Abstract
The aim of this research is to focus on under-researched questions of the formation rate of joint ventures (international and domestic) in the energy sector. The research also analyses the influence of financial crisis (2007-2010) on the formation rate dynamics. Following the existing literature, this paper conceptualizes the approach to the formation process and develops a model to evaluate the formation rate. The model has been tested on a sample consisting of the information about 805 deals. The information was obtained from the Zephyr database and the Thomson database. It has been cross-checked with the information from the Internet sources. Major findings indicate that the international joint ventures have a higher formation rate than domestic ones. Further, the crisis has a significantly negative impact on the formation rate. Finally, the deal’s parameters (deal size and deal duration) also influence the formation rate of joint ventures.
Acknowledgements
I owe my deepest gratitude to my supervisor, Dr W.Westerman. His professional and personal qualities, inestimable help and support made the conduction of this research much easier than was expected. His ability to explain and direct undoubtedly helped me to learn and understand a lot during this project.
I would like to thank my family members: my parents, Larysa and Nickolay, and my sister Anja. Their love and support brought me through all the difficulties. They always were and they always are with me in the toughest and brightest moments of my life.
Table of contents 1 Introduction ...5 1.1 Problem field ...5 1.2 Research question ...8 1.3 Research outline ...8 2 Theoretical background ...10
2.1 Joint venture alternatives...10
2.2 Domestic vs. international joint ventures...12
2.3 External factor – financial crisis ...15
2.4 Internal factors – deal characteristics...16
3 Research methodology...19
3.1 Sample and data...19
3.2 Event-study method ...20
3.3 Variables of the model ...21
3.4 Modelling procedure...23
3.4.1 Logistical Regression ...23
3.4.2 Model options ...24
4 Results and analysis...26
4.1 Descriptive analysis ...26
4.2 Results of tested hypotheses...27
4.2.1 Baseline model...27
4.2.2 Supplementary models ...29
5 Discussion of the results ...32
6 Limitations and directions for future research ...36
6.1 Limitations...36
6.2 Directions for the future research ...37
7 Conclusion and recommendations...39
1 Introduction
This section starts with description of the problem field, followed by the formulation of main research question and its sub-questions. The section will end with the detailed research outline.
1.1 Problem field
During the last 20 years the number of joint ventures (JV) has grown by more than 25 per cent annually (Bleeke & Ernst 1995). Joint ventures can be seen as an instrument by which companies can benefit from starting a new entity. It is the most common form of strategic alliance, suitable for both kinds of partnerships1: international (cross-border) and domestic (national) (Beamish, 1994). The term joint venture has been widely described in business literature. The most common definition is a hybrid form of organization which aims to combine skills and resources of two or more parent companies to accomplish particular objectives (Beamish, 1994; Duan & Juma, 2007). Any joint venture can be formed separately as an entirely new basis or as an association of several already existing divisions of two or more different companies (Zeira & Shenkar, 1990).
In the existing academic literature it is usual to distinguish between two inter-related definitions: equity joint ventures and non-equity joint ventures. The former refers to the creation of an entity with capital input of the partners, where each participates in the decision-making process of the joint entity (Geringer 1991). The latter refers to the agreement without creation of an entity, implying technical or construction service and management contracts (Prevot & Meschi, 2006).
Formation in the case of joint venture implies the successful deal completion and creation of a separate company within one country/region (domestic joint venture) or between geographically different partners (international joint venture) (Hanvanich, 2003; Egan, 2010). The result of joint venture formation is a separate entity with the shared control of assets (Grossman & Hart, 1986). As a result, the formation rate is the number of successfully completed deals to the total number of deals.
Formation itself is a considerably complicated process, including the second and the third stages of joint venture existence circle (Gray & Yan, 1997; Styles & Hersch, 2005). The first
1
stage includes the decision of partnership creation, while the second stage comprises of the negotiation process and the contract conclusion. The third stage involves the process of physical partnership creation followed by the fourth and final stage which is devoted to the maintenance and development of a partnership.
Traditionally, international joint ventures are more common than their domestic analogues. However, it is not necessary for companies to cross the border, as firms from the same country can benefit from a partnership not going global (Waggoner, 2005). Differences in culture, economic development, and infrastructure decrease the possibility of partnership creation (Dacin et al, 1997). As opposed to domestic joint ventures, international partnerships are presumed to be more complex to initiate and maintain. Despite this, international joint ventures can bring their parties significant benefits, not available in domestic joint ventures (Duma & Juan, 2007).
When firms from developed countries create a partnership with firms from less developed countries, they benefit mostly from significant costs savings (for instance, labour and materials). At the same time, firms from less developed countries have the opportunity to access technological know-how and the capital resources of foreign partners. Consequently, all partners benefit from taking part in a business alliance (Waggoner, 2005).
The reasons for starting joint ventures also differ on the domestic and international level. Foreign companies tend to start a joint venture in order to enter a particular market, using their local partner as an entry mode assistant. (Yan, 1998; Farrell et al., 2004). At the same time, domestic joint ventures are more likely to be formed when partners are trying to share the risks of any kinds of local operations (Meschi & Edson, 2008). Neither existing literature nor practical observations are able to provide an answer as to which type of joint ventures is more efficient. This is also a matter of industrial division, as the business sector considerably affects the process of formation and the performance of joint ventures (Janakiramanan, 2005; Egan, 2010).
unique combination of tremendous resources, part of which has not yet been discovered, and governing system including national states and multinational corporations (Jarvis, 2009).
Partnerships in the energy sector differ from those in other industries, where both partners contribute resources and tend to govern the partnership. The so-called ‘OBO model’ (operated by others) is widely spread in the energy sector, where one partner plays a dominant role over the other(s). Today, partnerships, namely joint ventures, are still preferred to acquisitions and divestitures, as they reduce the value, tax and regulation problems, and allow partners to contain reserves (critically important in the energy sector) as a hedge against price increases (Ernst & Steinhubl, 1997).
Initially, joint ventures in the energy sector have the main aim of sharing risks that are relatively high due to difficulties in extracting the resources. Discovered during the last decades, gas and oil fields are generally hard to reach, demanding the sophisticated systems of extraction and pipelines, and as a result bringing a potential risk to companies. Therefore, joint ventures are seen as a preferred option of sharing risks in the energy sector. Furthermore, there is another reason for international joint ventures to be important in the energy sector. As gas and oil fields are located globally, the partnerships are able to cut as many costs as possible, while exploring the resources together. Takeovers, presented by mergers and acquisitions, are less likely to appear in the energy sector due to the fact that gas and oil is related to the national sphere of interests, governed by the state. Therefore, multinational companies may face difficulties in obtaining the resources that belong to the state (Ruhl, 2010).
