The effect of equity control and national cultural distance on the performance of international joint ventures

(1)

The effect of equity control and national cultural distance on

the performance of international joint ventures

An empirical study of international joint ventures with at least one Dutch

partner (1985-2006)

Niels Kleijn University of Groningen Faculty of Economics and Business

Korenmolen 74

2406 KD Alphen a/d Rijn, the Netherlands Niels_kleijn@hotmail.com

Date: 12-1-2008

Study: International Economics and Business Course: Master thesis

(2)

Abstract

The last decennia, there has been an enormous increase in the number of alliances. Scholars study on what the performance of alliances depends. Two factors that are still under discussion are the effect of cultural distance between partners and control over the alliance. Studies are limited in the amount of available data. This thesis overcomes that problem by developing a database of international joint ventures with at least one Dutch and one non-Dutch partner during the period 1985-2006. The descriptive statistics of 752 international joint ventures give an overview of characteristics of international joint ventures set up by Dutch firms. The regression results of the Cox proportional hazard models show that performance, measured by duration, is not decreased by cultural distance in general. These results may demonstrate that the effect of specific knowledge of the partner or other country characteristics than cultural distance, like economic growth, is more important for the performance. Furthermore, the effect of equity control is investigated. The models including only international joint ventures in non-industrialized countries lead to significant results, while these are not found in the other models. It appears that in non-industrialized countries equity control is more important than in industrialized countries.

(3)

1. Introduction

The last decennia, there has been an enormous increase in the amount of alliances (Contractor and Lorange, 2002). This started during the period 1960-1980, but was firstly ignored by theory (Hagedoorn and Osborn, 2002). The subject received more attention after the large increase of alliances in the late 1970’s (Hagedoorn and Osborn, 2002; Jagersma, 1995). Following, scholars studied different forms of alliances, reasons for undertaking them and factors that determine their success. A joint venture is one of these forms, in which partners commit equity. Hence, a joint venture is not a loose contact, but an integrated part of the firm strategy.

Although joint ventures have been very popular the last decades, their survival rate is relatively low (Kogut, 1989). Gomes-Casseres (1987) argues that this is between one and two third divided over different studies, however higher (Xu and Lu, 2007) and even lower survival rates are argued (Kogut, 1989). Studies concentrate on commenting why the survival rate of joint ventures is lower compared to other business forms. A logical explanation is that not only the failure of the joint venture, but also conflicts between partners affects the survival rate. Gomes-Casseres (1987) finds that this is the case if the goals between partners are not in line. This can be from the set up of the alliance or later, after a strategy change of one of the partners. A consequence of this is that one partner leaves the joint venture and the business entity remains a subsidiary of the other partner(s).

Scholars study on what the performance of joint ventures depends. Although the amount of literature is large, there is still no consensus on many factors. This thesis takes into account the effect of national cultural distance and equity control. Much work has been performed on these topics, but stil, there is room for refinement to study the effect of national cultural distance and factors that are set up in the alliance development (Olk, 2002).

(5)

non-industrialized countries, but this thesis analyzes IJV’s in both country types. The characteristics of industrialized and non-industrialized countries differ on economical and cultural level. Therefore, this thesis tests models that enter IJV’s either in all countries, only in industrialized countries or only in non-industrialized countries.

A large variety of performance measures is possible (Olk, 2002). The performance of an IJV can be judged according to its duration. Performance measured by duration is correlated with other performance measures (Geringer and Hébert, 1991; Barkema et al. 1997; Koh and Venkatraman, 1991). Besides that, the data that is required to determine the duration of the IJV is available from the same source as from where most of the data for other variables are collected. For these two reasons, the duration of IJV’s is considered as the performance criteria.

Hagedoorn and Osborn (2002) argue that there is a lack of data on a longer period. This thesis collects IJV-data from more than two decades. For the relatively new topic of IJV’s, this is a long period. Jagersma and Bell (1992) and Boersma (1999) outline characteristics of IJV’s with at least one Dutch partner and one non-Dutch partner during the period 1985-1996. They allowed access to their data and this thesis continues on their studies with data over the period 1985-2006.

The lack of data on a longer time frame and the need for more research on characteristics that determine the performance of IJV’s, lead to the two main research questions of this thesis: -What are the characteristics of the partners and the international joint ventures with at least one Dutch partner during the period 1985-2006?

-What are the effects of national cultural distance and equity control on the performance of international joint ventures?

(6)

(7)

2. Theory and hypotheses

This section describes underlying theories in the field of the performance of alliances and (international) joint ventures and comes to hypotheses of the effect of national cultural distance and equity control on IJV performance. The point of departure (2.1) discusses reasons for the formation of joint ventures, how their performance can be measured and definitions for (national) culture and (equity) control. The theoretical foundations of the effect of national cultural distance and equity control on performance are set up in 2.2. These theories and related studies result in the formulation of hypotheses on what the effect of national cultural distance (2.3) and equity control (2.4) on the performance of IJV’s is. The control variables are discussed in 2.5.

2.1 Point of departure

Alliances are a broad subject and can be defined as ”any interfirm cooperation that falls between the extremes of discrete, short-term contracts and the complete merger of two or more organizations” (Contractor and Lorange, 2002, p.4). Every alliance should have at least three characteristics; there has to be a common goal; working together should assist in attaining this goal; there should be a need to achieve this goal quickly (Maljers, 1999).

Within the topic of alliances, a joint venture is a form, which lies relatively close to creating a new firm or to a merger between two firms in respect to other forms of alliances (Figure 1). In some cases, a joint venture is a stage before a take-over or a merger of certain activities of two firms. Besanko et al. define a joint venture as “a particular type of strategic alliance in which two or more firms create, and jointly own, a new independent organization” (2004, p.612). This thesis defines an international joint venture (IJV) as a joint venture in which at least one partner is Dutch and at least one is not.

(8)

Koh and Venkatraman (1991) give four potential benefits for joint ventures, i.e., economies of scale, access to complementary assets, cost or risk sharing and shaping the scope of activities and as basis of competition. These benefits lead to two main reasons for the formation of joint ventures; to obtain market access or for joint R&D (Koh and Venkatraman, 1991). They argue that these can be split into fast access to new technologies, fast access to new markets, benefits from economies of scale in joint research, sources of knowledge outside the firm and sharing risks that are beyond the scope of a single organization. The only way to reach these goals is to cooperate. The development of such synergy effects is a justified reason for the formation of a joint venture.

Joint venture performance depends on a large array of characteristics. Olk (2002) outlines different approaches how to evaluate this performance. In the case of alliance optimization, this is mainly based on different factors determining the stability or duration (e.g., Barkema et al., 1997). For studies on partner optimization, more measures are available, including several financial indicators, such as turnover and stock market reaction (e.g., Schut and Frederikslust, 2004). For studies on the strategic interest of the partners or the alliance, executive surveys are most compatible (e.g., Pothukuchi et al., 2003). Olk (2002) also mentions multi-interest or sequential approaches, which are combined in or have a sequence of different performance measures.

