Pergamon
0969~5931(94)E0010-I
International Business Review Vol. 3, No. 3, pp. 201-217, 1994 Copyright 0 1994 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0969-5931194 $7.00 + 0.00
International Joint Ventures and Performance: an Interorganizational Perspective Andrew C. Inkpen* and Julian Birkenshaw-f *School of Business and Management, Temple University, Philadelphia, PA 19122, USA t_Western Business School, University of Western Ontario, London, Ontario, N6A 3K7, Canada Abstract - Using data collected from a sample of North American-Japanese joint ventures (JVs), this paper reports an investigation of JV performance within an interorganizational framework. A structural equation modeling approach was used in conjunction with multiple methods of construct measurement. The findings support the view that the type and nature of interfirm interactions influence the parents’ assessment of JV performance. The study found that the quality of the exchange relationship was an important factor, supporting the argument that compatibility between the partners is a crucial factor in determining the partners’ relative bargaining power and operating relationship norms. Key Words - International Joint Ventures, Interorganizational Exchange, Joint Venture Performance, North American-Japanese Joint Ventures, Partial Least Squares, Structural Equation Modeling.
Introduction Joint ventures (JVs) have become increasingly popular in recent years and for many firms, have moved to the mainstream of domestic and international corporate activity. In a JV, two or more organizations are brought together because of their complementarity and their differences. The result is a “hybrid” organization (Borys and Jemison, 1989) that from an international perspective represents a governance mode for transactions between the polar opposites of arms-length market contracts and those conducted within the multinational enterprise. JVs have achieved a reputation for being difficult to manage (Geringer and Hebert, 199 1). The primary problems in managing JVs stem from one cause: there is more than one parent (Killing, 1982, p. 121). The owners of JVs are often visible and powerful; they can and will disagree on just about anything. Given the potential for interpartner conflict and disagreement, JV performance is often less than satisfactory. However, research efforts to Andrew Inkpen is an Assistant Professor in strategic management and international business in the School of Business and Management, Temple University. His primary research interest is the management of international joint ventures and strategic alliances. Julian Birkenshaw is a doctoral candidate at the Western Business School, University of Western Ontario.
201
International Joint Ventures and Performance
202 International Business Review 373
identify specific determinants of JV performance have been limited. Using data collected from a sample of North American-Japanese JVs, this paper reports an investigation of JV performance within an interorganizational framework. Although interorganizational exchange between JV partners is at the heart of a JV’s collaborative management, there is a paucity of empirical work in this area. An implicit use of an exchange theoretic perspective can be found in the works of several researchers interested in the performance of JVs. For example, Beamish (1988) described process level factors and Chowdhury (1989) examined interfirm-organization determinants of international JV successes and failures. We found no prior studies of international JVs in which the focal issue was the examination of JV performance in the context of exchange theory and interorganizational relationships. For this study, a JV is defined as a means of performing activities in combination with one or more firms instead of autonomously. A JV occurs when two or more distinct firms (the parents) pool a portion of their resources within a separate jointly owned organization. This definition excludes other forms of cooperative agreements such as licensing, distribution and supply agreements, research and development partnerships or technical assistance and management contracts. Theoretical Background As voluntary interorganizational relationships, JVs are formed by firms to achieve a strategic objective not attainable by acting autonomously. The strategic objectives include the reduction of risk, economies of scale, access to technology or markets, and the search for legitimacy (Contractor and Lorange, 1988; Harrigan, 1985; Henna& 1988; Porter and Fuller, 1986). A key element of any interorganizational relationship is the interaction process between organizations. The interactions constitute an exchange relationship which is a formal or informal set of arrangements between organizations involving the transfer of resources or services (Seabright et al., 1992, p. 122). Thus, a social exchange approach with its emphasis on purposeful coordination (Schmidt and Kochan, 1977), interactions, and noneconomic outputs from the relationship (Dwyer et al., 1987) provides a useful input for the conceptual model underlying this research. Analyzing JVs from an exchange perspective focuses attention on the interaction and relationship dimensions of the interorganizational relationship. The interaction and exchange of information between JV partners is viewed as a necessary precondition for a successful partner relationship. An exchange approach is particularly applicable when the JV is international because as Toyne (1989) argued, the issues of central importance in international business revolve around exchange relationships and the social actors involved in the exchange. An Interorganizational Model of JVs Building on the previous discussion,