1.2 Research question
Based on previous studies, it can be concluded that domestic joint ventures in the energy sector should be easier to start and handle when compared with international joint ventures. This is due to the knowledge about the business environment, established relationships with the local government and existing production and distribution systems. However, this question remains under-researched, leaving us with a lack of knowledge regarding joint ventures formation rate across borders. Therefore, this research aims to answer the following question:
Which type of joint ventures, international or domestic, has a higher formation rate in the energy sector?
Being a rather complicated process, the formation of joint ventures can be affected by different factors. Special attention has to be devoted to the business environment (external factor) and specific deal characteristics (internal factor). As a result, those factors should be included in the model, posing several additional sub-questions:
Which internal and external factors influence the joint venture formation rate? How does the business environment influence the formation of joint ventures? To what extent do deal characteristics affect the formation rate of joint ventures?
The relevance of this research lies in the evaluation of the relationship between the joint venture formation rate and the type of partnership, with a further assessment of the latest financial downturn and its influence on the formation rate in the energy sector. Furthermore, this paper aims to help future researchers to understand better the processes taking place within the joint ventures formation and provide recommendations for the practitioners.
1.3 Research outline
incentives to launch a joint venture. After the literature analysis several hypotheses will be developed.
In the second part of the research, the methodological background of the study will be detailed with an emphasis on the method and variables. Additionally, the process of data collection will be described, providing more details on the sample of 805 joint ventures. Furthermore, special attention will be devoted to the description of the energy sector and the research time frames (2000-2010).
2 Theoretical background
This section provides information on alternatives to joint ventures and conditions necessary for their creation. Then, the comparison between international and domestic joint ventures is presented, including the formulation of the first hypothesis. Further, the influence of external conditions (crisis) on the joint venture formation is described, leading to the second hypothesis. This is followed by the explanation of the deal characteristics impact on the formation process, resulting in formulation of the other three hypotheses. The theoretical part will end with conclusions, summarized in the table.
2.1 Joint venture alternatives
Joint ventures have a number of advantages: transfer of knowledge and technologies, lower labour costs, restructuration of business portfolio, and access to new markets (Meschi, 2005). Despite having considerable advantages, joint ventures are considered to be relatively unstable business units (Duma & Juan, 2007). Previous studies on the survival of joint ventures in emerging markets have shown that around 60 per cent of announced deals were completed, while approximately 40 per cent of joint ventures were sold, bought out, or dissolved by the partners within 5 years after the formation (Miller et al, 1997; Yan, 1998; Kale & Anand, 2001; Nakamura, 2005).
Some joint ventures, in order to maximize their performance, can be governed in a way which places one partner in a more dominant position that the other. In the case of domestic joint ventures, it appears as though the bigger partner is in the dominant position. In terms of international joint ventures, it is often the foreign partner influencing their local counterpart. However, this can also influence the success of a partnership, as balance in shared management is necessary for the completion of long-term goals (Beamish & Killing, 1997). According to the transaction costs theory, there is a choice between full and partial ownership that depends on the benefits and costs of sharing ownership. Full ownership comprises takeovers and the establishment of wholly-owned subsidiaries, while shared ownership includes joint ventures and sometimes takeovers (Hennart, 1991).
All three alternatives have their advantages and disadvantages and should be described in detail.
The first alternative, the establishment of a wholly-owned subsidiary, implies that the firm operates in the foreign market and has headquarters in another country. The advantages of wholly-owned subsidiaries include full ownership and control by the parent company, expertise and consulting assistance. On the other hand, the major disadvantages are related to the high costs and considerable risks of exploration of a new market (Chan & Timsawat, 2001). An establishment of a wholly-owned subsidiary is favoured when the company possesses the relevant knowledge about the country’s market. Additionally, a successful establishment implies that the company has an antecedent relationship history with the host country’s government (Belberdos & Zou, 2007).
The second alternative to joint ventures is takeovers, mostly presented by mergers and acquisitions. Lambrecht (2007) describes takeover as a purchase of one company (target) by another company (acquirer). Takeovers are more long-term oriented, while joint ventures are likely to be used to obtain the knowledge about a local market in a short period of time (Inkpen & Beamish, 1997). Further, takeovers provide a greater level of control within the entity due to the hierarchical division of partners (Thomson, 1999). Belberdos and Zou (2007) found that joint ventures are less flexible than wholly-owned subsidiaries in responding to changing business conditions.
Departing from the transaction costs theory, full takeover is less efficient than joint venture when the desired assets cannot be detached from unwanted assets, and when the increase of management costs result in a decrease of management incentives (Hennart, 1991). Traditionally, takeovers are widely favoured in countries with developed markets and strong corporate control, while joint venture mode is preferred outside Europe, Japan and the United States (Thomson, 1999). Kogut and Singh (1988) argue that companies prefer joint ventures over takeovers in case of vast difference in national cultures (international joint ventures) or in case of high costs of project conduction (domestic joint ventures).
marginal costs (Hennart, 1991). In practice, this option is rarely used due to the high costs of resources.
2.2 Domestic vs. international joint ventures
The theoretical background for the formation rate of joint ventures derives from transaction cost economics, where the relationships between companies are a combination of opportunism and bounded rationality attributable to the participation in transactions (Williamson 1975; Williamson 1985). When it comes to partnerships, mangers tend to build trust, combining it with control mechanisms which reduce the probability of opportunistic behaviour (Parkhe, 1998).
Control itself and its mechanisms are widely described in the literature. The major finding is, the more there is of trust, the less there is of control, and vice versa (Inkpen & Currall, 1997; Leifer & Mills, 1996). It has also been proved that numerous partnerships suffer from poor control, partner’s conflict of interests, slow decision making process or a lack of trust (Groot & Merchant, 2000). For example, Zhang and Rajagopalan (2002) found in their research that inter-partner credible threat and form of control choice play the most important role in the partnership formation. Furthermore, Ding (1997) found that dominant foreign control has positively influenced the performance of international joint ventures (US-China). As Luo et al. (2001) specified, US partners have preference of dominant control, while Chinese partners tend to have control in functional areas related to the technology transfer.