The effect of national cultural distance and equity control on performance are central in this thesis. Culture is a difficult and complex concept. It can be defined in many forms (Hofstede, 2001), e.g., “the collective programming of the mind which distinguishes the members of one category of people from another” (Hofstede, 1994a, p.1) or the textbook definition of Schermerhorn, “culture is a shared set of beliefs, values and patterns of behavior common to a group op people” (2002, p.128). The unit of analysis for this thesis is the national culture. The Hofstede index has four dimensions1in which national culture per country is assigned a value2. The Kogut and Singh (1988) index combines these four dimensions into one variable.

Additionally, the effect of equity control is investigated. In the organization of an IJV is the amount of control an important influence. Control refers to the process by which one entity

1_{There is also a fifth dimension. This was later added and available for less countries. For these reasons it is left}

out of consideration.

2_{See appendix 1 for a short description of the Hofstede dimensions. For more in depth information the large}

(9)

influences the behavior of another (Geringer and Hébert, 1989). Control in an IJV is most clearly reached by the amount of equity ownership. Control can however, also be obtained by controlling an important asset, managerial control or by undefined factors, such as better bargaining skills. E.g., one of the analyzed newspaper articles includes an interview of a manager, who spoke about an eighty percent minority share in an IJV with an active bargaining family business as partner. This thesis only considers the amount of equity control.

2.2 Theoretical foundations

For the explanation of the effect of national cultural distance and equity control on the performance of joint ventures are two main theories central in this thesis. On the one hand, the transaction cost theory (Coase, 1937; Williamson, 1996, 2005) is combined with strategic management theories in contingency theories in which the effect of (equity) control in combination with other partner/IJV characteristics on performance is argued (Koh and Venkatraman, 1991; Geringer and Hebert, 1989; Pangarkar and Klein, 2004; Lu and Hebert, 2005). On the other hand, an adapted Uppsala model is commonly used to argue the negative effect of (national) cultural distance on performance (Barkema et al., 1997; Kogut and Singh, 1988).

The transaction cost theory and the Uppsala model are two central theories in the process of market entry and internationalization (Whitelock, 2002). He argues that both the transaction cost theory and the Uppsala model put the focus on the firm. In the Uppsala model, the internationalization process increases over physical distance (Whitelock, 2002) and activities (Johanson and Wiedersheim-Paul, 2001) through time. Related to the effect of national cultural distance on IJV performance, this theory can be shifted from physical distance to cultural distance or other differences between partners (Barkema et al., 1997; Kogut and Singh, 1988). In this adapted Uppsala model, the difference between partners is a barrier. Difficulties arising with this barrier decrease the IJV performance.

(10)

are the preferred business form, when the transaction costs are lowest for this business entity. The decision of the partner(s) and the control mechanism within the IJV determine the possibility of opportunistic behavior (Geringer and Hebert, 1989). Scholars argue that in specific circumstances (equity) control reduces the chances of opportunistic behavior, resulting in an increase in performance, while in other circumstances it is the opposite. Research is aimed at a contingency approach in which a combination of variables detects such a circumstance and the effect of (equity) control on performance can be determined.

Barkema et al. (1997) and Schut and Frederikslust (2004) perform an analysis on the performance of international joint ventures with at least one Dutch partner. Their data collection methods and period differ from this thesis, but the focus of their studies resembles the focus of this thesis. Therefore, it may be expected that their results are related to the results of this thesis. Schut and Frederikslust (2004) consider the effect of shareholder value on the decision process surrounding the creation of a joint venture. In their event history analysis of 233 non-financial joint ventures, different factors in the decision process influence shareholder value. They find that on average the announcement of a joint venture increases the stock-market price. This increase is higher if the goal is development, the activities are closer to the core activities of the Dutch firm, if the foreign partner is of large size and the partners country of origin scores high on the Hofstede dimension of individualism. They conclude that equity control by one of the partners is increasing shareholder value more than equal ownership does. This is especially the case if the company is controlled by the Dutch firm.

(11)

(12)

dimension is highly significant in their results. For individualism this is only the case if their model is adapted. On other Hofstede dimensions, Barkema et al. (1997) do not come to significant effects.

2.3 National cultural distance

Here we continue on the earlier introduced theories in combination with related papers (Table 1), to hypothesize the effect of national cultural distance and equity control on performance. The related works of Schut and Frederikslust (2004) and Barkema et al. (1997) argue that cultural distance of certain Hofstede dimensions decreases the performance of an IJV. The paper of Kogut and Singh (1988) is one of the most used references on this topic. In this relatively early study, the Uppsala model is adapted for cultural distance, coming to the argument that a higher cultural distance increases the probability of a joint venture over an acquisition. Their results support that argument. If they only consider the uncertainty avoidance index, the possibility of a joint venture increases even more. Other authors, on the contrary, do not find an effect of cultural distance on performance, not for shareholder value (Merchant and Schendel, 2000), nor manager satisfaction (Pangarkar and Klein, 2004). Pothukuchi et al. (2002) argue with results of an executive survey of 127 Indian firms that organizational cultural distance explains employee satisfaction better, while for national culture (Hofstede) this is the case for operational and strategic outcomes. Duration is a strategic outcome and following the argument of Pothuchucki et al., a negative effect of national cultural distance on the performance of an IJV can be expected in this thesis. Olk (2002) argues in his literature based paper that minimizing cultural differences improves IJV performance in most studies. Although there is not a clear consensus of the effect of national cultural distance on performance, this thesis argues that overall national cultural distance, measured with the Kogut and Singh (1988) index decreases performance.

Hypothesis 1: The performance of the IJV will increase with a lower overall national cultural distance between partners.

(13)

uncertainty avoidance index determines the level in which uncertainty is reduced by means of rules, laws, policies and regulations in most situations (Hofstede, 2002). A large cultural distance on this dimension would cause contrasting expectations and behaviors between the partners concerning the way in which is operated in the environment (Barkema et al., 1997; Barkema and Vermeulen, 1997). A difference between expectations of the partners decreases the performance of the IJV. This thesis follows the argument of Barkema et al. (1997) that performance is especially limited by a national cultural distance on the Hofstede dimension uncertainty avoidance index.

Hypothesis 2: The performance of the IJV will increase with a lower national cultural distance between partners on the Hofstede dimension “uncertainty avoidance index”.

In two studies including IJV’s with at least one Dutch firm there is a significant increase of performance with a lower cultural distance on the dimension of individualism (Barkema et al., 1997; Schut and Frederikslust, 2004). The Netherlands have one of the highest scores on this dimension. Due to the high score on this dimension, it can be stated that individualism has become an important aspect in Dutch culture and business. Partners from countries with a high score on this dimension should have fewer problems to deal with this aspect. Hence, an IJV with a partner which is closer related on the Hofstede dimensions individualism will have a better fit with the Dutch firm. This results into a better performance of the IJV.

