Joint Venture, an Alternative for Knowledge Learning

Abstract

Joint venture as a form of business cooperation is a valid alternative for acquiring external knowledge and particularly so when it is unavailable on the market. Joint venture constitutes an interesting instrument for learning in firms that belong to high technology sectors; it not only enables access to learning but can also help its assimilation and even its subsequent application to new uses. To analyze this learning, we focus on the elements that make it up, the nature of transferred knowledge, communication between partners, the motivation of the partner to learn, and lastly, the context in which this occurs, such as the type of organizational structure of the partner firm. We obtained a sample composed of 74 firms that have been involved in at least one joint venture and that recognize the fact that they have obtained knowledge from their partner. We applied structural equations via the EQS 6.1 program to contrast our hypotheses. Among the more notable results is the fact that motivation on the part of the learning partner plays a relevant role in this process and that this motivation depends, to a large extent, on the complexity of the knowledge to be transferred. Copyright © 2012 John Wiley & Sons, Ltd.

INTRODUCTION

Organizations cannot find all the resources in the market; they need to be prepared for changes that may occur. When imperfections exist in the market that hinder or impede the acquisition of certain knowledge, organizations must consider other alternatives for securing these types of resources, such as internal development, mergers and the acquisition of all or part of a business, or cooperation with other firms or institutions (Menguzzato and Renau, 1991).

The research carried out by Tidd and Trewhella (1997) reveals that establishing alliances or cooperation has been the most favorable strategy for acquiring very specialized knowledge, whereas establishing strategic cooperation is one of the most effective alternatives for generating new knowledge (Teng and Das, 2008). In knowledge-intensive sectors, cooperation or alliances can be attractive for projects that are too expensive or that require technologies that are beyond the reach of a particular firm (Zysman, 1993; Hennart and Reddy, 1997). They have an enormous potential for generating mutual benefits for partners, as they can reduce the amount of money needed for investment in research and enable the transfer of knowledge that is either scarcely codified or tacit (Park et al., 2004). Moreover, current knowledge needs can be different to future ones, which is why acquiring a business can be a sword of Damocles for the acquiring firm (Badaracco, 1992).

An important advantage of cooperation for acquiring knowledge is its flexibility in adapting to changes that occur in the environment. In general, cooperation between firms is suitable for environments with a high degree of uncertainty (Menguzzato and Renau, 1994) because of the fact that they are suitable when knowledge is widespread, as they provide greater strategic flexibility and make knowledge learning more agile (Hoffmann and Schaper-Rinkel, 2001).

However, the use of alliances or cooperation agreements can be differentiated to increase access to knowledge of other firms and of alliances for acquiring new knowledge. Certain types of alliance are more suitable for simply accessing knowledge (Grant and Baden-Fuller, 2004).

In this study, we analyze cooperation in general and, in particular, joint ventures (JVs) as a quick way of acquiring knowledge through knowledge transfer from one firm to another, enabling “close contact” with the partner (Badaracco, 1992). We therefore focus on the study of JVs,* as partners in such cases have a closer relationship that enables learning. It should be noted that in this form of cooperation, the degree of commitment of partners is much higher than in other forms of cooperation in such a way that both problems arising from poor management and weaknesses in the firm's operational side have more serious repercussions. However, the chances of success in learning are normally higher in JVs than in other forms because of the greater degree of commitment involved (Menguzzato, 1992). It can equally be considered a “mutual hostage” situation (Kogut, 1988) that reduces opportunistic behavior among partners.

Joint venture is preferred over other forms of cooperation in terms of knowledge transfer when (Sampson, 2004)

1. knowledge is tacit and complex and should be displayed and transferred between partners,
2. the use of knowledge that is complementary to this tacit, complex knowledge must be used in the way foreseen by the partner that possesses it, and
3. the set of knowledge possessed by each of the partners is different, to the extent that partners find it difficult to incorporate.

Joint ventures offer an opportunity for generating the partner's competences and skills, as this type of cooperation represents an effective vehicle for transmitting knowledge between partners, although it also increases the likelihood of the potential danger that transforms a partner into a new competitor (Lei et al., 1997); cooperation can be a double-edged sword, and it is therefore essential to know how to wield it (Menguzzato, 1995).

THEORETICAL FRAMEWORK

Knowledge learning between partners in a joint venture

As we have already mentioned, learning as a cooperation objective (Hamel, 1991) is the definitive acquisition and internalization of knowledge for its use, not only within the framework of that cooperation but also of all activities outside it (Menguzzato, 1995). An organization's capacity to learn knowledge is based on the capacity to acquire and develop knowledge to explore it in a first stage and then go on to exploit it. Prior to beginning the learning process, it is necessary for the partners to possess a sufficient common knowledge base to subsequently look for applications of the knowledge acquired by means of cooperation (Lane et al., 2001), besides having a clear, strategic vision of its current capabilities in its respective organizations and of others it may need in the future. Learning processes develop dynamic capabilities in organizations (Collis and Montgomery, 2008).

Some studies, such as those of Lane et al. (2001) or Kale et al. (2002), suggest that an important way in which organizations can capture, integrate, and disseminate external knowledge through cooperation is via the creation of a separate organizational unit charged with the responsibility of capturing previous experience. Equally, there should be a belief that, via cooperation with the partner, the firm will have sufficient absorption capacity to assimilate the new knowledge (Cummings and Teng, 2003). These learning processes are involved in the development of the dynamic capabilities of organizations and directly influence the firm's operative capacity (Wang and Ahmed, 2007). It is not only important to be in contact with knowledge, but equally important is the partner's capacity to interpret it (explore it) and exploit it internally, a determining factor for its optimum use (García Canal, 1996). In short, the capacity to learn manifests itself in the firm's skill in recognizing the value of new knowledge.