Another stream of studies is devoted to the question of partnerships survival in the business environment (Lu & Hebert, 2005; Duan & Juma, 2007; Lowen & Pope, 2008). Some papers are devoted to the question of international versus domestic joint ventures survival (Meschi, 2005; Hanvanich et al., 2005), finding no significant proof of better performance by one of the partnership types. Lu & Hebert (2005), for instance, suggest that in the presence of high asset specificity, high levels of foreign equity control can lead to higher survival rates of joint ventures. Duan & Huma (2007) found that the existence of inter-partner credible threat, and high product relatedness between foreign parent and international joint venture, positively affects the survival rate of partnerships.
companies (Koh & Venkatraman, 1991; Min & Prather, 2001), while others reported the opposite (Reuer & Miller, 1997; Chang & Cheng, 2002). However, most of the researchers stated that a deeper research of the process within the joint ventures formation and related factors is necessary.
Relatively large group of studies examined the effect of cultural distance on the joint venture performance (Kogut & Singh, 1988; Park & Ungson, 1997; Meschi & Edson, 2008). The results of the conducted studies are quite contradictory. Eden & Miller (2004) found that an increase in national culture differences is caused by high unfamiliarity of the business environment. As a result partners need more time for understanding and maintaining the relationship. Barkema et al. (1996) claimed that the cultural differences can affect the long-term survival of the partnership. Lowen and Pope (2008), in contrast to previous literature, found that cultural distance between firms is not statistically significant for the survival of joint ventures. Although the cultural issue has been described in the literature, no results found in relation to its influence on the formation process of partnerships.
The institutional approach in organizational theory states that companies relying on ‘culturally’ approved resources are legitimized by such organizations as the state and regulation agencies (Meyer & Rowan, 1977; DiMaggio & Powell, 2983). Consequently, they can access markets to obtain resources relatively easily (Scott, 2001). Their environment is more stable and their chances of survival are greater. However, the institutional approach is not well adjusted to the crisis management, mainly because of the lack of flexibility and a continuous learning process (Ouedraogo, 2007).
Finally, a group of authors has studied the motives and conditions of partners pushing them to start a joint venture (Park & Zhou, 2005; Estrada et al., 2010). For instance, Koza and Lewin (1998) and Park et al. (2002) proved the need to consider market and firm conditions together to explain alliance formation. Garcia-Pont and Nohria (2002) showed that companies in order to maintain and increase their market share are ready to form a partnership with potential competitors. Moreover, companies need a specified strategy of the formation process, helping them to avoid additional expenditures of time and money in the future (Estrada et al., 2010).
joint ventures, arguing that abnormal returns influence the formation rate of joint ventures in a positive way. Pan and Chi (1999), for example, have discovered in their research of joint ventures that the most profitable partnerships were likely to be those focused on the local Chinese market. At the same time, there are a number of studies disclaiming this fact. Child and Yan (2003), in contrast, have found that domestically formed partnerships have the same formation rate as international partnerships. In fact, despite the numerous literature sources on joint ventures, specific studies addressing the issue of formation rate are necessary.
The relation between the formation rate and the type of joint venture, international versus domestic can be found in studies devoted to cultural similarity. Thus, in joint ventures where partners represent the same culture or nation, a higher tendency towards agreement between parties is expected (Geringer & Hebert, 1991). In joint ventures with partners from different cultures, the agreement process is relatively more complicated. Though, as it was described by Anderson and Weiz (1989), cultural similarities promote a higher level of communication between partners, what will result in higher formation rate of joint ventures. The major supporting argument is a study conducted by Park and Ungson (1997), where 186 international and domestic joint ventures were studied. The major findings indicate that international partnerships Japanese) were easier to form and maintain than domestic (US-US) partnerships.
The localization of joint ventures is an important factor inseparably linked with the type of partnerships, influencing the joint venture formation rate (Dunning, 1988). Following the Hanvanich et al. (2003; 2005) research, location of the partners in this research also represents such variables as cultural difference in deeper issue rather than just division of partnerships into international and domestic. Therefore, Lu and Hebert (2005) proposed the idea that investigation of joint ventures, established by firms from other countries in other regions of the world, is crucial to understand the influence of location on the formation rate.
Based on the literature, domestic joint ventures have a higher chance to survive than international joint ventures due to the fact that it is considerably easier for local partners to start a partnership within one country (region) than search for the partner(s) outside (Hanvanich et al., 2003; Duan & Juma, 2007). Cultural differences can lead to instability in the negotiation process, resulting in the better performance of the partnerships, facing no cultural dissimilarities (Woodcock et al., 1994). The lack of cultural distance results in the minimization of managerial complexity (Makino & Beamish, 1998). Therefore, based on the existing literature, the first hypothesis is developed as follows:
H1. Domestic joint ventures in the energy sector have a higher formation rate than international joint ventures.
2.3 External factor – financial crisis
Another important factor in the area of partnership formation is the external conditions in which it has to operate. It has been proved by numerous studies that the well-being of companies is significantly affected by the business environment (Chan & Timsawat, 2001; Pangarkar, 2007; Somers, 2009). Taking into consideration the consequences of the latest financial crisis (2007-2009), its influence on the joint venture formation has to be investigated in depth.
The majority of previous studies on environmental shocks is based on the ecological theory (Hannan & Freeman, 1984), arguing that companies in times of distress are tend to use inertial tendencies. Those tendencies have impact on internal mechanisms of companies, resulting in reduced resistibility during the times of financial downturn. Furthermore, industrial organization economics models conclude that the performance of a partnership is affected by financial downturns. This is due to the lower volume and losses of scale economies, reduced capacity utilization and increased competition (Van Witteloostuijn, 1998). Finally, the managerial theories also find support that partnerships are considerably influenced by external conditions. The agency theory, for example, states that managers within the partnership operating in times of crisis could have their own view of the situation different from the vision of shareholders (Pangakar, 2007; Somers, 2009).
alliances formed by Singapore companies, found that those partnerships that are unable to react to the changing conditions of the business environment have fewer chances to start long-term relationships during the times of financial distress. Furthermore, the extent to which joint ventures are affected by crisis is important, as some types of joint ventures are more vulnerable than others (Ravichandran, 2009). Finally, a high economic uncertainty is associated with a risk rate of joint ventures that decreases with time (Meschi, 2005).
While no evidence of influence of the financial environment on the joint ventures formation was found in the literature, it can be concluded from the literature devoted to the survival of partnerships during the crisis, that financial distress negatively influences the formation rate of joint ventures (Chan & Timsawat, 2001). Thus, the second hypothesis has been formulated as follows:
H2. Financial crisis has a negative impact on the joint ventures formation rate in the energy sector.