Hypothesis 3: The performance of the IJV will increase with a lower national cultural distance between partners on the Hofstede dimension “individualism”.

(14)

2.4 Equity control

Control in general does not have a clear effect on the performance of an IJV (Xu and Lu, 2007). Geringer and Hebert (1989) suggest that control should be part of firms’ strategy. Accordingly a firm should take control in the IJV when it fits into its strategy. Geringer and Hebert (1989) argue that if the fit between strategy and control is good, the performance will be good as well. This is the case if technological knowledge is involved (Lu and Hebert, 2005) or if the partners are related (Xu and Lu, 2007).

It is better that Dutch firms take control to avoid bureaucracy (Groesbeek, 1995; Daling, 2006). The argument is that the partner from the stronger economy should take control to improve performance. The United States, Japan and the Netherlands are industrial countries operating in different types of countries. On the one hand, Merchant and Schendel (2000) find a positive influence of control by firms from the United States. On the other hand, Xu and Lu (2007) argue that control by a foreign firm in a developing country leads to a weaker link between parents and find a negative effect for Japanese control in Chinese-Japanese joint ventures. In case of Dutch firms, Ellen (2004) argues that Dutch firms take a larger equity share in undeveloped countries and in a research by Dutch students it is argued that Dutch firms should take control in China, in order to perform better (Groesbeek, 1995). Schut and Frederikslust (2004) argue that control by one firm increases the shareholder value more than if no firm is in control. That effect is even larger if the controlling firm is Dutch. This thesis expects to come to the same results as Schut and Frederikslust (2004).

Hypothesis 4: IJV’s controlled by the Dutch partner will perform better than IJV’s that are not.

(15)

expected to have more related and unrelated experience. Combining these two arguments, this thesis expects better performance of IJV’s controlled by Fortune global 500 firms.

Hypothesis 5: IJV’s controlled by a large firm will perform better than IJV’s that are controlled by a small firm or have shared equity control.

Lu and Hebert (2005) argue that in the case of large cultural distance, it is not efficient that non-local firms have high ownership levels. This is because of the better knowledge of the business environment by the local party. However, Lu and Hebert (2005) do not come to a significant result. They argue that the inclusion of only Asian partners might explain this result. In such a limited sample, national cultural distance is relatively small. This thesis uses observations in which partners are originated in a larger variety of countries and in which national cultural distance is relatively larger than in their study. This thesis follows the argument of Lu and Hebert (2005) and argues that it will conclude to significant results with a database including partners from more diverse countries.

Hypothesis 6: More national cultural distance will have a negative moderating impact on the relationship between the level of Dutch ownership and IJV performance.

2.5 Control variables

(16)

thesis tests models that enter IJV’s in all countries, only in industrialized countries and only in non-industrialized countries. In the models in which all countries are included, there will be a control variable for the industrialization of the country in which the IJV is located.

2.6 Theoretical model

These expectations result in the models below, where the + or - sign in the model expresses an expected increase or decrease on performance respectively. The control variable of industrialization is the same for all IJV’s in the models including only IJV’s in industrialized or non-industrialized countries, therefore industrialization is in those models excluded. There is no expectation of the effect of industrialization, hence the +/- sign is used. The first model includes overall cultural distance measured with the Kogut and Singh index and the second model includes the Hofstede dimensions separately. The methods and variables to test these expectations will be outlined in section 3, the results will be discussed in 4.2.

1) P = -β1KS + β2CD + β3CL – β4C*C + β5II + β6SI + β7G +/- β8IND + ε

2) P = -β1PDI - β2IDV - β3MAS - β4UAI + β5CD + β7CL – β8C*C + β9II + β10SI + β11G +/- β12IND + ε

Where,

P = Performance of the international joint venture KS = Overall cultural distance (Kogut and Singh index) PDI = Cultural distance of the power distance index IDV = Cultural distance of individuality

MAS = Cultural distance of masculinity

UAI = Cultural distance of uncertainty avoidance index CD = Control by the Dutch partner

CL = Control by a large firm

C*C = Interaction variable of cultural distance and control II = Industry relatedness between Dutch firm and the IJV SI = Size of the two partners

G = Goal of the IJV

(17)

3. Data and methods

This section describes the data collection process and the methods that are used to analyze the data. With these data, the research questions that are outlined in the introduction will be answered in section four. Section 3.1 shows how the data are collected. Section 3.2 and 3.3 describe the variables and regression methods that will be used to answer the second research question in 4.2.

3.1 Data collection

All data are collected from the same Dutch financial newspaper, “Het Financieele Dagblad”, also known as the Dutch Financial Times, in the period of January 1985 through December 2006. This newspaper is also used by other authors for their data collection (e.g., Jagersma and Bell, 1992; Schut and Frederikslust, 2004). The newspaper articles are selected if there is at least one Dutch partner and one non-Dutch partner in the announced IJV start-up or termination. The IJV has to be in the selected period and at least contain enough information to do an analysis. The selection criteria for this thesis differ from the criteria used by Schut and Frederikslust (2004). They select joint ventures of which the stock of at least one partner is traded at the Amsterdam exchange, there are no relevant announcements made at the same time by the partners in “Het Financieele Dagblad”, all partners are not financial institutions and the location of the joint ventures is supplied in the announcement. Although the criteria are different, this thesis comes to similar findings in the overlapping period. The entire database of this thesis includes start-ups of joint ventures during the period 1985-2006 and terminations during the period 1990-2006, as Jagersma and Bell (1992) dit not include terminations.

This thesis leaves the databases of Jagersma and Bell (1992) and Boersma (1999) largely unchanged, although some changes are made. Firstly, the turnover variables are left out due to too many missing values. Secondly, for the timeframe of 1990-20063, the names of the joint ventures are collected in order to find out how long they existed during the period 1990-2006. The data is updated for the concerning files and variables. Thirdly, for the data of Boersma (1999), the few files, which are from other sources4 than “Het Financieele Dagblad” are left out. These data are only collected from these other sources for two years. Lastly, this thesis

3_{This is not done for the period 1985-1989 as for this first data collection only IJV start-ups are included, not}

terminations. This period will not be included in the analysis for the second research question.

(18)

adds values for national cultural distance and elaborates on industry, equity control and control variables. These changes are necessary to update the data, make the analysis of the next section possible and combine the databases properly.

The data include 752 observations in the period 1985-2006. These data include variables of the most active Dutch firms, size of the partners, locations of the partners and the IJV, industries of the partners and the IJV, ownership of the IJV and the direction, goal and activities of the IJV. Of these 752 observations, 139 are terminated at the end of the observation period. The reasons for these terminations are included in the data. The descriptive statistics of all these variables will be discussed in 4.1. To answer the second research question and test the hypotheses, four models will be used. The variables have to be transformed in order to use these models. The methods of these models will be discussed in 3.2 and the transformed variables in 3.3.