To analyze knowledge learning, we focus on knowledge transfer through JV, following in the footsteps of the study by Simonin (1999), which examines the nature of the knowledge transferred, that is, it focuses on its dimensions, its tacit component, its specificity, and its complexity. The study by Cummings and Teng (2003) also provides relevant research as it classifies variables according to the source, recipient, and context of knowledge transfer between partners. We thus propose the study of possible determining factors of knowledge learning according to the following:

1. the characteristics of the source: communication
2. the characteristics of the recipient: motivation
3. the characteristics of the context: the structure of the partner firm
4. the characteristics of the knowledge: specificity, tacitness, and complexity

Within this framework, we have introduced the variables that we have found to be more relevant in transferring knowledge, summarized in Table 1. The knowledge acquired by a partner via transfer through a JV should be assimilated in such a way that a minimum level of learning occurs and thus, its subsequent application to other processes, uses, or products depends upon the absorption capacity of the acquiring partner.

Table 1. Variables that influence knowledge learning

Authors	Factors	Structure of knowledge transfer
Source: author's own work
Menguzzato (1992)	Communication	Characteristics of the partner source
Levinson and Asahi (1995)
Doz (1996)
Yoshino and Rangan (1995)
Doz (1996)	Motivation	Characteristics of the partner recipient
Lei et al. (1997)
Cummings and Teng (2003)
Grant (1996)	Structure	Characteristics of the context
Martínez and Ruiz (2003)
Cohen and Levinthal (1990)	Nature of the knowledge: specificity, tacitness, and complexity	Characteristics of knowledge
Levinson and Asahi (1995)
Menguzzato (1995)
Doz (1996)
Powel et al. (1996)
Mowery et al. (1996)
Lei et al. (1997),
Inkpen (1998, 2000)
Lane and Lubatkin (1998)
Simonin (1999)
Cummings and Teng (2003)
Kogut (1988)	Form of cooperation
Hamel (1991)
Doz (1996)
Levinson and Asahi (1995)

Characteristics of the source: communication

According to Pisano (1988), the capacity to learn knowledge through JV, among other things, depends upon the “good will” of the source partner for the knowledge to reach the recipient. It is also necessary for partners to have the “good will” to enable the acquisition of knowledge that has already been negotiated in the cooperation agreement. In knowledge transfer, the source partner must lend the necessary support during the initial stages of using the knowledge being transferred (Szulanski, 1996). Good communication is vital between issuer and receiver for knowledge transfer to occur, and communication is made easier when partners share the same language, we can see in Table 2.

Table 2. Technical datasheet of the empirical investigation

^a In this case, the sample is made up of simple units that are accessible or favorable (Miquel et al.,1997: 146).
Universe and scope of the investigation	1210 firms
Sample size	74
Level of confidence	90% p = q = 0.5
Sample error	±9%
Sampling procedure	Convenience samplinga
Geographical scope	International
Sample unit	Firms involved in at least one joint venture
Type of interview	Structured questionnaire in web and/or Word format, according to interviewee preference
Profile interviewed	A manager from the firm involved in the joint venture

Communication is crucial not only for effective use but also for the assimilation of knowledge into the organization because, as Badaracco (1992) rightly points out, “for an organization to be in a position to acquire knowledge that is embedded in the routines of another, a complex and intimate relationship must be formed”. There must be flowing feedback between partners for knowledge transfer to occur.

It is thus important for firms to carefully select the knowledge they are to transfer to partners. They must establish what information is necessary for taking decisions that are relative to the activities affected by the cooperation agreement (Yoshino and Rangan, 1995), how frequently and to whom this information should be provided, the form it should take, how the information should be passed on to the partner, who should authorize the provision of information that is not specified in the agreement, at what level of responsibility communication between managers should take place, the relative importance of formal and informal relationships between the staff involved in the cooperation, and how frequently meetings should take place between the managers of the organizations involved (Menguzzato, 1992; Yoshino and Rangan, 1995).

The intensity or frequency and knowledge of the information and knowledge that is transmitted are important for strengthening the relationships between the source and the recipient. Intensive communication creates strong links and allows information and knowledge to be shared, thereby making knowledge transfer more effective (Ghoshal et al., 1994) and generating greater opportunity for knowledge learning.

HYPOTHESIS 1. (H.1) Communication between partners of a JV positively influences knowledge learning.

Characteristics of the recipient partner

The analysis of motivation in learning is fundamental, as simply exposing individuals, groups, or the organization to relevant knowledge is insufficient unless an effort is made to internalize it (Kim, 1998). The organization that receives the knowledge must be motivated to understand, retain, and assimilate knowledge acquired from the partner and not make do with simply accessing the partner's knowledge. Considering motivation as a determining factor for knowledge learning is justified because therein lies the true capacity for development, both in the interests of the individual or groups and those of the organization (Argyris, 1994).

Motivation is related to what Nonaka and Takeuchi (1995) call the intention or purpose of a firm in achieving its goals. A firm's motivation is understood as the aspirations of the organization in achieving its objectives, involving employees in actions and commitment in reaching goals and objectives. A lack of motivation can bring about passivity, hidden sabotage, or rejection in the application and use of this new knowledge (Szulanski, 1996). One of the ways of motivating is the appropriate introduction of a reward system (monetary or otherwise). The reward system (Lei et al., 1997) has a direct impact on behavior that can enable organizational learning to a large extent.

In knowledge transfer, at times, the “not invented here” syndrome can crop up when staff are not motivated. This syndrome entails the rejection of the use of external knowledge, as it has not been created within the firm, considering it, among other things, as a threat (Davenport and Prusak, 1998; Gupta and Govindarajan, 2000). Another reason for not accepting external knowledge is that the source of the knowledge in question is perceived as being unreliable (Zander and Kogut, 1995).