2.4 Internal factors – deal characteristics
Next to external factors, internal factors influence the process of joint venture formation. Such internal factors include: deal size, number of partners, and duration of deal. Following the transaction cost theory, a company should maximize the result (deal size), whilst minimizing the time frames (deal duration). For negotiations, the ultimate goal is to reach an agreement between the optimal number of partners (Yang & Tien, 2005).
According to the literature, deal size is one of the most important factors influencing the joint venture creation. Gupta and Misra (2007) found that mergers of smaller companies are faster and more efficient than mergers of large companies. Furthermore, Greenberg (2010), in the separate investigation of mergers and acquisitions, stated that smaller deals were more successful than larger ones. Additionally, the size of any deal, including joint venture formation, can be affected by the latest financial crisis (Bank Loan Report, 2008). Although mechanisms within mergers and acquisitions are not the same as those in joint venture deals, we assume that smaller deals have higher formation rate than larger deals. Therefore, the next hypothesis is suggested:
H3. Joint ventures with smaller deal size have a higher formation rate in the energy sector.
participants increases, transaction costs also tend to increase, as well as the complexity of negotiation process between parties (Beamish & Lupton, 2009). Moreover, the chance that a dysfunctional pairing will destabilize the joint venture increases with the number of parties involved in that process (Park & Russo, 1996).
Several studies have discovered that the presence of costs related to the control over the joint venture formation decreases the willingness of partners to create a joint venture, as with the increase of partners the costs will grow respectively (Lu & Hebert, 2005). Supporting this idea, Makino and Beamish (1998) discovered that joint ventures with only two partners are likely to outperform those with three or more partners. Hence, a hypothesis is formulated as follows:
H4. In the energy sector joint ventures with less number of partners have a higher formation rate.
The third factor, deal duration, is also relatively important for joint venture creation (Yang & Tien, 2005). It is assumed in the literature that deal duration is the time necessary for deal completion and calculated as the difference (in days) between announcement of joint venture creation and its official completion (Dikova et al., 2010). However, it should be mentioned that not all announced deals are completed. Some announcements can be out of date, while others have non-actual duration due to undisclosed information.
Several studies discovered that longer negotiations between joint venture partners slow down the process of partnership creation (Lowen & Pope, 2008). At the same time, Miller (1997) found no relationship between the length of time required to complete the deal and the venture creation. Dikova et al. (2010) discovered that the duration of cross-border deals increases in institutionally closer environments, but shortens in institutionally distant environments. However, it should be noted that both latter studies are not related to the field of joint venture formation. Following the previous studies the next hypothesis can be formulated as follows:
H5. Joint ventures with shorter duration of deal have a higher formation rate in the energy sector.
findings were made in spheres connected with the joint venture formation process. The summary of major findings is presented in Table 1.
Table 1. The theoretical summary of developed hypotheses.
Hypothesis Literature support/rejection Major findings
Support: McConnell &
Nantell (1985); Styles & Hersch (2005)
Anderson & Weiz (1989); Geringer & Hebert (1991)
There is a positive market reaction to the announcements of domestic joint ventures
In joint ventures with culturally close partners a higher tendency towards agreement is expected H1. Domestic joint ventures have a higher formation rate than international joint ventures
Rejection: Park & Ungson
(1997)
Child & Yan (2003)
International partnerships are easier to form and maintain than domestic partnerships
Domestically formed partnerships have the same formation rate as international partnerships
H2. Financial crisis has a negative impact on the joint ventures formation rate in the energy sector
Support: Pangarkar (2007);
Ravichandran (2009)
Partnerships that are unable to react to the changing business conditions have less chances to start the cooperation during the times of financial distress
H3. Joint ventures with smaller deal size have a higher formation rate in the energy sector
Support: Gupta & Misra
(2007); Greenberg (2010)
Mergers of smaller companies are faster and more efficient than mergers of larger companies
Support: Park & Russo
(1996); Makino & Beamish (1998)
Joint ventures with a small number of partners are likely to outperform those with a higher number of partners
H4. In the energy sector joint ventures with less number of partners have a higher formation rate
Rejection: Beamish &
Kachra (2004)
The relationship between the partnership performance and the number of partners is insignificant
Support: Lowen & Pope
(2008); Dikova et al. (2010)
Longer negotiations between partners slow down the process of partnership creation H5. Joint ventures with shorter duration of deal have a higher formation rate in the energy sector
Rejection: Miller (1997) The relationship between the time
3 Research methodology
This section of the research has the following structure. Sample and data features are to be discussed first, followed by the description of an event study method and variables included in the research. Then modelling section appears, consisting of two sub-sections: logistic regression characteristics and description of supplementary models.
3.1 Sample and data
The main sources of information about deals in the energy sector are the Zephyr database and the Thomson database of strategic alliances. Both databases use a variety of sources, including international news, media and trade publications, to collect the information on joint ventures, mergers and acquisitions world-wide. Additionally, it should be stressed that all the information obtained from the databases has been cross-checked with the help of companies’ web-pages and annual reports. The major purpose of this step is an attempt to avoid the misinterpretation or double counting of the obtained data.
According to the existing literature, the energy sector includes coal mining, electric services, pipelines, gas and oil extraction (Egan, 2010). Following the Standard Industrial Classification (SIC) classification the next industries were included in the research: ‘Coal mining’ (SIC 12), ‘Oil and gas extraction’ (SIC 13), ‘Petroleum refining and related industries’ (SIC 29), ‘Pipelines, except natural gas’ (SIC 46) and ‘Electric, gas and sanitary services’ (SIC 49). Alternative sources of energy (nuclear energy, solar, wind and water energy) were excluded from the sample as they apply a different system of taxation and regulation, influencing the process of partnerships creation.
Primarily the dataset consisted of 4067 deals, announced in the energy sector between 2000 and 2010. The number of deals with announced deal size included in the sample was 1160 observations. Finally, out of 1160 deals 355 were excluded due to the unsatisfactory parameters (absence of the duration information, deals in a process of current formation, etc.), resulting in the total sample of 805 deals. The time frames for the sample were built as follows: from the 1st of January 2000 till the 31st of December 2010, comprising all stages of economic cycle, namely: two large financial downturns of 2001 and 2008 and a financial boost of 2006.