3.2 Methods

The hypotheses of the second research question will be tested with the performance measure of duration. The methods of analysis for duration as the dependent variable are survival or hazard models. These models adapt for the censored cases in which the observations have not yet been terminated. The Cox proportional hazard model (also called Cox regression) deals with the effect of covariates on the survivor function (Tabachnick and Fidell, 2007). The goal of this thesis is to determine the effect of national cultural distance and equity control on performance or otherwise formulated, on the survivor function. Hence, the Cox proportional hazard model is the required method for this thesis. Besides that, related studies with duration as the performance measure also test their hypotheses with this method (Barkema et al., 1997; Lu and Hébert, 2005; Xu and Lu, 2007).

In the entire data set, 752 IJV’s are included. The period 1985-1990 is excluded from the regressions on the Cox proportional hazard model, because this period does not include terminations. IJV’s that cannot be checked for their development through time5are excluded. Furthermore, all observations with missing values are left out. Considering these three additional criteria, 173 IJV’s are included in the analysis of the Cox proportional hazard

5_{Their development could be checked if the name of the IJV is included in the announcement of “Het}

(19)

models. Of these 173 IJV’s, 66 are located in non-industrialized countries and 107 in industrialized countries.

This thesis uses four different models. In two models, the Kogut and Singh (1988) index is included and in two models, the Hofstede6national cultural dimensions are included instead. For both variations, there is a model for all countries, only industrialized countries and only non-industrialized countries. This results in six groups (Table 2). The analyses are done by using the statistical program SPSS.

Table 2. The four regression models.

Model Culture Type of country

1 K&S All countries included 2 HOFDIMS All countries included 3a K&S Industrialized countries 3b K&S Non-Industrialized countries 4a HOFDIMS Industrialized countries 4b HOFDIMS Non-Industrialized countries

The Cox proportional hazard model uses a regression in which covariates partly determine the probability of the hazard or termination of the IJV. Hence, the slope coefficients are opposite in sign compared to ordinary least squares regression, which would analyze the effect of the survival (Kiefer, 1988). Therefore, variables of which an increase of performance is expected are expected to have a negative sign in the results. These variables will be discussed below.

3.3 Constructs and measures

The variables that are included in the models can be split into three categories, i.e., national cultural distance, equity control and control variables (Table 3). For national cultural distance this thesis uses the Hofstede dimensions7 _{(Hofstede, 2002). These dimensions are widely}

accepted and used in many other studies. The initial four Hofstede dimensions are included separately as the difference of the score between the Netherlands and the nationality of the largest non-Dutch partner. The overall national cultural distance is measured with the Kogut and Singh (1988) index. The later developed fifth dimension is excluded for all models, because this dimension is only available for a smaller number of countries.

(20)

Table 3. Variables included in the Cox proportional hazard models.

Variable name Description Source Prediction

Dependent variable

Diff (D) Duration of the IJV in the period 1985-2006

Het Financieele Dagblad Independent Variables

Culture PDI (PDI)

Cultural distance between the Dutch firm and the non-Dutch firm on the Hofstede

dimension of power distance index Geert-Hofstede.com

Negative effect on performance. Culture IDV (IDV)

dimension of individuality Geert-Hofstede.com Negative effect on performance Culture MAS (MAS)

dimension of Masculinity. Geert-Hofstede.com

Negative effect on performance Culture UAI (UAI)

dimension of uncertainty avoidance. Geert-Hofstede.com Negative effect on performance

Culture K&S (KS)

A combination of the four Hofstede dimensions in an index developed by Kogut and Singh (1988).

Geert-Hofstede.com, Kogut and Singh (1988)

Negative effect on performance Control Dutch firm (CD) The amount of equity control in the IJV by the Dutch firm. Het Financieele Dagblad Positive effect on performance

Control large firm (CL)

A dummy variable in which for 0 there is shared equity control or equity control by a small firm and for 1 a large firm has equity

control. Het Financieele Dagblad and Fortune Positive effect on performance

Control*Culture UAI (CC)

The amount of equity in the IJV owned by the Dutch firm times the cultural distance on the Hofstede dimension uncertainty avoidance index.

Geert-Hofstede.com and Het Financieele Dagblad

Negative moderating effect on performance Control Variables

Industry relatedness IJV (II)

A dummy variable in which for 0 the IJV and the Dutch firm are not in the same industry and for 1 they are.

Het Financieele Dagblad and SBI (1993)

Positive effect on performance Size (SI)

A dummy variable in which for 0 is at least one partner is large and for 1 are both partners small.

Het Financieele Dagblad and Fortune

Negative effect on performance

Goal (G)

A dummy variable in which for 0 the goal is development and for 1 the goal is not development. Het Financieele Dagblad Negative effect on performance Industrialization (IND)

A dummy variable in which for 0 the IJV is in an industrialized country and for 1 the IJV is in a non-industrialized country.

Het Financieele Dagblad and Boersma

(1999) No prediction

(21)

dimension in doing international business (Kogut and Singh, 1988; Barkema et al., 1997; Barkema and Vermeulen, 1997).

This thesis uses four control variables. Industry relatedness between the IJV and the Dutch firm is a dummy variable in which the Dutch firm and the IJV are in the same two code industry classification of the “standaard bedrijfsindeling 1993”8 or not [1 = IJV and Dutch firm in the same industry, 0 = IJV and Dutch firm not in the same industry]. Size of the partners is a dummy variable, in which Fortune global 500 firms are large [1 = both partners are small, 0 = at least one partner is large]. Goal of the IJV is a dummy variable, where the announced goal of the IJV is development or not [1 = the goal is not development, 0 = the goal is development]. For the models that include IJV’s in all countries, there is a control variable for the industrialization of the IJV location [1 = IJV is in a non-industrialized country, 0 = IJV is in an industrialized country].

(22)

4. Results

This section discusses the results of the statistics on the two research questions. The descriptive statistics of the data collection are discussed in 4.1; these results answer the first research question. The results on the Cox proportional hazard models are discussed in 4.2, these results answer the second research question.

4.1 Descriptive statistics

Below we present the results of the descriptive analyses on the characteristics, the ownership and the termination rate of international joint ventures (IJV’s) with at least one Dutch partner in the period 1985-2006. This is a follow up on the studies of Jagersma and Bell (1992)9and Boersma (1999)10. This thesis updates their data until 2006 and does a comparable analysis.

4.1.1. Characteristics

Jagersma and Bell (1992) argue that the number of IJV start-ups drops after the 1980’s. Boersma (1999) shows, however, that the number of IJV start-ups increases in the 1990’s. After her observation period 1989-1996, the higher number of IJV start-ups remains high in the end of the 1990’s, followed by a drop at the start of the millennium (Figure 2). Whether this decrease is a trend or not should be subject of future research.