HYPOTHESIS 2. (H.2) Motivation to learn from the partner in a JV has a positive influence on knowledge learning.

Characteristics of the context of joint venture: the structure of the learner partner firm

Contextual conditions influence knowledge learning to such an extent that the advantages that can be gained from learning can disappear because of a sterile organizational context (Szulanski, 1996). In this sense, we consider organizational structure to be the context where learning is developed in the organization. According to Grant (1996), integrating strategic knowledge into the organization entails two aspects, establishing flatter structures, based upon teams where the role of employees is emphasized to better articulate tacit and complex knowledge, and the decentralization of knowledge-related decision-making.

If the partner's structure is flat, this will enable the relationship between the various grouped units, and therefore, it becomes easier to integrate knowledge into the structure. Generally speaking, the successful diffusion of knowledge in a JV requires greater decentralization on the part of the partner who is acquiring the knowledge, as long as the staff involved in the JV are skilled workers.

Organizations that want to learn knowledge must allow their skilled staff to act with the greatest degree of freedom possible, as this enables the acquisition of new knowledge to increase and, consequently, the creation of new opportunities, innovations, and new products. Autonomy drives and gives meaning to the personal commitment that must be managed by the organization (Nonaka, 1994).

Skilled staff that are involved in the JV must make more accurate decisions on their tasks by being prepared to acquire specific, skilled knowledge to make judgements and make decisions on complex problems. This knowledge can be interpreted and assimilated, adapting it to the needs of the firm and its subsequent “routinization”, going on to become the property of the learning partner.

HYPOTHESIS 3. (H.3) Decentralization in the structure of the learner partner in a JV has a positive influence on knowledge learning.

Characteristics of knowledge: specificity, “tacitness”, and complexity

At the start of a cooperation agreement, each firm possesses knowledge, “internal knowledge” or the partner's “self-knowledge”. This knowledge includes information on products, processes, and markets, as well as perceptions on cultural behavior, available within the limits of each of the organization partners.

The acquisition of knowledge on the part of the partner depends on the nature of the knowledge contributed to the JV in relation to the knowledge possessed by the partner (Inkpen, 2000), as having adequate previous knowledge of the partner can allow for the effective use of the knowledge learnt (Cohen and Levinthal, 1990; Simonin, 1999). To acquire knowledge via a JV, this knowledge should, at the very least, be accessible. Two factors limit this accessibility: protectionism on the part of the partner and the degree of tacitness inherent in the knowledge concerned (Inkpen, 1998); the more tacit knowledge is, the more difficult it is to acquire. In this section, apart from examining the degree of tacitness of the knowledge being transferred, we incorporate the study of the complexity and specificity of knowledge.

The specificity of knowledge in joint ventures

The specificity of knowledge refers to the degree to which knowledge can be put to alternative uses without losing its value. The idiosyncratic nature of knowledge constitutes a barrier for transfer and thus the chances of alternative uses in an organization other than the one that produced it. Specificity is understood as the discrepancy between the first and second use or value of an asset, which means each partner is different in terms of the knowledge they possess.

Sometimes, knowledge is not specific to the partner but to the country they are based in, which is why costs derived from knowledge transfer can be especially high. Attempts on the part of firms to exploit their innovative potential can only be fruitful if they are accompanied by direct activity in the target countries. In this sense, specificity is considered to be a source of ambiguity and a barrier to transfer and thus impedes the imitation or internalization of knowledge (Simonin, 1999), and for this reason, close cooperation between the partner that possesses the knowledge and the acquiring partner is necessary to reduce causal ambiguity.

The advantage of JVs in terms of knowledge transfer is that the partner in possession of specific knowledge helps the other to better understand the knowledge they contribute to the cooperation, and the acquiring partner improves understanding and assimilation of knowledge, thereby increasing knowledge learning (Cummings and Teng, 2003), whereas for competitors, it still constitutes acquired specific knowledge.

HYPOTHESIS 4. (H.4a) The more specific the knowledge being transferred, the greater the degree of knowledge learning via the JV.

The “tacitness” of knowledge transferred to the joint venture

If knowledge is tacit, it can be more easily understood or more precisely copied if knowledge transfer is carried out via a JV (Hennart, 1988), as it involves the collaboration of the partner. When the level of tacit knowledge is extremely high and is thus not suitable for imitation, the JV may represent one of the few paths available for learning this type of knowledge (Pisano, 1988). JVs make knowledge more accessible for the partner, although still bring inaccessible to competitors.

Hence, Simonin (1999) claims that the tacit component is a source of instability or conflict in the cooperation and is a manifestation of the difficulty and frustration inherent in knowledge learning. However, in JVs, the partner who has the tacit knowledge provide it learning.

The risk of tacit knowledge lies in the fact that the knowledge transferred to the cooperation fades or gets lost. JVs are thus an instrument for satisfactorily transferring tacit knowledge. However, it is not only considered to be a means of acquiring the tacit knowledge of partners but also of assimilating or internalizing them (Menguzzato, 1995).

HYPOTHESIS 5. (H.4b) The more tacit the knowledge being transferred, the greater the degree of knowledge learning via the JV.

The complexity of knowledge transferred via the joint venture

Two basic elements can be distinguished with regard to the complexity of knowledge: technical complexity and social complexity. The first refers to the number of resources related to the knowledge to be learnt, and the second concerns the degree of interaction between those resources. Authors such as García et al. (2008) discuss endogenous and exogenous complexity; the endogenous complexity refers to the knowledge that have been generated internally and are not coded, but are difficult for competitors to identify and imitate, whilst the exogenous complexity to refers to the difficulty involved in identifying and understanding the value and use a firm makes of the knowledge diffused in the industry. The latter is basically what we refer to with specificity.