3.2 Event-study method
Taking into consideration the fact that this research includes such parameters as the duration of deals, an event-study method is the sufficient way of analysing the data. One of the most important issues in the event study is to select relevant time frames (Koh & Venkatraman, 1991). Almost all articles concerning the announcements of deals and their duration are based on the research of the event-study method conducted by Brown and Warner (1985). However, this model has been adjusted, as in the current research we measure the formation duration, but not the wealth effect of announcements on shareholders’ value.
The day of joint venture formation is considered as the event date (day ‘0’ or t = 0). The days prior to the formation are: day ‘– 1’, day ‘– 2’, and so on, up to the announcement date. As it was mentioned previously, in order to eliminate the influence of time frames on the sample structure, only joint ventures with the duration longer than 30 days were included in the sample. A special concern was the withdrawn deals, since the term itself implies the deals that are out of date, as they have never been completed2. Since the number of observations with clearly specified duration period is 233 deals, the rest of the sample (572 deals) had to be adjusted. Based on the previously conducted studies and the collected information from the databases, we made a range of assumptions, concerning the duration of these deals.
acquisitions despite their industry relatedness is approximately 180 days or 6 months (Ekelund et al., 2001; Dikova et al., 2010).
The second assumption is pertinent to the duration of uncompleted deals. Since the number of uncompleted deals with an announced duration is relatively small (12 deals with average duration of 660 days), it would be statistically incorrect to make an assumption on such a small sample. Therefore, we made an assumption by extending the time frames for completed deals, applying the duration of uncompleted deals of 365 days or one year. The analogous number in the literature appears to be around 180-270 days for mergers and acquisitions (Ekelund et al., 2001). However, we assumed the number to be higher here, since the process of joint venture formation is relatively more time consuming than mergers and acquisitions, since it requires the creation of a separate entity. Additionally, this number is closer to the calculated duration of uncompleted deals from the sample, namely 660 days.
Furthermore, the deals with the same announcement and formation days were excluded from the sample in order to prevent the influence of such deals on the research outcome. The possibility that these deals were conducted within one day is considerably low. It is most likely that there were no announcements and rumours previously to the formation. Additionally, there were several situations when all companies taking part in the deal represent the same country, but one of them is registered in the off-shore zone. Nominally these deals can be considered as international. However, in reality they pursue the only aim of reducing the tax burden. Thus, these deals were excluded from the sample either. We also excluded stock market returns to deal announcements, which have been widely described in the literature. The focus of the current research is on the formation process rather than on abnormal returns caused by the announcements.
3.3 Variables of the model
The main dependent variable, formation versus non-formation, is dichotomous (formation = 1, non-formation = 0) and has been named FORMATION. While analysing the data, joint ventures were indicated to be formed if they have a date of completion, official address, web-site or any other relevant information about the partnership existence. If such information was absent or unavailable, then deals were excluded from the final sample.
CROSS-BORDER DEAL. If at least one of the partners, despite its equity share, represents the country different from the country of other partners, we indicated this partnership as cross-border or international. At the same time, if partners originate from the same country, the joint venture is considered to be national or domestic (Hanvanich, 2005).
Based on hypotheses developed we implemented four other independent variables. The first one is deal size, estimated in Euros (DEAL SIZE). The higher the amount of investments provided by the parties, the lower the possibility of joint venture formation, due to high risks caused by building trust process between partners (Hanvanich, 2005). The second independent variable is deal duration, which is calculated as difference (in days) between the dates of completion and announcement of joint venture creation (DURATION OF DEAL). The third variable, number of partners, can have a double-sided effect (NUMBER OF PARTNERS). On the one hand, a higher number of partners bring higher input in the joint venture, both during the creation and operation. On the other hand, it raises the complexity of the formation process, due to the difficulty of combining interests of all partners participating in the joint venture (Meschi, 2005; Pangarkar, 2007).
The last independent variable indicates the presence of the latest financial crisis (crisis time = 1, non-crisis time = 0). The variable has been named CRISIS in the model. The inclusion of this variable can be explained by the fact that the time frames of the sample reflect the times of the latest financial distress. We believe that the slowdown of the business activity caused by the crisis can have significant influence on the formation rate of joint ventures in the energy sector. In order to determine time frames of the crisis we adjusted Pangarkar (2007) research that studied the influence of the Asian crisis on strategic alliances. Additionally, a study conducted by Ravichandran (2009) has also been implemented to set the crisis borders. Based on these two studies, the time frames for crisis factor are: before the crisis January 2000 – October 2007 and during the crisis October 2007 – December 2010.
consist of two parties. Thus, if the partner of joint venture represents the region it has been coded as 1, otherwise = 0. It should be mentioned that this variable also reflects the cultural aspect to some extent. As we can assume that countries within the same geographical region have more similar cultural backgrounds than countries from different regions (Meschi & Edson, 2008). The variables in the model were coded as AFRICA, ASIA & OCEANIA, E.EUROPE, MIDDLE EAST, N.AMERICA, S.AMERICA and W.EUROPE for the first partner; and AFRICA 2, ASIA & OCEANIA 2, E.EUROPE 2, MIDDLE EAST 2, N.AMERICA 2, S.AMERICA 2 and W.EUROPE 2 for the second partner.
3.4 Modelling procedure 3.4.1 Logistical Regression
In total, the model together with the dependent variable includes 20 variables, with 14 control variables representing the regional division of partners. Most of the variables are dichotomous (binary coded), except deal size, deal duration and number of partners. In order to eliminate the negative influence of the number of dummy variables on the model output, we used the Logistical Regression (LR) estimation instead of standard Ordinary Least Squared (OLS) estimation.
Additionally, it should be mentioned that in order not be caught in a so-called ‘dummy
variable trap’ (the case of perfect multicollinearity between parameters) one of the
geographical regions had to be excluded from the model. Otherwise, the probability of this variable would always be close to 1, since the sum of all regions in a binary coded model is always 1. Hence, we excluded the least significant region of the sample – Africa (1.74 per cent of the total sample), resulting in the decrease of total number of variables from 20 to 18 (while 12 variables represent the geographical region of the partner).
One more concern in the model is the comparability of the dataset. Since the majority of the variables in the model are dichotomous, such values as deal size and deal duration have to be adjusted to make the dataset possible for comparison. Hence, the logarithm of parameters mentioned has been used to keep the data level. It is also worth mentioning that the number of partners was excluded from the logarithming, since its value varies between 2 and 11, thus being suitable for the estimate.