Figure 2. Number of IJV start-ups per year

2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 60 50 40 30 20 10 0

(23)

Size

Boersma (1999) argues that in most years, Philips set up the highest number of IJV’s. In the later years, the (petro) chemical firms, Royal Dutch Shell, Akzo Nobel and DSM are more active. In the new millennium, the large financial companies also become more active. In 2002 and 2004, ING and Fortis were the most active firms respectively. Philips was until 1995, the most active firm with 56 IJV start-ups, from 1995 on, Royal Dutch Shell was more active with 32 new IJV’s. Table 4 shows in which year, what Dutch firm was the most active.

Table 4. Dutch firm that set up most IJV’s by year

Year Firm N 1985 Philips 8 1986 Philips 9 1987 Philips 7 1988 Philips 3 1989 Akzo Nobel 2 1990 Philips 6 1991 DSM 3 1992 Philips 6 1993 Unilever 4 1994 Philips 7 1995 Akzo Nobel 7 1996 DSM, Shell 3 1997 DSM 6 1998 DSM, KPN, Shell 2 1999 DSM 4 2000 Hoogovens/Corus, Shell 3 2001 Shell 8 2002 ING 5 2003 Philips 4 2004 Fortis 2 2005 Shell 3 2006 Shell 3

(24)

experience continue to set up more IJV’s. However, the new millennium does not show this continuing growth.

Figure 3. Number of IJV’s set up by small and large Dutch firms per year

2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 40 30 20 10 0 not in G500 list in G500 list Dutch partner: in Global 500 list? Location

(25)

Figure 4. Continent of the foreign partner Australia & New Zealand Latin America Africa Eastern Europe Asia USA Europe 300 200 100 0

Figure 5. Continent of the IJV

Australia & New Zealand Latin America Africa Eastern Europe Asia USA Europe 300 200 100 0

In most years Europe is, followed by Asia the most popular location. Although globalization is argued to be an important factor, it cannot be concluded from these results that the locations have become more varied through the years. Besides that, most IJV’s are located in industrialized countries (Figure 6), just as the Netherlands is.11 This results into the observation that IJV locations are mostly chosen in related economies and on a small geographical distance from the Netherlands. This does not correspond with the expected increasing amount of globalization.

(26)

Figure 6. Location of the IJV per type of country Countries in transition Developing countries NIC's Industrialized countries 400 300 200 100 0 .

Boersma (1999) finds that in 1995 the amount of IJV start-ups increases, especially in China. China has the last decades experienced a large economic growth. Besides that, foreign firms have been forced to work together in joint ventures for a long period (Daling, 2006). One might expect a large increase of IJV start-ups during this period. It can be observed that the amount of IJV start-ups in China has remained high after the increase in the mid 1990’s (Figure 8). The IJV’s in China are 46% of all IJV’s in Asia. China entered the WTO in 2001 and hence, China has to decrease their restrictions on control. Firms will more and more be allowed to choose for other possibilities. With this new policy, it remains to be seen if firms will still choose IJV’s or more often for other business forms.

Figure 7. Number of IJV start-ups in China per year

2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 12 10 8 6 4 2 0 China country of the joint

(27)

Industry

The industry is measured by the two code “standaard bedrijfsindeling 1993” 12 _{and a}

classification in primary, secondary and tertiary sectors. The industry of the Dutch firm, the foreign firm and the IJV are highly correlated. For the classification in sectors, this is the case between the IJV and the Dutch firm, but less between the foreign partner and the IJV. Most IJV’s are set up in the secondary sector (54%), followed by the tertiary sector (43%) and only 3% in the primary sector. It can however be observed that the amount of IJV’s in the tertiary sector is higher since the second half of the 1990’s (Figure 8). This is due to the relative increase in the amount of IJV’s in the financial, energy and (tele)communication industry (Table 5). Until the beginning of the 1990’s, most IJV’s were set up in the electronics industry and during the 1990’s mostly in the chemical industry. This also corresponds to the most active Dutch firms in these periods (Table 4). The increase in amounts of IJV’s in the tertiary sector might be explained by the privatization of these industries and the rise of internet and mobile communication.

Figure 8. Amount of IJV start-ups in the three sectors per year

2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 40 30 20 10 0 tertiary sector industrial sector primairy sector

sector of the IJV

(28)

Table 5. Number of IJV start-ups in the four most active industries by year

chemical products

electronics and optical

apparatus financial institutes

transport, storage and communication 1985 ₄ ₈ ₀ ₂ 1986 5 10 1 0 1987 3 6 0 2 1988 ₂ ₂ ₀ ₁ 1989 ₂ ₁ ₂ ₀ 1990 7 7 5 6 1991 6 2 3 3 1992 ₅ ₈ ₄ ₅ 1993 ₆ ₂ ₃ ₂ 1994 6 6 1 6 1995 17 5 2 5 1996 11 3 6 7 1997 ₉ ₇ ₄ ₂ 1998 5 2 5 10 1999 7 3 5 10 2000 2 3 2 8 2001 ₀ ₅ ₆ ₃ 2002 1 0 8 4 2003 1 4 4 1 2004 1 3 4 2 2005 ₂ ₁ ₃ ₁ 2006 2 1 5 1

Goal, direction and activities

(29)

Figure 9. Goal of the IJV’s in different country types both obtaining knowledge continuing product/process 600 500 400 300 200 100 0 Countries in transition Developing countries NIC's Industrialized countries

type of country of the joint venture

Horizontal cooperation is most common (88%) in the period 1985-2006. In 12% of the cases there is a vertical cooperation and in 10 % of the IJV’s a diagonal cooperation. Through time, this does not differ much. Only in the beginning of the period was this relatively higher (Figure 10).

Figure 10. Direction of the cooperation in the period 1985-2006

(30)

The IJV’s are set up for different activities. Production is the most announced activity for the IJV’s in this study. For 39% of the IJV’s is production the only announced activity and in 21% of the IJV’s are producing and selling the announced activities. This shows that in at least 60% of the cases production is an activity. Selling is in 7% of the cases the only activity, but in 35% of the case, selling was one of the announced activities. Service is in 20% of the cases an announced activity. Trade and transport are rarely one of the main activities of the IJV. Development is in only 15% of the cases an activity and mostly not the only activity. Scholars however argue that we have entered a period of knowledge based economies (Contractor and Lorange, 2002). Following Hagedoorn and Osborn (2002), the results not necessarily contradict the knowledge based theory. They argue that most research partnerships are on a looser base than IJV’s are and that the percentage of research partnerships that are based on equity is decreasing.

