECD country risk classification which features data from 1999 to 2013. If year the entry mode was not included within the 1999-2013 time span, then the country risk rating in 1999 was used as an indication. This seems to be a proper measuring method, because most countries appear not to have changed significantly over the 14 years from 1999 to 2013. Other authors that used a similar index include Mutinelli and Piscitello (1998), and Delios and Beamish (1999).
An alternative measurement is the one used by Gatignon and Anderson (1988) and Goodnow and Hanz (1972). They made 3 clusters: high risk, medium risk, and low risk. However, their classification of the countries within those clusters is one that is over 40 years old. Hence, the data provided by the OECD seems more accurate. However, it is just one of the many ways of measuring host country risk, because there is no universally accepted measure (Jung, 2004).
6.1.4 Real effective exchange rate
The real effective exchange rate per country will be obtained from the World Bank
Development Indicator Database which is common in research (Buckley et al., 2007; Cleeve, 2008; Banerjee, Oetzel & Ranganathan, 2006). The real effective exchange rate is “a measure of the value of a currency against a weighted average of several foreign currencies divided by a price deflator or index of costs” (Worldbank.org).
6.1.5 Legal restrictions on foreign direct investment
Legal restrictions will be measured using Koyama and Golub’s (2006) OECD’s FDI Regulatory Restrictiveness Index. This index has been proposed by scholars such as Karabay (2010). Other authors that used legal restrictions data from third parties include Padmanabhan and Cho (1996), and Delios and Beamish (1999).
41
been used by, for example, Gomes-Casseres (1989, 1990), but is insufficient according to Brouthers and Hennart (2007), especially now the list is so outdated.
6.1.6 Industry type: manufacturing vs. service industries
The industry type of the foreign firm will also be extracted from Zephyr and Orbis. Zephyr is a widely known database with data on thousands of mergers, acquisitions, joint ventures etc. and was used by scholars such as Sanchez and Criado (2009). The owner of the database is Bureau van Dijk whom are well known for their other database Orbis, a database which is also widely used (Brouthers, 2002; de Jong, Phan & van Ees, 2011; Brouthers, Brouthers & Werner, 2003; Dikova and Witteloostuijn, 2007). To illustrate the databases, some screenshots of both the Orbis and Zephyr database have been included in the appendix (refer to appendix 2).
Subsequently, the categorization used by Ekeledo and Sivakumar (1998, p.278) will be applied to determine whether firms are in manufactured goods, hard services, or soft services. This distinction was proposed by Erramilli (1990), but more often used later on by various scholars. If firms are in the soft services, hard service, or manufactured goods business, a 0, 1, or 2 will be assigned respectively.
6.1.7 Asset specificity
Asset specificity has most often been measured by the advertising intensity and/or the R&D intensity. Both measures have been used both as substitutes (Alt et al., 1999; Canabal & White III, 2008; Pan, 1996; Kogut, 1989), as well as complements (Zhao, Luo & Suh, 2004; Delios & Beamish, 1999; Hennart, 1991; Tsai & Cheng, 2004). Due to the fact that the data on R&D expenditure was not available, advertising intensity is measured as a substitute rather than as a complement for R&D intensity. The data on ratios will be obtained from the Almanac of
42 6.1.8 Relative size of the partner / target firm
The relative size of the partner / target firm is calculated as the assets of the target / partner firm divided by the foreign firm’s (entering firm) total assets, which is in line with previous research (Padmanabhan & Cho, 1996). This data is obtained from both Zephyr and Orbis.
6.1.9 General international experience
The measure for the general international experience of a firm will be the number of subsidiaries a firm has outside of its home country of which the firm has at least 10% ownership. This measure deviates slightly from the number of countries in which a firm has subsidiaries minus the home country which is a measure applied by scholars such as Caves and Mehra (1986), and Kogut and Singh (1988). However, due to time constraints, it was not possible to check in how many countries a firm had subsidiaries, because in some cases a firm had close to, or even over, 1.000 subsidiaries.
6.1.10 Host country-specific experience
This variable is measured as the amount of years the firm has had a subsidiary, in the host country, which is still operating. Years of operation in a country or region is a commonly used measurement by scholars such as Padmanabhan and Cho (1996), Dikova and Witteloostuijn (2007), Delios and Henisz (2003), and Hennart and Larimo (1998). This data will be obtained from Zephyr and Orbis.