a model
of the JV relationship
is
203 International Joint Ventures and Performance
Fig. 1. Conceptual Model
introduced (Fig. 1). The communication between the partners is a key element in a JV and one of the constructs that underlies the process of working together within a partnership. In a truly cooperative relationship, extensive communication between the partners is an essential feature of the relationship (Anderson and Narus, 1990; Hall et al., 1977). The nature of the communication may range from operational information exchanges necessary to run the JV to the sharing of more strategic information (Kanter, 1989). An important aspect of communication is the quantity and extent of communication between the partners at a managerial level. For example, do both partners contribute to the JV’s decision making process? We predict that the partner communication will influence the relationship openness. Openness reflects the willingness and ability of the JV partners to share information, and in particular, information or knowledge embodied in organizational skills and routines (Hamel, 1991). In an open venture, firms are able to penetrate the social context surrounding their partners. The extent of managerial communication should influence openness because partner firms that are regularly interacting at an operational level may lose their wariness about revealing information to a partner. Thus, a positive path from communication to openness is predicted. For various reasons, a firm may wish to restrict its flow of knowledge to a JV partner, including a desire to maintain its influence over the other partner(s) (Hamel et al., 1989). While openness between JV partners may be a function of cultural and language differences (Kanter, 1989), partner sharing of information is more likely to be based on the relative competitive positions of the partners (Hamel, 1991). Firms will likely be reluctant to share information with a JV partner if the partner is perceived as a potential competitor. The willingness to share information in dyadic relationships has been studied from several perspectives. Gupta (1987) examined openness in the relationship between corporate and business unit managers. Openness was defined as “the degree to which relations between business unit managers and their corporate superiors are open and informal and allow for spontaneous and open exchange of information and ideas” (Gupta, 1987, p. 479). John and Reve (1982) incorporated the concept of openness in a study of dyadic
204 International Business Review 393
relationships in marketing channels. Openness was one of the items used to measure norms of exchange. We posit a positive path from openness to trust and also from communication to trust. Trust reflects the belief that a partner’s word or promise is reliable and that a partner will fulfil its obligations in the JV relationship. The need for trust between JV partners has been identified as an important element of long-term JV relationships (Beamish, 1988; Harrigan, 1986; Sullivan and Peterson, 1982). In a JV, trust reflects the belief that a partner’s word or promise is reliable and that a partner will fulfil its obligations in the relationship. Breakdowns in the value creation process in JVs often stem from problems in managing interdependencies; a lack of trust between partners is one such problem (Borys and Jemison, 1989). As a current state, trust is dependent on previous partner interactions; hence, the path from communication to trust. However, an atmosphere of trust should also contribute to future exchanges of information between committed exchange partners since the decision-makers do not feel that they have to protect themselves from the others’ opportunistic behavior (Blau, 1964; Jarillo, 1988). Without trust, information exchanged may be low in accuracy, comprehensiveness, and timeliness (Zand, 1972). Considering the consequences of trust, we predict a positive relationship between trust and cooperation (Anderson and Narus, 1990). Cooperation is defined as coordinated actions taken by firms in interdependent relationships to achieve mutual or singular outcomes (Anderson and Narus, 1990). Trust helps to establish expectations of reciprocity or mutuality that is then reflected in cooperative behavior within a JV (Buckley and Casson, 1988). Thus, trust should influence the extent of partner cooperation. Openness is also predicted to influence cooperation because openness, like trust, is an expectation of future behavior that is grounded in the