4.1.2 Ownership of the IJV

During the period 1985-2006, only 13 % of the IJV’s have more than two partners. In 8.3% of the cases there are three partners, 2.9 % of the IJV’s have four partners and 1.7 % of the IJV’s have more than four partners (Table 6). The relative amount of IJV’s with more than two partners remains about the same through time (Figure 11). The percentage of IJV’s with more than two partners does not rise above 26% and in only one year there are IJV’s with more than two partners. Boersma (1999) finds that IJV’s with more than two partners are mainly located in non-industrialized countries. During the period 1985-2006 this is also the case; IJV’s in industrialized countries (10%) have significantly less often more than two partners than IJV’s in non-industrialized countries (21%) (t-value: 4,003; p<0,01).

Table 6. Percentage of IJV’s with two and more than two partners Amount of

partners % of theIJV's

2 87.0

3 8.3

4 2.9

(31)

Figure 11. Number of IJV’s with 2 and more than 2 partners per year 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 60 50 40 30 20 10 0 more than 2 2 partners number of partners

A large part of the IJV’s has shared equity. In a small part of the IJV’s is control shared unequally by more than 2 partners. Most IJV’s are not controlled by one of the partners, in only 48% of the cases is the IJV controlled by one of the partners. In 30% of the IJV’s, the Dutch firm is the controlling firm, compared to 18% for the non-Dutch partner (Figure 12).

Figure 12. Equity control

control foreign firm control unequally by more

than 2 firms control shared equally

(32)

Jagersma and Bell (1992) argue that this large part of shared control is due to habit and trust towards each other. However, they find that Dutch firms are taking more control in the later years of the 1980’s. In the period 1985-2006, the amount of IJV’s controlled by foreign firms is quite stable. The amount of IJV start-ups with Dutch equity control or shared equity control is more volatile. In the 1990’s, in which many IJV’s were set up, Dutch firms take relatively more often control, while before and after this period, it is the opposite (Figure 13).

Figure 13. Equity control in the period 1985-2006

2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 25 20 15 10 5 0

control foreign firm control unequally by more than 2 firms control shared equally control dutch firm

control by amount of equity

(33)

countries and 44% in transitional countries (Table 7). Dutch firms take significantly more often equity control in non-industrialized countries than in industrialized countries (t-value: 3.911; p<0.01)

Table 7. Control in the different country types

Control Dutch firm Control shared equally

Control shared unequally by more

than 2 firms Control foreign firm Industrialized countries 23% 51% 7% 19% NIC's 30% 59% 4% 7% Developing countries 40% 30% 11% 20% Countries in transition 44% 31% 13% 11%

Equity control can be obtained with only 50.1% of the equity, but in many cases, the equity ownerships are higher. Other motivations for equity ownership than obtaining control are for example the higher value of admitted resources, that one partner is able to invest more or to obtain a larger part of the profit. Dutch partners have relatively larger parts of equity as can be expected from table 7. Dutch partners have a significantly higher equity percentage (49.6%) than non-Dutch partners (46.3%) have (t-value: 3.554; p<0.01). Figure 11 shows that in about half of the years the percentage of ownership of the Dutch firm was above 50%. In the years 1990, 1994, 1995 and 1996, the equity percentage was significantly higher of the Dutch partners than of the non-Dutch partners (t-values: 2.731; p<0.05, 1.913; p<0.1, 2.644; p<0.05, 1.983; p<0.1).

Figure 11. Average equity share of Dutch firms during the period 1985-2006

(34)

Especially in the secondary sector, Dutch firms have a significantly higher equity percentage (52.3%) than non-Dutch partners (45.4%) have (t-value: 5.199; p<0.01). In the primary and tertiary sector that is not the case. Dutch firms in the chemical industry and the electronics industry, which are industries in the secondary sector, take a large part in that by taking respectively 54,2% and 50.9 % of the shares compared to 44.6% and 45.2% of the foreign partners (t-values: 4.642: p<0.01, 1.980; p<0.1). From table 7, it is not surprising that Dutch firms have significantly higher amounts of equity than the Dutch partners have in non-industrialized countries (t-value: 3.880; p<0.1), while that is not significantly higher in industrialized countries. Dutch Fortune global 500 firms and Dutch small firms have significantly higher average equity percentages than their partners (t-values: 2.776; p<0.01, 2.254; p<0.05). Large Dutch firms do not have significantly higher equity percentages than smaller Dutch firms. Foreign large firms however take significantly higher equity percentages than small foreign firms (t-value: 2.148: p<0.05). This is resulting into the finding that large foreign firms have, although insignificant, a larger equity share than Dutch partners, while small foreign firms have a highly significant smaller equity share than Dutch partners (t-value: 3.655; p<0.01). These results suggest that large firms prefer taking more control than small firms do.

4.1.3 Terminations

The first data collection of Jagersma and Bell (1992) excludes terminations of IJV’s. Hence, the statistics of terminations are only included in the period 1990-2006. Boersma (1999) wonders how many of the IJV’s would be terminated after her study. Figure 12 shows that after the large number of IJV start-ups in the 1990’s, the number of terminations increased by a large amount. In 2000, the maximum number of terminations was reached, with a total of 18. As can be expected, many of these terminated IJV’s were set up in the 1990’s13, as can be seen in figure 13. The percentage of terminations in the entire period 1990-2006 was 18.5% while in the set of 1990-1996 collected by Boersma this was only 10.9%14. The reason for this is that it Boersma includes a smaller observation period. The termination rate of the periods resembles if the entire data set is used. The termination rate per year in which the IJV’s were

13_{The later set up IJV’s only exist for a few years now. Figure 13 might therefore mislead that the IJV’s from}

the new millennium are more robust. It should be taken into consideration that in the coming years these IJV’s have a larger hazard rate than the earlier IJV start-ups.

14_{In the unpublished paper of Boersma (1999), the period 1985-1990 is also included in the percentages. This}

(35)

set up (except for the last three years) is between 5% and 25%, in which the termination rate is highest for the IJV’s set up in the period 1995-1997 and in 2001.

Figure 12. Number of IJV’s terminated per year

2007 2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 20 15 10 5 0

Figure 13. Year in which the terminated IJV’s were set up.

2006 2005 2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 1990 1989 1988 1987 1986 1985 12,5 10,0 7,5 5,0 2,5 0,0 termination

(36)

(34%), to concentrate more on core activities (25%) or in order to restructure the company (9%). In a small number of cases was the IJV left due to a conflict of interest (3%). These results correspond to the findings of Gomes-Casseres (1987).

The percentage of IJV’s that are ended in industrialized countries (23%) is significantly higher than in non-industrialized countries (12%) (t-value: 3.559; p<0.01). Boersma (1999) argues that most IJV’s are terminated in the tertiary sector. This thesis finds that the percentages of terminations in the primary, secondary and tertiary sector were respectively 18.2, 20.3 and 19.7 %. The termination rates of the sectors are not significantly different. Terminations are more frequent if the goal is not in development (t-value: 1.939; p<0.1). This corresponds to Schut and Frederikslust (2004), who find that shareholder value is increased more if the goal is development.