6.1.11 Equity stake foreign firm
The equity stake of the foreign firm will be extracted from the Zephyr database which was already mentioned and explained above.
6.1.12 Firm size (SME/MNE)
43
small firm is one with less than 50 employees, a medium firm one with less than 250 employees and a large firm one with 250 employees or more (Commission Recommendation, 2003). If a firm is small, medium, or large, a 1, 2, and 3 will be assigned respectively.
6.2 Sampling
6.2.1 Sample size
The sample size that will be used for this research is 150. This size seems appropriate, because a sample size around 150 is a commonly used sample size (Agarwal, 1994; Erramilli, 1991; Domke-Damonte, 2000; Hennart, 1991).
6.2.2 Sampling method
The sampling method that will be applied is quota sampling. This is a type of non-probability sampling and refers to selecting a predetermined number of each type of case. In this case it meant that of the 150 samples, 75 were intended to be joint ventures and 75 acquisitions. This type of sampling was chosen, because the database used had much more data entries on acquisitions. This could result in a domination of the sample by acquisitions which could possibly bias the results. Despite the fact that both 75 joint venture and acquisitions were intended, the database was able to provide 73 joint ventures, hence 77 acquisitions were included rather than the 75 ones that were intended in the first place. However, since the difference is minor, this is not expected to pose any issues.
6.3 Data analysis
The data will be analyzed by means of a 2 step analysis method. Each step will briefly be discussed in the next section.
6.3.1 Step 1: Antecedents of market entry mode (binary logistic regression)
44
(Canabal & White III, 2008) is because a logistic regression is appropriate when there are multiple independent variables and one or more non-metric dependent variables which are to be tested for a relation. This is often the case in market entry mode research, because market entry is a non-metric variable. There are two types of logistic regressions: binary logistic and multinomial logistic regressions. Binary logistic regression is the one that will be applied here, because this type of regression is applicable when there is one dependent variable which is a metric or categorical variable. Multinomial regressions are regressions with multiple non-metric dependent variables, which is not the case here. Within the binary logistic regression, categorical variables can be added. These are variables which show whether there is a difference between categories. The control variable firm size will be used as a categorical variable to see whether entry modes differ between small, medium and large firms.
Finally, the binary logistic regression will be performed by means of enter wise regression: not all variables are entered at once, but entered step by step. Due to the limit of steps allowed, the variables were added by category; step 1: country-level variables, step 2: industry-level variables, step 3: firm-level variables, and step 4: the control variable.
6.3.2 Step 2: The impact of the choice of entry mode under various conditions on the
performance of the market entry mode (multiple regression)
45
but if its negative impact is higher in the acquisition case, then the choice for a joint venture seems to be better relative to an acquisitions with respect to asset specificity.
Finally, the equity stake of the foreign partner will be added as an independent variable, but only when testing the joint venture sub-sample.
6.3.3 Evidence supporting the methodology of data analysis
In order to assure that the proposed two steps are the correct way of analyzing the data, the author has not only consulted various experts in the field of data analysis, research
methodology, and statistics, but has also found evidence from various scholars applying the same research technique (Brouthers, 2002; Erramilli, 1991; Bradley & Gannon, 2000; Schrader, 2001).
6.3.4 Threshold
Like most commonly applied, an alpha (α) of .05 will be used as the threshold. However, if a variable does not meet the threshold of alpha .05, but does meet the threshold of alpha .10 or any number close to that, it will be said to be, respectively, supported at the .10 level, or marginally significant. The latter means the variable is not found to be significant at the .05 or .10 level, but the relationship still seems quite strong.
6.4 Testing the sample
6.4.1 A three-step testing of the sample
Prior to the analysis of the data, the sample is tested for its representativeness and appropriateness in three steps.
46
Furthermore, there may be little ambiguity on how to measure entry mode choice, but there is much ambiguity on measuring performance, as discussed earlier (Murphy, Trailer & Hill, 1996). Hence, the second step is checking for the appropriateness of using either growth, or both growth and return on assets (ROA) as a performance measure. Earlier in the methodology performance was explained to be measured by both growth (of sales) and return on assets. However, in this section both types are tested by means of a simple regression. If both measures of performance correlate significantly with one another, it may be redundant to check both performance measures. If both performance do not correlate at all, then this may indicate that not both measure performance, but possibly only one of them does. If both correlate, but not significantly enough, this would indicate that using both growth and return on assets as a performance measure would indeed add value to the research, because both measure performance, but in a slightly different way.