In fast developing sectors, knowledge can be complex and widely dispersed. As Powell et al. (1996) propose, innovations in the biotechnology sector involve innumerable interrelations between scientists and biotechnology firms, pharmaceutical companies, universities, and other institutions that, via the use of cooperation agreements, raise the complexity of the knowledge used by them, and consequently, the knowledge handled by these knowledge-intensive sectors tends to be highly complex.

Empirical studies such as that carried out by Simonin (1999) highlight the tremendous importance of the complexity component together with the tacit nature of knowledge transfer, in contrast to the scant influence of specificity. Mowery et al. (1996) confirm the fact that JVs appear to be more suitable for the transfer of complex knowledge than cooperation agreements that do not result in the creation of a new entity. Complexity basically affects the understanding of the knowledge being transmitted, and it seems clear that JVs enable the acquisition and understanding of complex knowledge from other partners (Inkpen and Beamish, 1997), thereby reducing the causal ambiguity of the transfer.

HYPOTHESIS 6. (H.4c) The more complex the knowledge being transferred, the greater the degree of knowledge learning via the JV.

METHODOLOGY

To carry out this research, it was necessary to select a sample of firms that had been involved in at least one JV. We found a database that fitted our criteria, the ZEPHYR database, where we found a population of firms that had carried out a JV. In this database, we found 1837 firms across the five continents that met the criteria for this research. This database was filtered using data from other bases such as Amadeus and Thomson One Banker, all of which are available online, along with consultation of the web pages of each of the firms that make up the database. The questionnaire was drawn up in three different languages, Spanish, English, and French, and was sent via email and/or via postal mail. In the case of Spanish firms, contact was also made via telephone to make contact with a high ranking manager involved in the JV.

Out of the total number of firms in the database, we were not able to contact 231 firms because of incorrect contact information or because the firm no longer existed, and a total of 396 firms declared that they did not fit the profile of a JV as it was defined in the covering letter sent along with the questionnaire, despite forming a part of the database consulted. The total number of firms whom we contacted in reality came to 1210. Finally, the sample obtained is 74† firms (51 Spanish firms and 23 from the rest of the world). Table 1 shows the technical datasheet with the details of the sample.

Measurement instruments: creation and validation of the measurement scales

Empirical studies of a quantitative nature are extremely scarce, but we cite those that are most representative such as those by Lane and Lubatkin (1998), Simonin (1999), Lane et al. (2001), Martínez and Briones (2004), among others. We agree with the view expressed by Montes et al. (2002) that, generally, studies are characterized by an insufficient psychometric evaluation of scales. As a result, we opted to use our own scales, which allowed us to specify the domain and dimensionality of each construct. In this respect, Chin and Marcolin (1995) claim that the psychometric properties of the scales must be verified within the model itself that is to be contrasted, as the reliability and validity of the constructs can change according to the theoretical model they are applied to.

Analysis of the reliability of the measurement instrument

Generally speaking, in the field of business management, not only one but several constructs are involved. The Cronbach alpha for each factor calculated separately does not take into account the influence on reliability of the other constructs. Consequently, Fornell and Larcker (1981) suggest the calculation of the composite reliability index (CRI) for each factor, which is interpreted in exactly the same way as the Cronbach alpha. The same authors also propose the calculation of the variance extracted index (VEI).

The calculation of the Cronbach alpha for the “knowledge learning” scale has an acceptable value of 0.860. We firstly check that none of the items should be eliminated to improve this indicator, as can be seen in Table 3, where we summarize the results obtained for all the factors once the items that did not surpass a value of 0.7 have been removed.

Table 3. Cronbach alpha factor indices

Factors or scales	N of items	Cronbach alpha	Items
Knowledge learning	11	0.860	V22–V32
Nature of knowledge	7	0.761	V01–V07
Partner motivation	5	0.765	V08–V12
Communication between partners	5	0.890	V13–V17
Organizational structure	4	0.779	V18–V21

Authors such as Fornell and Larcker (1981) claim that composite reliability is a superior measurement to that of the Cronbach alpha, in so far, as the latter does not take into account the influence on reliability of the other constructs, and moreover, composite reliability has the advantage of not being influenced by the number of existing items in a scale (Cepeda and Roldan, 2004). To obtain the necessary data, both for the calculation of the CRIs and for (convergent) validity, it is firstly necessary to do the confirmatory factor analysis (CFA), which was estimated using the EQS 6.1 program (Software multivarado, Inc. Encino, California, USA). The CFA was estimated using the maximum likelihood procedure, verifying the goodness of fit and interpreting the model. This analysis shows the need to eliminate certain items to achieve a good fit. The remaining items are used to calculate the composite reliability.

In Table 4, we summarize the reliability values of the different scales that go to make up the measurement instrument we have created for this investigation.

Table 4. Composite reliability of the measurement instrument

Factors or scales	N of items	Cronbach alpha	N of items	CRI	Significant items
CRI, composite reliability index.
Knowledge learning	11	0.860	6	0.837	V22, V26, V27, V28, V29, V30
Nature of knowledge	7	0.761	3	0.920	V4, V5, V6
Partner motivation	5	0.765	2	0.652	V08, V10
Communication between partners	5	0.890	2	0.954	V13, V14
Organizational structure	4	0.779	1	1.000	V18

As can be seen in Table 4, the CRIs of the factors are higher than the cited threshold, and thus, we opted to maintain this scale. Composite reliability is also carried out using the measurement developed by Fornell and Larcker (1981) called the average variance extracted index (AVE).‡ The values obtained must be greater than the threshold of 0.5 suggested by Bagozzi (1981), which indicates that more than 50% of the variance of the construct is because of its indicators.

Generally, the results of the AVEs shown in Table 5 are reasonably satisfactory, except for that corresponding to knowledge learning that has a value of 0.472 but is still very close to 0.5, which is why we maintain all the scales.