The final formula after all necessary adjustments has the following look:
FORMATIONi,t = β1*CROSS-BORDER DEALi,t + β2*CRISISi,t + β3*DEAL SIZEi,t +
β4*DURATION OF DEALi,t + β5*NUMBER OF PARTNERSi,t + β6*ASIA&OCEANIAi,t +
β7*ASIA&OCEANIA2i,t + β8*E.EUROPEi,t + β9*E.EUROPE2i,t + β10*MIDDLE EASTi,t +
β11*MIDDLE EAST2i,t + β12*N.AMERICAi,t + β13*N.AMERICA2i,t + β14*S.AMERICAi,t
+ β15*S.AMERICA2i,t + β16*W.EUROPEi,t + β17*W.EUROPE2i,t + εi,t
(Eq. 1)
Where i – deal, t – time, β – coefficients, ε – standard error.
3.4.2 Model options
It should be noted that the number of observations and variables in this research is considerably high. The results of tests can vary depending on the data parameters and variables included. We decided to run several models changing the number of variables and the data features in order to check the model performance under different circumstances. However, the total number of possible options is countless. Therefore, we propose the baseline model (model with the original data from the sample) and two other adjusted models that will be compared with the baseline model and each other. Two adjusted models are briefly described in this section.
for the deal size, the median of all deals was taken into consideration, namely – 20 mln Euros. The median instead of average has been used due to the high volatility of this parameter. Hence, all deals with the size higher than 20 mln are considered to be large deals (coded as 1) and those below 20 mln to be small deals (coded as 0).
The same technique with the cut-off point has been used for the number of partners. Since 70 per cent (562 observations) consist of two partners, these partnerships were coded as 0. Joint ventures with the number of partners higher than two were coded as 1. Finally, based on the assumptions made in the related studies, for the duration of deal the cut-off point of 365 days has been used. Thus deals longer than 365 days are considered to be relatively long and the opposite to be short. The former were coded as 1, the latter as 0. The major aim of the described model is to level the whole dataset to be the binary-coded and to check the influence of this model on the final outcome.
The second model reflects the observation that can be made from Figure 1, implying that the majority of deals in the energy sector in 2000-2010 took place in the Asia-Oceania region – almost 62 per cent. Thus, the dataset has been lightened by excluding all control variables concerning the geographical region, leaving only two of them: ASIA & OCEANIA and ASIA & OCEANIA 2. This adjustment helps to avoid the overload of the model by variables related to the geographic region and, as a result, to check the model performance, keeping the major research question untouched.
61,96 16,44 7,85 5,11 3,49 3,42 1,74 0,00 10,00 20,00 30,00 40,00 50,00 60,00 70,00 Asia & Oceania Western Europe North America Eastern Europe Middle East South America Africa
4 Results and analysis
In this section the results of the research are discussed, starting with the descriptive analysis and followed by the results of tested hypotheses. The results are derived from two types of models, namely, the baseline model and the supplementary models.
4.1 Descriptive analysis
The energy sector is one of the most profitable and perspective markets in existence. However, as it was mentioned previously, the energy market is also considerably unpredictable and volatile. Indeed, 48.8 per cent of all joint venture deals in the last ten years were not completed, while only 22.9 per cent (74 bln out of 323 bln Euros) of all announced equity took place in the completed deals. These relatively low figures might have scared the managers of the energy companies, but the benefits of starting a joint venture in this field are deemed to exceed the expected losses. Hence, the prognoses for the near future are quite optimistic, despite the unstable situation in the market within the last ten years.
As can be seen from Figure 2 the market of joint venture deals is highly exposed to the external business situation. It was affected by the crisis of 2001 and the latest financial crisis of 2007-2010. There is of course a lag in the crisis influence and the deal size decrease, as the formation process is considered to be rather time consuming. Thus, the deal size started to decline in 2003, falling to 50 per cent of the 2002 volume. It is a remarkable fact that while the deal size plummeted, the number of deals went up, hitting its historical maximum in 2004 – 70 deals per year. This paradox can easily be explained by an attempt of the energy companies to keep the success rate of deals by switching to smaller deals that were presumed to be more efficient in the crisis time (Ruhl, 2010). However, this was not true for the market and the number of deals went down to its average of 40 deals the next year (2005).
small-sized deals has been tried out by companies, resulting in a high gap between deal size and number of deals in 2009. Finally, in 2010 with the slight recovery of the world’s economy the dynamics of deals in the energy sector has returned to a positive pace. Despite the decrease in number of deals, the deal size is back on the growing trend, giving reason for optimism in the near future. 0 10 20 30 40 50 60 70 80 0 2000000 4000000 6000000 8000000 10000000 12000000 14000000 16000000 18000000 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Deal size (in thd. EUR) Number of deals
Figure 2. The dynamics of completed deals (deal size and number of deals) in the energy sector in 2000-2010.
4.2 Results of tested hypotheses
To test the hypothesis three models were run, using the Eviews package. The first model is the baseline one, while two others are supplementary models and were run in order to check the dataset sensitivity towards changes. We assume the results of the baseline model to be the major results of the research that would be compared with the results of supplementary models. The description of the obtained results starts with the baseline model.
4.2.1 Baseline model
the confirmation of suggested hypotheses. Therefore, it shows that three hypotheses out of five were confirmed.
Table 2. Results of regression analysis (baseline model)
Variable Coefficient Standard Error
Cross-border deal 0.545* 0.194
Crisis -0.891* 0.182
Deal size -0.001** 0.001
Duration of deal -0.011* 0.001
Number of partners 0.040 0.080
Asia & Oceania 1.632** 0.838
Eastern Europe 1.693*** 0.917
Middle East 1.462 0.928
North America 2.216** 0.900
South America 0.843 1.010
Western Europe 2.198* 0.842
Asia & Oceania 2 -1.723 0.872
Eastern Europe 2 -1.196 0.962 Middle East 2 -0.696 1.009 North America 2 1.724*** 0.925 South America 2 -1.421 1.049 Western Europe 2 1.191 0.865 * − p<0.01 %; ** − p<0.05%; *** − p<0.1%
The coefficient +0.545 has a positive sign, meaning that cross-border deals do not have a lower formation rate than domestic deals. This implies that every upcoming international deal has 54.5 per cent probability to be completed, holding everything else equal. Hypothesis 2 (crisis), Hypothesis 3 (deal size) and Hypothesis 5 (duration of deals) were confirmed, having negative coefficients of -0.891, -0.001 and -0.011 respectively. The fourth (number of partners) hypothesis has not been confirmed due to its insignificance.