The third step is more statistical in nature. In this step the data will be tested using factor analysis to see whether there are issues of multicollinearity and whether there are any obvious factors that could solve this multicollinearity. If issues of multicollinearity are found, then factors need to be composed, and tested by means of reliability analysis.
6.4.2 Step 1: Tests of representativeness of the population
The frequency tables can be found in appendix 3. Since the goal of this section is to briefly discuss the country frequencies, only the higher frequencies of around 10% and higher will be discussed.
47
Subsequently, the host countries are tested (refer to appendix 3, table 7.2). The most popular countries amongst the host countries are led by the United Kingdom, and followed by China, Russia, Spain, and Sweden. Comparing these statistics with those from the OECD, it becomes clear that the sample may not fully optimally represent the population. The United Kingdom is by far the leading host country. Even though it’s an important host country to FDI inflows, according to the OECD’s statistics, it’s not amongst the top 5. Hence, the United Kingdom may be somewhat overrepresented. China, on the other hand, seems properly represented, because it’s the second largest host country in the sample, and it’s also the second largest host country in the OECD’s statistics. The number one being the United States. Somewhat odd is the fact that the United States do not frequently appear in the sample as a host country. Russia and Spain seem quite properly represented, because their frequency as a host country in the sample seems quite similar to the OECD’s statistics on FDI inflows. Sweden, like the United Kingdom, seems somewhat overrepresented.
To conclude, the sample seems fine in terms of home countries, but in terms of host countries some countries may be somewhat overrepresented (the United Kingdom and Sweden) and others underrepresented (the United States). However, other countries such as China, Russia, and Spain seem properly represented. These limitations should be taken into consideration. However, it should also be noted that the OECD’s statistics are from 2013, and the entries are from 1997 up till 2011. Hence, the OECD’s statistics may be not fully be a proper benchmark. Overall, the sample fit compared to the population seems quite reasonable.
6.4.3 Step 2: Test of performance measure
48
insignificant, but not negligible, nature of the relationship between the two measures of performance, both measures will be incorporated in the statistical analysis.
6.4.4 Step 3: Factor analysis and reliability analysis
The last step is the testing of the independent variables for multicollinearity. This concept refers to relations between independent variables that cause highly significant relationships to be found even though these may be the result of independent variables being related with one another. Factor analysis is a common statistical method of testing for this multicollinearity. In case multicollinearity is found, variables that highly correlate with one another can be put in a common factor which represents all the variables it is comprised of. The commonly used and recommend threshold with a sample size of 150 is an extraction of at least ,450 (Hair et al., 1998, p.112). The factor analysis and the resulting tables show that all independent variables have extraction values above the threshold of ,450 (refer to appendix 3, table 7.3).
Since all independent variables have extraction levels above the threshold, there is no need for creating factors and reliability analysis. However, it is interesting to briefly discuss the
49
7. Results
7.1 Setup of results
In section 6.3 it was discussed that the data was to be analyzed by means of a two-step data analysis; 1) analysis of the antecedents of market entry mode choice, and 2) a comparison of the impact of these antecedents on performance of both joint ventures and acquisitions. Step 1 will be discussed in section 7.3, and step 2 will be discussed in section 7.4 and 7.5. Please note that performance has been divided into joint venture performance and acquisition
performance. This way, it is easy to see what factors lead to preference for a joint venture and what factors lead to preference for an acquisition in terms of performance.
7.2 Summary of findings expressed in p-values and B’s
The table in appendix 8 has summarized all findings as a result of the data analyses. For ease of finding, the table has been included at the very end of the report. However, this table may require some explanation. The column ‘independent’ shows which independent variable(s) is/are tested. The row ‘dependent’ shows the whether the independent variables have been tested on the choice of market entry, or the performance of the market entry (either measured as growth or return on assets). The row ‘sample’ shows whether the whole sample has been used, or a subsample (e.g. joint ventures or acquisitions). The column ‘sign.’ shows the
significance of the relationship. Variables that are (marginally) statistically significant at either the .05 or the .10 level are marked bold and orange. Finally, ‘B’ shows the direction of the relationship, and the numbers have colours assigned to make them easier to interpret: green
50 Table 1: Statistical results
Finally, since the variable ‘firm size’ is applied as a control variable, this variable has not been separately tested, but only in the model. Furthermore, ‘firm size’ was used as a categorical variable in the case of entry mode choice, and therefore has also not been included in the case of entry mode choice in the summary table in appendix 8.