previous interactions between the partners. A relationship that has been open should contribute to an expectation of cooperative behavior between the partners. We predict that cooperation and JV integration will be positively related. The integration construct reflects the strategic relationship between a firm and its JV with an emphasis on the parent’s role in the decision-making processes of the JV. The more critical the JV-parent linkage is to the strategy of the parent organization, the greater the integration (Harrigan and Newman, 1990; Shortell and Zajac, 1988). This construct is similar to project commitment, used by Beamish (1988) in a study of JVs in developing countries. Beamish’s construct dealt with a parent firm’s readiness and willingness to provide resources and capabilities to the JV. If a JV is strategically integrated with its parent, the parent should be prepared to contribute resources to the JV. Since the degree of integration is a reflection of the parent’s commitment to its JV, it represents an outcome of the JV interactions. If there is an expectation of uncooperative partner behavior, the parent may be reluctant to contribute further resources. Thus cooperation should influence integration. We posit a positive path from cooperation to satisfaction, the focal
205 consequence of JV relationships
in our model. Satisfaction has been defined as “a positive affective state resulting from the appraisal of all aspects of firm’s working relationship with another firm” (Anderson and Narus, 1990, p. 66). Satisfaction in the JV context is a subjective assessment of JV performance; it reflects the parent’s satisfaction with overall JV performance and in comparison with original objectives. An evaluation of outcomes relative to objectives is a critical variable in social exchange theory (Anderson and Narus, 1990; Kelley and Thibault, 1978). Reported satisfaction with JV performance and performance relative to initial objectives has been found to be closely related to objective measures of JV performance (Geringer and Hebert, 1991). The perception that a partner is cooperative and contributing to mutual JV outcomes should result in satisfaction with the relationship. Two additional relationships involving satisfaction are predicted. An open relationship should contribute to partner satisfaction, particularly if knowledge acquisition and learning are JV objectives.* Thus, we predict a positive path from openness to satisfaction. Finally, a JV that is closely integrated with its parent should be one that is perceived to have satisfactory performance; hence, the predicted positive path between integration and performance. Methods The Empirical Base The empirical analysis is based on data collected from a sample of North American-Japanese JVs located in North America.? A single industry, the automotive supply industry, was used as the source of sample firms. Using a single industry with a homogeneous set of organizations imposes certain constraints. In particular, theory development is restricted to limited domain or middle-range theories (Pinder and Moore, 1980) and generalizablity is limited to other industries sharing similar structural characteristics. A single industry does, however, provide a greater degree of control over market and environmental peculiarities (Conant et al., 1990) and increases the internal validity of the study. This was considered important since JVs are a particularly complex form of hybrid organization (Borys and Jemison, 1989) and given the limited research on the antecedents of international JV performance. A cross-sectional field study incorporating multiple methods of data collection was used. The data for the study were collected from senior managers associated with JV firms. Two data collection methods were used: self-administered questionnaires and field interviews. While the use of field interviews limited the sample size, a multimethod research design was chosen
*See Kogut (1988) and Westney (1988) for a discussion of learning as a JV objective. “yAlthough there were two Canadian partner firms in the sample, for brevity we refer to “American” rather than “North American” in future references to the sample of firms.
International Joint Ventures
and Performance
206 International Business Review 333
for several reasons. First, the use of multiple measurement methods increased the validity of the research model. Specifically, the integration and satisfaction measures incorporated data generated by both interviews and questionnaires. Second, the two methods allowed descriptive realism to be combined with the data used to test the hypotheses (Lee, 1991). Third, participation rates for other research studies using interview-based approaches have been much higher than those using exclusively mail questionnaire designs (e.g. Geringer, 1988). For logistical reasons the geographic scope of JV location was restricted to Ontario, the United States midwest (Ohio, Michigan, Illinois, and Indiana) and the United States upper south (Kentucky and Tennessee). The number of partners was held to two because of the complexity of ventures with more