(37)

4.1.4 Conclusion for the descriptive statistics.

During the period 1985-2006 there are between ten and sixty IJV’s set up on a yearly basis (Figure 2). Almost half of these are undertaken by only ten large Dutch firms (Figure 3). This relative large part of IJV’s set up possibly reflects the international business environment of the Netherlands, where large firms may have a relative large part in. Verspage and Duyster (1994) for example argue that two out of the three foreign patents are obtained by only five large Dutch firms.

It stands out in the descriptive statistics that in the 1990’s a relatively high number of IJV’s was set up. This large increase in IJV start-ups was mainly done by small firms (Figure 3) in Europe and Asia. A large percentage of the IJV start-ups in Asia were set up in China (Figure 7). However, most IJV’s were set up in Europe. Most of the IJV’s in the 1990’s were in a horizontal cooperation in order to continue with their product of process. This is also the case for other periods (Figures 9 and 10).

Table 8. A comparison between the three observation periods

Dimension Period 1 Period 2 Period 3

Period 1985-1989 1990-1996 1997-2006

Number of observations 80 325 357

Average share Dutch partner 45.8% 51% 49.6%

Country foreign partner USA (22.5%) China (12.6%) USA (13.9%) and China (13.3%)

Country IJV USA (18.3%) Netherlands (17.6%) China (16.5%)

Continent IJV

Europe, Asia (both

36.2%) Europe (40.1%) Europe (45.8%)

Activity

production and selling

(33.8%) production (37.2%) production (39.5%)

Industry Electronics (36.6%) Chemicals (19.7%)

Financial institutes (15%) and Logistics and telecommunication (14.4%)

(38)

For future studies, it will be interesting to follow developments in which industries most IJV’s will be set up and what the characteristics of the partners and the IJV’s will be. The more important aspect for managers will however be, if this growth in the amount of terminations persists and on what it depends. The regression results of the Cox proportional hazard models in section 4.2 shed light on what the survival rate and the duration of IJV’s depends.

4.2 Regression results

The regression results of the Cox proportional hazard models are presented in table 9. Below, we discuss the results for the effect of national cultural distance, equity control and the control variables, respectively partner size, industry relatedness, the goal of the IJV and industrialization of the country in which the IJV is located, on the duration of the IJV’s.

4.2.1 National cultural distance

Scholars argue a negative effect of cultural distance on performance. This thesis does not come to the same results. The first hypothesis considers the effect of overall national cultural distance, measured with the Kogut and Singh (1988) index. Although the variable has the expected sign, the results are insignificant and hypothesis 1 cannot be accepted. The insignificant results of the Kogut and Singh index can be explained by the different signs of the dimensions. In the second model, including the Hofstede dimensions, cultural distance on two dimensions has a positive effect, while cultural distance on two other dimensions has a negative effect. Only the result for masculinity is significant with the expected sign. This corresponds to the results of Merchant and Schendel (2000). The cultural distance on individuality and the uncertainty avoidance index is hypothesized to have a negative effect on performance. The regression results of model 2 show, however, that this is the opposite, although insignificant. If the control variable of goal or industrialization is removed, this becomes significant for individuality. This supplies evidence for the opposite of the expectations of hypotheses 1.2 and 1.3. It also brings up the point that different dimensions might have different effects on IJV performance.

(39)

terminations, shorter duration and a smaller cultural distance15. A separation between IJV’s in industrialized and non-industrialized countries might supply a deeper understanding.

Table 9. A summary of the regression results of the Cox proportional hazard models 16 17 18 Expected

sign Model 1 Model 2 Model 3a Model 3b Model 4a Model 4b

Countries All All Ind. Non-Ind. Ind. Non-Ind.

Cultural distance PDI + 0,022 -0,036 0,070 (0,019) (0,039) (0,045) IDV + -0,018 -0,012 -0,036 (0,011) (0,014) (0,026) MAS + 0,0156* 0,006 0,086** (-0,009) (0,011) (0,044) UAI + -0,054 0,037 -0,466*** (0,045) (0,064) (0,180) K&S + 0,049 0,038 0,255 (0,085) (0,106) (0,221) Equity control

Control Dutch firm - -0,009 -0,025 -0,020 0,012 -0,021 -0,204**

(0,013) (0,019) (0,015) (0,029) (0,023) (0,089)

Control Large firm - -0,706 -0,743 -0,673 -1,403 -0,728 -1,167

(0,473) (0,489) (0,630) (0,888) (0,635) (1,069)

Control*Culture UAI + 0,000 0,001 0,000 0,000 0,000 0,009**

(0,000) (0,001) (0,000) (0,001) (0,001) (0,003) Control variables

Industry relatedness IJV - -0,044 -0,024 0,000 -1,057 0,177 -1,693*

(0,370) (0,377) (0,417) (0,855) (0,425) (0,978) Size + -0,460 -0,541* -0,194 -1,325* -0,178 -1,712** (0,282) (0,285) (0,314) (0,314) (0,328) (0,857) Goal - 0,535 0,566 1,172 -1,610 1,163 -2,963* (0,726) (0,741) (1,016) (1,139) (1,025) (1,540) Industrialization ? -0,954*** -0,945** (0,342) (0,476) Model statistics -2log likelihood 528,383 522,274 372,363 84,458 368,891 73,021 P model 0,021 0,009 0,447 0,214 0,445 0,056 n 173 173 107 65 107 65 *** significant at 1% level ** significant at 5% level *significant at 10% level

The results for both country groups are still insignificant for the Kogut and Singh index (model 3). The results of model 4 come to more interesting findings. An evident difference

15_{See appendix 3 for the results of the t-tests.}

16_{The summary statistics of the selected cases are available in appendix 2.} 17_{The complete regression results are available in appendix 4.}

(40)

between industrialized and non-industrialized countries arises. Firstly, the model fit of industrialized countries is lower. Secondly, it can be observed that the sign of the power distance index and the uncertainty avoidance index is different for the two groups. Finally, the results on cultural distance for IJV’s in industrialized countries are all insignificant. For the IJV’s in non-industrialized countries they are not.

The results of model 4b show a positive effect of cultural distance on performance for individualism and the uncertainty avoidance index, while a negative effect for the power distance index and masculinity can be observed. The findings for masculinity and the uncertainty avoidance are significant. The regression results of IJV’s in non-industrialized countries confirm the earlier findings that the effect of national cultural distance differs between the dimensions (model 4) and that overall cultural distance does not significantly decrease performance (model 3). However, a negative effect can be argued for some dimensions. The results of model 4b show in non-industrialized countries for the Hofstede dimensions individuality and the uncertainty avoidance index a positive effect on performance, while the opposite is hypothesized. This concludes to a rejection of the hypotheses of the effect of national cultural distance, i.e., hypotheses 1-3.