Table 5. Comparison between the CRI and the AVE

Factors or scales	N of items	Cronbach alpha	N of items	CRI	AVE
CRI, composite reliability index; AVE, average variance extracted index.
Knowledge learning	8	0.860	5	0.837	0.472
Nature of knowledge	7	0.761	3	0.920	0.798
Partner motivation	5	0.765	2	0.652	0.546
Communication between partners	5	0.890	2	0.954	0.912
Organizational structure	4	0.779	1	1.000	1.000

Validity of the measurement instrument

The validity of a polyhedric concept has diverse dimensions that must be explained and analyzed separately, such as the validity of content, validity of concept or construction (convergent and discriminatory), and the validity of criteria (Vila et al., 2000).

In our case, with regard to the validity of content, we have reviewed several theoretical and empirical studies and, above all, the latter to attempt to reveal the dimensions of each scale. We have opted to summarize them in Table 6.

Table 6. Research related to the construction of scales

Scales	Studies that have been fundamental in creating the items in the construction of our scales
Knowledge learning	Cohen and Levinthal (1990), Levinson and Asahi (1995), Menguzzato (1995), Lane and Lubatkin (1998), Kale et al. (2000), Galende (2001), Montes et al. (2002), Bontis, Crosaan, and Hulland (2002), Martínez and Ruiz (2003), Tippins and Sohi (2003), Cummings and Teng (2003)
Nature of knowledge	Szulanski (1996), Simonin (1999), Lane et al. (2001)
Motivation	Szulanski (1996)
Communication	Menguzzato (1992), Mohr and Spekman (1994), Levinson and Asahi (1995), Montoro (2000), Montes et al. (2002), Koka and Prescott (2002), Martínez and Ruiz (2003), Martínez and Briones (2004)
Structure	Nonaka and Takeuchi (1995), Peris and Herrera (1998), Lane and Lubatkin (1998), Cruz and Camps (2003)

A scale has validity of construction when it has convergent validity and discriminant validity (Vila et al., 2000). Convergent validity is determined by reviewing the t statistics of the factor loadings. If all the factor loadings that measure the same construct are statistically significant, the convergent validity of these indicators is verified (Anderson and Gerbing, 1988). Apart from being significant, Hair, Anderson, Tatham and Black (1999) recommend that the average loadings for each factor should be greater than 0.7 and that the minimum level of factor loadings should be 0.4. In our case, to carry out the convergent validity analysis, we performed the CFA.

In the knowledge learning model, we show the results of the CFA in Table 7. As can be observed in the table, these results display extremely sound estimations with a high level of significativity (as all the t statistics are greater than 3.291 and are thus significant for p < 0.001) and standardized λ with high values, all of which are above 0.4, with the exception of one, with value of 0.304.

Table 7. Confirmatory factor analysis of model 2 for knowledge learning

Variable		t	Standardized ʎ	Level of fit indices
AGFI, adjusted goodness of fit index; GFI, goodness of fit index; RMR, Root Mean Square Residual. Levels of significativity: * p < 0.5. ** p < 0.1. *** p < 0.001; (based on two-tailed t tests (499) ). t(0.05, 499) = 1.964. t(0.01, 499) = 2.585. t(0.001, 499) = 3.291.
Knowledge learning (F5)
V22 F5	2.340***	6.780	0.645	χ^2(9 degrees of freedom) = 14.556
V26 F5	2.054***	6.265	0.606	Bentler–Bonett normed fit index = 0.936
V27 F5	2.996***	9.704	0.840	Bentler–Bonett non-normed fit index = 0.957
V28 F5	3.058***	10.513	0.886	Comparative fit index = 0.974
V29 F5	1.558***	5.056	0.506	Lisrel GFI fit index = 0.951
V31 F5	1.931***	5.614	0.553	Lisrel AGFI fit index = 0.885
				StandardizedRMR = 0.050
Nature of knowledge (F1), motivation (F2), communication (F3), and structure (F4)
V4 F1	2.020***	7.737	0.687	χ^2(21 degrees of freedom) = 26.167
V5 F1	3.078***	13.928	1.000	Bentler–Bonett normed fit index = 0.908
V6 F1	2.212***	7.799	0.649	Bentler–Bonett non-normed fit index = 0.963
V8 F2	3.451***	13.928	1.000	Comparative fit index = 0.979
V10 F2	1.270***	3.420	0.304	Lisrel GFI fit index = 0.948
V13 F3	3.968***	11.655	0.937	Lisrel AGFI fit index = 0.888
V14 F3	4.198***	12.402	0.730	Standardized RMR = 0.064
V18 F4	3.653***	13.928	1.000

On another note, the statistics for the goodness of fit show a value of around 0.9. For knowledge learning, the indicators of the goodness of fit generally reach the desired values, except for the adjusted goodness of fit index, with a value of 0.885, which is still close to 0.9. For the variables of the nature of knowledge and motivation, there is a good fit, except again for adjusted goodness of fit index, which is very close to 0.9 but with a value of just under 0.888, and standardized root mean square residual, which slightly exceeds the limit of 0.05 at 0.064. Lastly, the third level, where we have included the structure and the environment, values do not quite reach the desired levels but are reasonable.

In short, we can state that the measurement model also has an acceptable fit. We can thus state that the convergent validity of this model is confirmed.

We analyze the discriminant validity by comparing the correlations of the factors and the Cronbach alpha in each case. If the latter is greater, discriminant validity is ensured (Sánchez and Sarabia, 1999).

Table 8 shows the comparison matrix between the correlations and coefficient values of the Cronbach alpha.