than two partners. To identify qualifying JVs, a Toronto-based organization called Pacific Automotive Cooperation (PAC) was contacted. PAC’s mission is to introduce North American and Japanese JV partners and to promote Japanese investment and the purchase of Canadian-made parts by Japanese automakers. PAC monitors and tracks Japanese automobile-related investment in North America. PAC also helps firms learn about Japanese management through workshops on topics such as total quality control and continuous improvement. PAC agreed to supply a list of Japanese auto-related investments in North America. The PAC list included both wholly-owned Japanese firms and Japanese-North American JVs. In addition to the PAC list, two other sources were used to ensure completeness: URIS Business Dateline CD-ROM and Ward’s Auto World. Business Dateline is a comprehensive collection of regional business publications in Canada and the United States. Articles, including announcements of new business developments such as JVs, are stored in full text and are accessed using key word searches. Ward’s Auto World is a monthly automotive industry publication that includes a summary of automotive supplier firm activities (e.g. JVs, acquisitions, mergers). A total of 54 JVs with the specified attributes were identified as the population of ventures within the restricted geographic scope. Of the 54 ventures contacted, three did not qualify for the study (e.g. one firm was a JV between an American company and a Japanese-American JV). Overall, the study achieved a response rate of 82% with 42 of 51 qualifying JVs participating in the study. The final sample included 40 JVs; two cases were eliminated because of incomplete data. All sample JVs were suppliers to the automotive industry. With two exceptions, all JVs were startup or greenfield organizations. The primary JV motive for most American partners was market access to the Japanese transplant automakers in North America. Only five ventures supplied domestic automakers exclusively and 11 supplied a single transport exclusively. The prevalent opinion of managers involved with transplant customers was that American firm access to the transplant market would have been difficult without a Japanese partner. More than three-quarters of the ventures were first-tier suppliers dealing directly with the automotive
207 manufacturers. There were 17 cases of equal ownership, 15 cases of Japanese partner majority ownership, and eight cases of American partner majority ownership. The number of years the JVs were in operation ranged from one to six years. Two-thirds of the cases were between three and four years old and only two JVs were as old as six years. The mean number of JV employees was 206 and 15 JVs had between 150 and 300 employees. American partners initiated about the same number of JVs as Japanese partners and for one-third of the JVs neither partner was a clear initiator. The American JV parents were almost evenly divided between single business firms and divisions of diversified firms. There were 22 single business parents and 18 diversified parents in the sample. Data Collection Following a pretest, initial contact with the sample firms was made with a letter of introduction that briefly outlined the proposed research. Managers were then contacted by telephone to arrange an interview. After agreeing to participate in the study, informants were sent a confirmation letter with the self-administered questionnaire. Informants were requested to bring the completed questionnaire to the interview. The primary informant was the senior American manager in the JV. In most cases, this informant was the JV general manager or president, an approach supported by Geringer and Hebert (1991). With one exception, interviews were conducted in-person in the informants’ offices. Total interview time per scheduled interview ranged from 1 h to more than 4 h, with an average of 2 h. All follow-up interviews, carried out for about one third of the cases, were done by telephone. The interviews followed a semistructured format based on an interview guide. There was both an open-ended sequence of questions and a focused set of questions designed to measure specific variables. The open-ended questions allowed theory development to be grounded in the experiences and terminology of organizational participants (recommended by Beyer and Trite, 1982). For all interview data, data reduction began immediately following the interview and helped bring the raw data into a manageable form. Within a 24h period, the detailed interview write-ups were completed. The interview write-ups summarized the interviews in a consistent and logical manner. The main objective of the write-ups was a “product intelligible to anyone, not just the fieldworker” (Miles and Huberman, 1984, p. 50). All write-ups were reviewed for omissions and clarity problems with follow-up data collected if necessary. Analysis Procedure The Partial Lease Squares (PLS) method of structural equation modeling was used to represent the constructs of the model and test the hypotheses.* PLS *Version
1.8of the PLS program was used.