The results on national cultural distance are on the one hand disappointing as the hypotheses cannot be accepted. On the other hand, they are not, as this thesis comes to some interesting findings. The surprising result of having a negative effect of overall cultural distance, but having the opposite effect of cultural distance on the dimensions of individuality and the uncertainty avoidance index, points towards a different understanding. Culture is a difficult and complex concept. The idea to combine a complex concept into one index is probably oversimplifying. Lu (2006), for example, argues that scholars should stop using the Kogut and Singh index. The contradictory results of the effect on different dimensions support the argument of Lu.

(41)

firms might have come to the level of internationalization and size of the five largest Dutch firms. If that is the case, it might be that due to learning effects, national cultural distance is something that can be dealt with in business and is no longer a problem for these firms. The positive effect on performance of cultural distance on the dimensions of individualism and the uncertainty avoidance index is however not explained by this. Hence, other explanations are needed.

This thesis considers two possible explanations. Firstly, specific knowledge in certain environments may be more important than the difficulties arising with cultural distance on some dimensions. Performance is higher if the common goals are reached and the firms are able to do this better by working together (Maljers, 1999). The result could be that firms have improved their capabilities to work together, taking into consideration national cultural distance on certain dimensions. However, they still need the specific local or industrial knowledge. The effect being that for some dimensions, cultural distance would increase the performance in a certain environment. The earlier proposed effect on synergy would then be strongest with the combination of these different cultures.

A second explanation is that other country characteristics explain IJV performance. The results of the control variable of industrialization show such a characteristic. Such a cause might lie in the growth of the economy. Section 4.1.1 argues that the economy of China has grown fast and that the amount of IJV’s in China is a large part of the IJV’s in Asia. Other non-industrialized countries have also had a period of high economic growth. Gomes-Casseres (1987) argues that local partners are more likely to be bought out in low growth countries. This would have the consequence of a longer duration in high growth economies. Therefore, IJV’s located in regions of high economic growth are expected to perform better. The positive effect of cultural distance on individualism might therefore be explained by the fact that China has one of the lowest scores on this dimension;19 meanwhile China has been experiencing one of the highest economic growth rates. Hence, in this explanation the growth of the economy or other country characteristics affects IJV performance, not national cultural distance.

19_{China’s score on individualism is 20 compared to an average of Asia of 24, a world average of 43 and the}

(42)

4.2.2 Equity control

For the hypotheses of equity control, less conflicting results are found. The variables of equity control have in all models the expected sign. Significant results are less frequent. The first models, in which both country groups are combined, do not come to significant results. Only for the first model including the Kogut and Singh index, the value of control by large firm is significant, if the control variable of industrialization is removed. This only supplies limited response for hypothesis 5 and less support for the other hypotheses on equity control.

In model 4b, which includes IJV’s in non-industrialized countries, the effect of equity control is more apparent. Model 4a, which includes industrialized countries, holds poorly, like discussed earlier. Section 4.1.2 discusses that control in less developed countries is in the media argued to be more important. Different reasons can be brought forward, like more risk of loosing reputation or controlling patents in a less known environment. Among others, this is the reason why this thesis finds that Dutch control is higher in these countries (Table 7). The results suggest that higher Dutch equity control in these countries also significantly increases performance. Hence, hypothesis 4 can be accepted in the case of IJV’s in non-industrialized countries. This corresponds to the findings of Schut and Frederikslust (2004), who find a positive effect of Dutch equity control on shareholder value. Another aspect of non-industrialized countries is that it is should be perceived more risky by Dutch firms. The interaction term of cultural distance on the dimension uncertainty avoidance index and the amount of equity control shows a significant result with the expected sign. However, we should be cautious how to comprehend this finding. The effect of the uncertainty avoidance index is contrary to the expectations and if another measure of cultural distance, e.g. masculinity or the Kogut and Singh index is considered, the result changes considerable. In effect, hypothesis 6 cannot be accepted without doubt, even though the result is as expected.

(43)

significantly better, while others find that a large partner is more beneficial (Koh and Venkatraman, 1991; Schut and Frederikslust, 2004). Thus, the results of this thesis present that large firms in general perform worse, but IJV’s controlled by a large firm perform better. The results may show that the performance of large firms not in control is relatively low. If that is the case, one might wonder why large firms not always take control. Gomes-Casseres (1989) argues that even if firms would like to take control, governments not always allow firms to take it. China, for example has strict ownership restrictions in the observation period. However, other countries have similar restrictions in “key-industries”, like natural resources or defense. Such a restriction results in the dilemma for firms of not entering the market at all or to enter the market considering the ownership restrictions. Strategy and control should be in line with each other in order to obtain the best results (Geringer and Hébert, 1989). In this dilemma the firm has to choose between two evils. This results in worse performance of large firms, even though the firm might have taken the best choice. Another explanation is that two large firms working together have a large probability of being terminated. For the eleven IJV’s in which both firms are in the Fortune global 500, six have been terminated within the observation period. This can explain the findings in the earlier models as these terminations were in industrialized countries, while the first explanation is more plausible for IJV’s in non-industrialized countries.

(44)

5. Conclusions

5.1 Appraisal

The last decennia, there has been an enormous increase in the amount of alliances. (Contractor and Lorange, 2002). This started in the 1960’s and increased explosively in the late 1970 and in the 1980’s (Hagedoorn and Osborn, 2002; Jagersma, 1995). Scholars started to study the topic intensively and came to the conclusion that a relatively large percentage of joint ventures are terminated (Kogut, 1989; Gomes-Casseres, 1987). The reason for this is not that these have more liquidations, but that the partners come to different goals (Gomes-Casseres, 1987). Studies look for characteristics that increase the possibility of good performance in alliances, or more specifically in international joint ventures (IJV’s). This thesis analyzes two of these characteristics, i.e., national cultural distance and equity control. The relationship between national cultural distance and performance is set up by analyzing an adapted Uppsala model (Kogut and Singh, 1998; Barkema et al., 1997). The hypothesis is that national cultural distance is a similar barrier as physical distance. These barriers are expected to decrease performance in an IJV (Barkema et al., 1997). The second topic is the effect of equity control on the IJV performance. The relationship between equity control and performance is analyzed in a contingency approach in which control and strategy in line are expected to increase performance (Geringer and Hébert, 1989).

(45)

determine IJV performance. This thesis finds such an effect, IJV’s in non-industrialized countries perform significantly better. Another country characteristic that may determine IJV performance is the growth of the economy. Gomes-Casseres (1989) argues that IJV’s are less often left by the multinational corporations if the economy where the IJV is located grows more. In the last decades, economies with on some dimensions a large cultural distance and on other dimensions a small cultural distance have grown the most. The contradictory effects of the Hofstede dimensions on performance may be explained by the better performance of IJV’s in these fast growing economies, e.g. China. To test both explanations, a contingency approach would be in order.

The effect of equity control and national cultural distance on the performance of international joint ventures