Table 8. Correlation matrix and Cronbach alpha

	F1	F2	F3	F4	F5
VEI, variance extracted index. Levels of significativity: * p < 0.5. ** p < 0.1. *** p < 0.001; (based on two-tailed t tests (499) ). t(0.05, 499) = 1.964. t(0.01, 499) = 2.585. t(0.001, 499) = 3.291.
F1	1
F2	0.362	1
F3	0.208	0.468	1
F4	0.050	0.128	0.223	1
F5	0.427**	0.690**	0.578**	0.154	1
Alpha	0.764	0.765	0.890	0.779	0.860
VEI	0.798	0.546	0.912	1.000	0.472

The highest correlation in this matrix is 0.690, which corresponds to the relationship between F2 and F5. If we find the square of the correlation, it gives us a value of 0.470, which is lower than the VEI of F2 (0.546) and the VEI of F5 (0.472). These results therefore confirm the discriminant validity of the measurement instrument we are using.

It should be noted that the data used for the analysis of the measurement model have been statistically treated using the spss 12.0 and the EQS 6.1 programs. Certain indices have been calculated using the formulae we developed with the Microsoft Excel 2002 program.

In the structural model, we address the analysis of causal relationships, determined by the formulation of the hypotheses we have created in our theoretical framework. To do this, we use structural equation models.

The structural model

In this section, we go on to analyze the relations between the variables established for this model of knowledge learning through JVs, as we established in the hypotheses. For greater clarity in the proposed relationships, we have included Figure 1 as a graphic representation of these hypotheses.

Figure 1. Theoretical model

We introduce the proposed relationships into these hypotheses in the structural analysis and obtain indicators that are shown in Table 9.

Table 9. Goodness of fit indices of the “theoretical model”

	χ^2	gl	p	GFI	AGFI	SRMR
gl, degrees of freedom; GFI, goodness of fit index; AGFI, adjusted goodness of fit index; SRMR, standardized root mean square residual.
Theoretical model	12.357	2	0.002	0.967	0.537	0.083
GFI: close to 0.9
AGFI: close to 0.9
SRMR: below 0.05

The goodness of fit indicators of our “theoretical model” does not reach the desired values, and thus, we proceed with the analysis of the model. By reviewing the contrast of the Lagrange multiplier, we see the existence of relationships that have a theoretical justification and proceed with the restructuring. We introduce these new relationships in the structural part and observe that there is an improvement in the model, as shown in Table 10.

Table 10. Comparison of the goodness of fit indices of both models

	χ^2	gl	p	GFI	AGFI	SRMR
gl, degrees of freedom; GFI, goodness of fit index; AGFI, adjusted goodness of fit index; SRMR, standardized root mean square residual.
Theoretical model	12.357	2	0.002	0.967	0.537	0.083
Revised model	13.501	9	0.141	0.965	0.890	0.067
GFI: close to 0.9
AGFI: close to 0.9
SRMR: below 0.05

Figure 2 shows a graphic representation of the improvements introduced into the “revised model”, as an illustration of our proposal.

Figure 2. Revised model

Once we have established the “revised model”, we can contrast the proposed hypotheses for this second model.

RESULTS

In the second table, we provide a summary of the results of the contrast of our hypotheses, which we go on to comment upon.

As we can see in Table 11, H1 is significant. Communication is important for knowledge learning because, if there is no communication between partners, it is hard for knowledge to be transferred. Moreover, H2 is one of the more significant relationships found in our research as it has a value of 7.532, much higher than 1.96, and we can thus confirm this hypothesis in no uncertain terms. The firm that acquires the knowledge must have a proactive attitude towards its acquisition and assimilation if it wants to internalize it and subsequently apply it to other uses once it has been acquired through the JV.

The firm and its staff must be motivated to learn via a reward system (economic and noneconomic), which have a dual direct impact. On the one hand, economically, the managers of partners related to a JV that have rewards based on the hierarchy are well placed for learning new knowledge and behavior, thereby creating a greater relationship between the JV and these managers. On the other hand, the result would be that the managers involved in the JV, apart from being skilled staff, have a positive attitude and a resolute approach to problem solving. These are the people that allow knowledge learning to occur in their firms.

Table 11. Estimated parameters for the variables

Hypothesis	Influence	Standardized loadings	t
* p < 0.05; t > 1.964. ** p < 0.01; t > 2.585. *** p < 0.001; t > 3.291.
H.1. Communication between partners in a joint venture has a positive influence on knowledge learning	Communication in knowledge learning	0.452***	5.493
H.2. Motivation to learn from the partner in a joint venture has a positive influence on knowledge learning	Motivation in knowledge learning	0.615***	7.532
H.3. Decentralization in the learning partner's structure in the joint venture has a positive influence on knowledge learning	Structure in knowledge learning	0.063	0.880
H.4c. The more complex the knowledge being transferred, the greater the degree of knowledge learning via the joint venture	Complexity of knowledge in knowledge learning	0.206**	2.745

In addition, some very significant relationships that arose in the Lagrange multiplier (R1) were that motivation has a positive influence on the complexity of knowledge (R1) as the results on Table 12 show.

Table 12. Estimated parameters for the new relationships with motivation

Relationships generated in the revised model	Influence	Standardized loadings	t
* p < 0.05; t > 1.964. ** p < 0.01; t > 2.585. *** p < 0.001; t > 3.291.
R.1 Motivation to learn from the partner in a joint venture has a positive influence on knowledge learning	Motivation in the complexity of knowledge	0.362***	3.825

The R1 relationship is highly significant (t = 3.825) and go to further highlight the results previously described; the more motivated the partner firm is, the more complex knowledge is assimilated, that is, greater motivation is necessary on the part of managers of partner firm related via JV to learn complex knowledge contributed by the partner to the cooperation, we can see in (Table 13).