The tested model has the following look:
FORMATIONi,t = 0.545*CROSS-BORDER DEALi,t + (-0.891)*CRISISi,t + 0.001*DEAL
SIZEi,t + (-0.011)*DURATION OF DEALi,t + 0.040*NUMBER OF PARTNERSi,t +
1.632*ASIAi,t + (-1.723)*ASIA2i,t + 1.693*E.EUROPEi,t + (-1.196)*E.EUROPE2i,t +
1.462*MIDDLE EASTi,t + (-0.696)*MIDDLE EAST2i,t + 2.216*N.AMERICAi,t +
1.724*N.AMERICA2i,t + 0.843*S.AMERICAi,t+ (-1.421)*S.AMERICA2i,t +
2.198*W.EUROPEi,t + 1.191*W.EUROPE2i,t + εi,t
(Eq. 2)
Where i – deal, t – time, β – coefficients, ε – standard error.
The effectiveness of the logistical regression model is assumed to be appropriate for the baseline model. Pseudo-R2 (McFadden’s R2) is 0.304 which is just in the middle of the acceptable level (0.2-0.4). The Hosmer-Lemeshow test also indicates that the model fits well to the data (Chi-square statistic – 157.3, with the probability of 0.001 per cent). The likelihood ratio is also significantly higher than the norm – 338.3, while its probability is an infinitesimal number. The ‘White test’ did not provide any big difference for the results, with only a slight change in the probability of the number of partners (Appendix A). However, this parameter still remained insignificant, thus providing no changes to the model outcome.
Table 3 shows the correlation between the variables, included in the model, excluding the dependent variable. There is no high correlation (>0.8) found between the independent and control variables. We assume those coefficients to be not high enough to influence the outcome of the model. The highest correlation coefficient 0.676 was found between control variables representing geographical regions, namely Asia-Oceania. It can be explained by the fact that a significant number of deals in the region were presented by the same companies, ‘China National Petroleum Corporation’, for example.
4.2.2 Supplementary models
Table 3. The correlation matrix of the baseline model
** Correlation is significant at 0.01 level * Correlation is significant at 0.05 level
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17) (1) Cross-border deal 1 (2) Crisis -0.051 1 (3) Deal size 0.154** -0.016 1 (4) Duration of deal 0.143** 0.103** 0.172** 1 (5) Number of partners -0.047 -0.019 0.076* 0.047 1
(6) Asia & Oceania -0.249** 0.061 -0.229** -0.023 0.111** 1
(7) Asia & Oceania 2 -0.217** 0.072* -0.236** -0.088* 0.090* 0.676** 1
As it was expected to be, the effectiveness of the logistical regression model grew significantly, comparing with the baseline model. Pseudo-R2 (McFadden’s R2) raised more than twice – from 0.304 to 0.721, which indicates a high level of fitted data. The Hosmer-Lemeshow test also indicated that the model was fit to the data well (Chi-square statistic – 15.63, with the probability of 5 per cent). The likelihood ratio has also become significantly higher – from 338.3 to 802.7, while its probability is an infinitesimal number. The ‘White test’ resulted in the decreased probability of the crisis variable, leaving the values of coefficients and standard errors practically unchanged (Appendix B). The increase in the goodness-of-fit parameters and likelihood ratio is derived by a decrease of variables number.
Table 4. The results of regression analysis (first supplementary model)
Variable Coefficient Standard Error
Cross-border deal 0.890* 0.328 Crisis -0.811** 0.317 Deal size -0.695** 0.344 Duration of deal -8.364* 1.037 Number of partners 0.218 0.319 * − p<0.01 %; ** − p<0.05%; *** − p<0.1%
The second adjusted model implied the exclusion of all control variables, except for Asia and Oceania, since this region represents more than 60 per cent of all deals in the energy sector. The results of the regression are presented in Table 5. The major difference with the baseline model is the non-confirmation of Hypothesis 3 (deal size) due to its insignificance3. Other results of the regression are similar to the first supplementary model. Apart from slight changes, the effectiveness of the logistical regression model is almost the same as in the first supplementary model.
Table 5. The results of regression analysis (second supplementary model)
Variable Coefficient Standard Error
Cross-border deal 0.952** 0.331
Crisis -0.858*** 0.316
Deal size -0.166 0.151
Duration of deal -8.260* 1.032
Number of partners 0.056 0.127
Asia & Oceania -0.158 0.321
* − p<0.01 %; ** − p<0.05%; *** − p<0.1%
3
5 Discussion of the results
This section will provide the discussion on the results described in the previous section. Additionally, special attention will be devoted to the comparison of results obtained in the previous section with the analogous studies in the existing literature. The summary of all confirmed and not confirmed hypotheses is presented in Table 6.
Table 6. The summary of tested hypotheses and literature findings
Hypothesis and the result Main literature findings
H1. Domestic joint ventures have a higher formation rate than international joint ventures – Not confirmed
Park & Ungson (1997). International partnerships are
easier to form and maintain than domestic partnerships
Child & Yan (2003). Domestically formed partnerships
have the same survival rate as international partnerships
H2. Financial crisis has a negative impact on the joint ventures formation rate in the energy sector – Confirmed
Pangarkar (2007); Ravichandran (2009). Partnerships
that are unable to react to the changing business conditions have fewer chances to start cooperation during the times of financial distress
H3. Joint ventures with smaller deal size have a higher
formation rate in the energy sector – Confirmed
Moeller et. al (2004); Gupta & Misra (2007); Greenberg (2010). Deals between smaller companies
are faster and more efficient than deals between large companies
H4. In the energy sector joint ventures with less number of partners have a higher formation rate – Not confirmed
Beamish & Kachra (2004). The relationship between
the partnership performance and the number of partners is insignificant
H5. Joint ventures with shorter duration of deal have a higher formation rate in the energy sector – Confirmed
Lowen & Pope (2008); Dikova et al. (2010). Longer
negotiations between partners slow down the process of partnership creation
Gas and oil fields are located all over the world. It is obvious that not all countries have the same level of technological development to be able to extract resources without any losses and additional investments. Joint ventures in the energy sector can be seen as the preferred option of transferring technologies and sharing risks. Mergers and acquisitions are less common in the energy sector due to the fact that natural resources (especially, gas and oil) represent national interests of a state. Hence, multinational companies face problems in acquiring the right to use resources that belong to the state, while joint ventures have advantages for both: the domestic and the international partner (Ruhl, 2010).