7.3 Antecedents choice between joint venture and acquisition
7.3.1 Model test antecedents choice between joint venture and acquisition
The variables have been tested all together in the model with firm size as the controlling variable. This test resulted in the statistics of table 1. below, which can also be found in table 8.4 step 4 in appendix 5. Please note, the constant and the control variable have not been included in this table. In this model hypotheses 1,3,4,5,6, and 8 are (marginally) significant at the .05 level, but hypothesis 1,5, and 6 have a sign in the opposite direction.
Variables in the Equation
B S.E. Wald df Sig. Exp(B)
Host country risk ,825 ,403 4,200 1 ,040 2,283
Real effective exchange rate ,044 ,050 ,780 1 ,377 1,045
FDI restrictiveness -,228 ,077 8,800 1 ,003 ,796
Industry type ,814 ,421 3,732 1 ,053 2,256
51
7.3.2 Model fit antecedents choice between joint venture and acquisition
A common way of determining the goodness of model fit in the case of a logistic regression is to look at the classification table (refer to table 8.3 step 4 in appendix 5) and look at the overall percentage of observations that were correctly predicted by the variables in the model (Lemeshow & Hosmer, 1982; Hennart, 1991). The regression has been run with the control variable (firm size) both included and excluded, with the insignificant variables both included and excluded, and in various steps, but the total model with all independent and the control variables included still seems to have the best model fit. The total model correctly predicted 79,8% of all entry modes of which 74,4% for joint ventures and 84,4% for acquisitions.
7.4 Impact of antecedents on joint venture performance
7.4.1 Model test impact antecedents on joint venture performance
The variables seem to have no significant explanatory power of joint venture performance measured as growth. However, when measured as return on assets, industry type is significant at the .05 level, and asset specificity is marginally significant at the .10 level.
7.4.1 Model fit antecedents on joint venture performance
Not surprisingly, the model fit of the antecedents on joint venture performance is not very high when performance is measured as growth. An R of ,322 and an R square of ,103 is found when performance is measured as growth, and an R of ,649 and an R square of ,421, which is more reasonable, but still not strong, is found when performance is measured as return on assets.
7.5 Impact of antecedents on acquisition performance
7.5.1 Model test impact antecedents on acquisition performance
52
7.5.2 Model fit antecedents on acquisition performance
53
8. Discussion of findings
This thesis investigated the antecedents of the choice between joint ventures and acquisitions as modes of market entry and their impact on performance of both entry modes. The
hypotheses were derived from a literature scan which has been performed and is summarized by the tables in appendix 1. Only the most common hypotheses/variables that have been incorporated in entry mode research (applying transaction cost theory) have been
incorporated. First the antecedents will be discussed, and subsequently their impact on performance.
Looking at the four steps of enter wise regression (refer to appendix 5 table 8.3), it becomes clear that at each step the model fit has increased. Hypotheses 1 (host country risk), 3 (legal restrictions on FDI), 4 (industry type), 5 (asset specificity), 6 (relative size), and 8 (country-specific experience) found statistical support for a correlation with the choice of market entry mode. Hypothesis 2 (real effective exchange rate), and 7 (general international experience) were not supported. Although most hypotheses found statistical support, not all had signs in the expected direction.
Host country risk was significant at the .05 level, but whilst a positive relation with the choice of joint ventures was expected, a negative relation was found. An explanation for this may be the fact that under conditions of high country risk, firms are said to find it hard to predict future outcomes of a joint venture, and as a result of this the long-term contracts required for a joint venture cannot be properly drafted (Agarwal, 1994). Consequently, either partner can interpret ‘unspecified clauses’ to its benefits and act opportunistically (Agarwal, 1994). As a result of this, with high country risk firms could tend towards acquisitions rather than joint ventures.
The findings with respect to restrictions on foreign direct investment were in line with previous research (Delios & Beamish, 1999; Luo, 2001; Gomes-Casseres, 1990; Gatignon & Anderson, 1988; Brouthers, 2002): restrictions on foreign direct investment were found to lead to a preference for joint ventures over acquisitions.