Table 13. Summary of the contrast of the hypotheses of model 2

Hypotheses	Results
H.1. Communication between partners of a joint venture positively influences knowledge learning	Confirmed
H.2. Motivation to learn from the partner in a joint venture has a positive influence on knowledge learning	Confirmed plus R1
H.3. Decentralization in the learning partner's structure in the joint venture has a positive influence on knowledge learning	Rejected
H.4c. The more complex the knowledge being transferred, the greater the degree of knowledge learning via the joint venture	Confirmed

When it is a question of learning something new, it is essential to see the usefulness and end result of the knowledge (Bandura, 1977) and much more so if the knowledge is complex.

We cannot confirm H3, as the t statistic is not significant as we can see in Table 3. The existence of greater decentralization in the learner partner firm does not enable the acquisition of greater knowledge.

When analyzing the possible reasons for not being able to confirm this hypothesis, other authors have come up with similar results to ours. For example, Moreno-Luzón and Peris (1999) cannot confirm the hypothesis that suggests a relationship between decentralization and knowledge learning. In the study by Lloria (2003) on organizational design and knowledge creation, the hypothesis that proposes a relationship between decentralization and knowledge creation is finally rejected.

In an attempt to understand this result, we propose that the hierarchical position of people involved in the JV is not so vital to knowledge learning but rather that these people need to be skilled and committed to the organization and have the desire to better themselves. These people are usually managers.

To integrate complex, dependent, and sophisticated knowledge, the existence of skilled staff is required to establish organizational procedures, routines, complementary capabilities, and interpersonal relationships that may be necessary (Cohen and Levinthal, 1990).

It should be borne in mine that over half the firms that make up our sample are large firms with a rather bureaucratized structure, far from what is understood as an organic structure, and thus, decision-making is, to a large extent, centralized. However, the fact that it may be mechanical or bureaucratic is not incompatible with knowledge learning as a specific “workforce” or team that can be formed to strengthen coordination in the firm as Nonaka and Takeuchi (1995) propose in the hypertext structure by eliminating the drawbacks of this type of structure for knowledge learning through JV.

From the dimensions of the nature of knowledge, we have only been able to validate that which corresponds to the complexity of knowledge and thus can only contrast hypothesis 4c (H4c). Having said this, the relationship between the complexity of knowledge and knowledge learning is significant, t = 2.745, and is well over 1.96, as shown in Table 3.

On analyzing the items that measure the nature of knowledge, we can see that the two aspects we had previously foreseen for complexity were validated, both technically and socially although we have not been able to validate the specificity and degree of codification. This leads us to believe that both the specificity and the difficulty in codification in the case of JVs are compensated by the desire on the part of the source partner to provide the knowledge laid down in the agreement. Despite the fact that the knowledge to be transferred to the JV may be tacit, the source partner enables its transfer because of the commitment to the task, thereby minimizing the barrier to “tacitness”. When contact among partners is direct and does not entail the creation of a new entity, the importance of “noncodification” or “tacitness” of knowledge is greater than in the case of a JV.

The relationship established in hypothesis H4c means that the complexity of knowledge, both technically and socially speaking, has a good deal of influence on knowledge learning. With regard to the technical side, this is because of the relationship between this and other knowledge possessed by the partner and socially to the dependence of knowledge of the system where it is located. Both aspects of the complexity of knowledge make it difficult to acquire and assimilate without the cooperation of the source partner, but the JV is an instrument that makes it possible or, at least, easier by committing the partner to the transfer of certain knowledge. These results are consistent with those of Cohen and Levinthal (1990) who state that the characteristics that affect how easy learning is basically boil down to just two: the complexity of assimilated knowledge and the degree to which external knowledge is suited to the needs of the firm. Therefore, the acquisition of tacit and complex knowledge through a JV is important, as they can lead to the generation of ideas that pave the way for radical innovation (Castiaux, 2007).

In our case, we must bear in mind that when we talk about the nature of knowledge, we are referring to the dimension of complexity of knowledge, both in technical terms and social ones.

CONCLUSIONS

The results of the research carried out in the present study lead us to draw a set of conclusions in reference both to the theoretical framework and the empirical part of our study.

The enormous impact of partner motivation in this learning should be underlined in light of its high level of significance. It is necessary for the recipient partner to be willing to learn the specific knowledge proposed and even more so if the partner helps in the learning process, as if the opposite occurs, all efforts will have been in vain. If motivated, the recipient partner will acquire the knowledge and assimilate it then go on to internalize it into the routines of the organization. A second question concerns the use the partner makes of that knowledge and how the firm exploits it. For the recipient partner to acquire knowledge, there must be a minimum active, positive attitude towards the acquisition. In particular, motivation has to be greater when the knowledge being acquired is more complex (R1).

Parting from the idea that strategic knowledge is specific, complex, and largely noncodified, we have considered complexity and the degree of codification of knowledge as key factors for the interorganizational transfer of specific knowledge of a JV, with a slightly different slant. The fact that knowledge is not codified, that is, it is tacit, can be overcome via observation or via close contact with the partner, and thus, the complexity of knowledge appears as the key dimension for its adequate transfer. Hence, the capacity of the receiver partner to acquire specific knowledge will depend on the very characteristics of the knowledge and especially its complexity and the extent to which the contributing partner helps in easing, understanding and assimilation.

The complexity of knowledge, both in its technical and social aspects, has a positive influence on knowledge learning (H4c); the greater the complexity, the greater the amount of knowledge we learn, although this process is by no means an easy one. Another factor is the source partner's readiness and willingness to facilitate or make the effort to allow the knowledge transmitted to the partner to be adequately understood. Communication is a variable that can be considered significant for knowledge learning (H1). Indeed, if knowledge is complex, the receiver partner needs help on the part of the source partner to be able to understand and correctly assimilate the knowledge they are attempting to learn.