Nevertheless, the non-confirmation of the first hypothesis is still in line with several studies. These studies represent the minority of the existing literature stating that international partnerships can be more efficient than domestic analogues. For instance, Park and Ungson (1997) suggested, in contrast to the majority of studies, that international partnerships are easier to form and maintain than domestic partnerships. Child and Yan (2003) have also found that domestically formed partnerships have neither higher nor lower survival rates compared to international partnerships. Additionally, on the local Singapore study, Pangarkar (2007) found that those involving at least one partner from the Western country exhibited a better likelihood of survival during the Asian economic crisis. Hence, the mentioned studies determined that a reduction of failure risk can be achieved through international diversification.
The second hypothesis, that the crisis has negative impact on the partnership formation, has been confirmed. Indeed, the existing literature states that the external business environment may influence the formation and performance of joint ventures. When the economy is flourishing, the formation tends to be easier, since financial resources are available. In contrast, when the economy is declining, the formation is complicated, as resources are expensive and the financial performance is deteriorating. Thus, Pangarkar (2007) and Ravichandran (2009) indicated that more than 60 per cent of partnerships will not survive during times of economic downturn, due to the lack of financial resources and increased competition between partnerships. Both authors conclude that partnerships that can enhance revenue potential in the short-term are more stable to environmental shocks.
Gupta and Misra (2007) found that mergers and acquisitions with smaller deal sizes were more successful than large deals. It is obvious that the size makes the process of combining the assets relatively complicated. Moeller et al. (2004) found that there is a big difference in starting a partnership with total assets of 10 thd Euros and partnership that are worth more than 10 bln Euros. The benefits of the latter can be higher, while the complexity and deal expenses also increase.
The fourth hypothesis (number of partners) has not been confirmed due to its statistical insignificance in the model. Nonetheless, the non-confirmation of this hypothesis is still in line with the study of Beamish and Kachra (2004). Their findings indicated that the relationship between the number of partners and the alliance performance is insignificant. They argue that all parties in a partnership bring something to its performance. Control systems that are implemented in partnerships tend to decrease opportunistic behaviour and improve the performance.
The fifth hypothesis about the duration of deals has been confirmed. It should be noted that a range of assumptions have been made under this hypothesis. A lack of necessary information was the primary reason for these assumptions, which will be described further in the limitations section. The confirmation of the hypothesis reflects the findings of several studies. Lowen and Pope (2008), for example, discovered that longer negotiations between partners slow down the process of partnership creation. Furthermore, Dikova et al. (2010) found that mergers and acquisitions with significantly longer duration have lower chances of being completed in the future.
There are four significant coefficients constructed for the research model (the number of partners coefficient proved to be insignificant). Out of five tested hypotheses three were confirmed, namely: crisis (Hypothesis 2), deal size (Hypothesis 3) and duration of deal (Hypothesis 5). Two hypotheses were not confirmed: cross-border deals (Hypothesis 1) and number of partners (Hypothesis 4). Hypothesis 1 has not been confirmed because of the sign that appeared in the outcome of the test. This sign indicated the opposite correlation between the formation and international deals. Hypothesis 4 has not been confirmed due to its statistical insignificance.
6 Limitations and directions for future research
The analysis has shown that three out of five suggested hypotheses were confirmed. This can be deemed appropriate results, since the analogous studies in the sphere have relatively similar findings. Though the analysis has revealed several interesting results, acknowledgements of several limitations have to be made. They will be described along with the implications for future research in the following section.
6.1 Limitations
One potential limitation of the study is the dataset size, since only deals with announced deal size were taken into consideration. The total number of deals in the energy sector in the presented period is 4067, while the number of deals with available deal size is 805. This sample reflects nearly 20 per cent of the whole dataset. Compared to other studies this is presumed to be sufficient for the research, due to the fact that the number of observations included is relatively high.
Our second shortcoming may arise from the assumption made about the duration of deals. Only about 30 per cent of the deals included in the final sample contained information about duration. Most of the deals with announced duration were successfully completed deals. Companies tend to refrain from disclosing the information about uncompleted deals. To substitute missing numbers, the average duration of available successful deals was implemented, namely: 150 days for domestic and 273 days for cross-border deals. For the uncompleted deals we adopted the method used in previous studies that assumed the duration of uncompleted deals to be approximately one year or 365 days.
The third limitation relates to the fact that the dataset was calculated in Euros, while approximately 60 per cent of all deals were processed in currencies different than the Euro. It is difficult to say whether companies paid in Euros or whether they used a different currency. Moreover, it is impossible to detect which exchange rates were used in order to convert the numbers into Euros. An ideal situation would be to gather information about deals with home currencies and to add cross-exchange rates in the model. This limitation could therefore be seen as one of the opportunities for future research.
adjustments to their model. First, we united two regions, Asia and Oceania, due to the fact that Oceania had only 7 deals in the sample, which is less than 1 per cent of the total number of deals. Second, we assumed that Central America belonged to the North American region. This is done because the largest country of the region, Mexico, is a member of NAFTA. Third, Algeria was put into the African region, while Egypt was put into the Middle Eastern region. It should be noted that both countries are on the African continent. Finally, special attention should be devoted to the Eastern European region, as we decided to separate it from Western Europe. The main reason for this is the presence of Russia, the largest player in the market. Other countries included in this region represent Central and Eastern Europe, namely: new members of the EU (including Baltic states), former Yugoslavia countries and CIS countries (Appendix C).
6.2 Directions for the future research
The major difficulty faced by the authors in this research was a lack of some essential information. To fill this gap and complete the model several assumptions were made. Indeed, additional data on deals parameters would strengthen the research significantly. Thus, for the research improvement we suggest three steps.
The first step includes the improvement of research by conduction of a questionnaire. The questionnaire can be seen as a useful instrument in the process of gathering necessary data. The conduction of the questionnaire consists of four phases, namely: questionnaire preparation and design; training and fieldwork; data processing; final report, data preparation and dissemination. The questionnaire will give a chance to obtain insider information that is crucial for continuing the research. It should be sent to the partnership managers, aiming to obtaining missing information in different databases. This information may include the exact duration of a deal, the currency of a deal and geographical belonging of partners. The approximate response rate from this type of questionnaires is presumed to be around 20-30 per cent (Sykes, 1994). This means that in order to obtain at least 500 deals, the database should be increased to 2000 deals. That is possible to do, even when not extending time frames or industry borders. Indeed, the total preliminary number of deals obtained for this research comprised more than 4000 deals. More than 80 per cent of the sample was cut due to the lack of data.