Lastly, from the dimensions proposed in the organizational structure of the learning partner, we have only been able to validate the degree of decentralization. This variable does not appear as a determining factor for knowledge learning (H3).The nonconfirmation that a decentralized organizational structure enables knowledge learning is worthy of attention, as many arguments exist in the theory in favor of this idea. In a similar vein to certain previous quantitative studies, we have not been able to confirm this hypothesis. We believe that knowledge learning from a partner is independent of the type of structure that is dominant in the partner firm. However, this is only true as long as the people related to the JV have a certain degree of autonomy and function as an independent unit while the cooperation lasts.

These people are normally managers with more or less autonomy for decision-making with regard to the activities related to the JV; managers are skilled people who are capable of integrating complex knowledge acquired from their partner through the JV into their own firm. These managers act in terms used by Nonaka and Takeuchi as a specific “workforce”. Thus, if knowledge is highly complex and the source partner has facilitated this knowledge under the same conditions, the learning partner must use their full organizational capacity to be able to assimilate, transform, and apply the knowledge acquired through the JV within their organization. Such assimilation needs a greater degree of flexibility that the firm can reach through other available coordination mechanisms.

The results of this study apply some of the contributions from the academia to the business world. Indeed, we believe that determining which factors are the most important when choosing the JV as a form of cooperation in the interests of enabling learning to be of utmost importance. Thus, managers who charged with overseeing the JV can have a clearer vision of the key factors they need to strengthen such as motivation, in addition to the factors they should pay more attention to so as to avoid possible problems such as the generation of mistrust or the lack of communication between partners.

One limitation of this study lies in the size of the sample. Several circumstances have led to this situation: the great difficulty in obtaining a database where JVs are identified or in contacting the right person in the firm related to the JV, a fact that is particularly pertinent in large firms. However, the size of the sample is sufficient for the application of the methodology chosen although we assume the existence of a reasonably high margin of error, and thus, the results should be interpreted with caution.

It should be pointed out that the data gathered on JVs were facilitated by just one of the partners involved in this type of cooperation, a limitation that is hard to avoid and is common in studies on JVs and cooperation in general.

The development of this line of research has provoked greater interest among our research group, and we intend to continue working towards overcoming some of the limitations described and opening up other lines of related research.

• *

  We define a JV as an agreement through which two or more independent firms decide to create a new one as a legally established entity, whose capital, the partners own, to which they devote the necessary resources to run the new business, receiving in return the profits generated by the activity of this firm (Menguzzato, 1992), under the control of the competitive strategies of the owner firms (Fernández-Sánchez, 1996).

• †

  We found empirical studies on cooperation and learning with a similar sample size to this one, published studies, for example, in the Strategic Management Journal such as those of como los de Kale et al. (2002) with 78 firms, Colombo (2003) with 67, Lane et al. (2001) with 78 JV, and Lane and Lubatkin (1998) with 69.

• ‡

  The VEI or AVE provides the amount of variance a construct obtains from its indicators in relation to the amount of variance because of measurement error p (Cepeda and Roldan, 2004).

APPENDIX

(1) strongly disagree; (2) disagree; (3) indifferent; (4) agree; (5) strongly agree

	A	B
KNOWLEDGE	1 2 3 4 5	1 2 3 4 5
4.1. Your firm had to acquire specialist equipments to use this knowledge	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
4.2. Your firm had to foment specialized training of staff to use this knowledge	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
4.3. This knowledge was very concrete, very specific and only relevant to the activity in hand	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
4.4. This knowledge was very complex	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
4.5. This knowledge was related to the partner's skilled staff	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
4.6. This knowledge was closely related to other knowledge pertaining to your partner	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
4.7. This knowledge was extremely difficult to express in a code using numbers, words, symbols…	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
MOTIVATION	1 2 3 4 5	1 2 3 4 5
7.1. Your firm spent time and resources on analyzing the knowledge shared for your partner in the joint venture (JV)	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
7.2. Your firm took an interest in the functioning of the JV at all times	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
7.3. Your firm attempted to improve when carrying out activities	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
7.4. Your firm encouraged an atmosphere of self-improvement and learning amongst its employees	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
7.5. Your firm attempted to improve your products or services	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
COMMUNICATION	1 2 3 4 5	1 2 3 4 5
9.1. Meetings served to share decisions on the development of the JV	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
9.2. Meetings served to share information on the development of the JV	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
9.3. Meetings took place essentially at the moment of choosing the partner and on negotiation of the JV	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
9.4. Meetings were held throughout the duration of the JV	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
9.5. Meetings were only held when there was a problem to solve	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
9.6. Our advice and suggestions were well received on the part of the partner	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
KNOWLEDGE LEARNING	1 2 3 4 5	1 2 3 4 5
11.1. Your firm dedicated resources to analyzing knowledge contributed by your partner to the JV with a view to adopting the best aspects	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.2. Your firm assimilated knowledge exactly in the way it was contributed by your partner via the JV	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.3. At the end of the JV, your firm was capable of using knowledge gained, without the support of the partner and in the same conditions it was used during the JV	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.4. Your firm used some of the knowledge contributed to the JV by the partner under the same conditions, substituting what was previously in place	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.5.Your firm used some of the work practices contributed by your partner during the JV, under the same conditions, substituting those previously in place	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.6. Your firm dedicated the necessary means to apply the knowledge gained to other uses	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.7. Your firm has had to invest in specialized staff and machinery to exploit the knowledge gained	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.8. Your firm applies/applied knowledge gained to the firm's own products or procedures	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.9. Once you had established the best work practices learnt from the JV, your firm applied them to other uses	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.10. At the end of the JV, your firm was capable of using the knowledge gained without the support of the partner in conditions other than those present in the JV	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.11. Your firm has gained the knowledge it expected to learn via the JV	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐
11.12. Your firm has gained knowledge it did not expect to learn prior to the JV	☐ ☐ ☐ ☐ ☐	☐ ☐ ☐ ☐ ☐