International Joint Ventures and Performance
208 International Business Review 333
assigns values to the loadings of indicators (manifest variables) and paths between the constructs (latent variables). The main objective of PLS is to maximize the explained variance in dependent (endogenous) latent variables. PLS is a second generation multivariate technique that avoids the restrictive assumptions of first generation methods such as multiple regression (Fornell, 1982). For two main reasons, PLS is an appropriate technique when a complex model with multiple paths is hypothesized. (1) All relationships are modeled simultaneously, eliminating concerns about mulitcollinearity. (2) Constructs are defined within the theoretical context of the model such that there is an explicit error term. In contrast, multiple regression analyzes only a single dependent variable and assumes that the constructs are perfectly measured by single indicators. In comparison to other second generation multivariate methods like LISREL, PLS has several advantages. PLS makes limited demands about measurement scales, sample size and the distribution of residuals. On that basis, PLS trades parameter efficiency for accuracy, simplicity, and fewer assumptions (Fornell and Bookstein, 1982, p. 450). Construct Indicators The organization-level indicators for the constructs that were measured by questionnaire are summarized in the Appendix. All paths between the latent and manifest variables are reflective on the grounds that the constructs have theoretical meaning rather than being composed of several operational measures (i.e. formative). The communication indicators (Cl and C2) were based on measures used in a study by Hall et al. (1977). The openness indicators (01, 02, and 03) followed the definition developed by Gupta (1987) around the notion of partner willingness to share information. Since the informants were associated with the American partners, an additional question (03) was asked specifically about the Japanese partner’s willingness to share information. The trust indicators (Tl, T2 and T3) were drawn from Beamish’s (1988) study of JVs in developing countries and from the marketing channel partnership work by Anderson and Narus ( 1990). A potential limitation of the trust measure is its focus on trust as a current state rather than trust at the time of JV formation. The cooperation construct (Co1 and Co2) was also based on indicators adapted from Anderson and Narus (1990). This construct captures the degree to which the partners work towards mutual goals. The questionnaire indicators of satisfaction (Sl and S2) were based on venture effectiveness rather than longevity and stability. These indicators followed Geringer and Hebert’s (199 1) recommendation concerning the validity of subjective questionnaire measures of JV performance and were adapted from Geringer and Hebert ( 199 1) and Fomell et al. (1990). Given that JV performance measurement remains an area of some controversy, the questionnaire measure of satisfaction was supplemented by an indicator based on interview data (S3). This procedure took advantage of
209 the PLS property permitting ordinal and interval-scaled manifest variables. The use of multiple methods for construct measurement should also enhance the validity of the research model. Using the detailed interview data, the JVs were classified on the basis of American parent satisfaction with overall JV performance. A similar interview-based satisfaction measure was used by Beamish (1988) and Killing (1983). Hallen et al. (1991) used a similar methodological approach to measure firm adaptation and Ibarra (1993) used coded informant reports to measure administrative and technical innovation involvement. A three category ranking (high, medium, and low) was used for satisfaction. All codings of interview data for the purposes of the ranking were done independently of the questionnaire data. The 11 ventures classified as low satisfaction were those that had been, or would likely be, terminated because of unsuccessful working relationships. For these ventures, we looked for unplanned terminations that occurred because of serious problems in the working relationship between partners. The emphasis was on unplanned terminations because as JV researchers have argued (e.g. Anderson, 1990; Parkhe, 1991), success of a JV should not be judged solely on the basis of longevity. In all but one case, termination or likely termination represented a failure situation. In the nonfailure termination case, the JV was terminated following the takeover of the American firm and a policy decision to sell the JV interest. The classification of the 29 remaining ventures into medium and high satisfaction categories was based on several factors. One factor was the degree of satisfaction with JV financial performance. In 21 JVs out of the sample of 40, the American parent indicated a dissatisfaction with financial performance. Of the 21, 11 were ventures already classified as low satisfaction. The other 10 were classified as medium satisfaction ventures. In addition to financial performance concerns, we looked for evidence of American partner dissatisfaction in two key areas: (1) management of customer relationships; (2) management of the manufacturing process. These areas were selected because market access was the American partner’s primary formation motive in the majority of cases and the Japanese partner generally exercised control over manufacturing. Partner dissatisfaction in these areas resulted in eight additional ventures classified as medium satisfaction ventures, for a total of 18 medium satisfaction ventures. The remaining 11 JVs were classified as high satisfaction ventures. In several of these ventures, there were plans for further partner commitments, such as expansion of the JV business. This does not mean a complete absence of partner conflict in the successful ventures. However, these ventures did appear to be functioning as relatively stable businesses without signs of instability. The integration construct also incorporated multiple method measures. The degree of integration between the parent and its JV was considered indicative of the parent’s willingness to make a commitment to the venture. Following the discussion by Shortell and Zajac (1988), two questionnaire indicators
International Joint Ventures and Performance
210 International Business Review 373
reflected the involvement of parent managers in JV decisions (11 and 12). A third indicator was based on the frequency of communication between JV and parent managers as measured by a series of questionnaire responses (13). The fourth indicator was derived from the interview data (14). The JVs were classified into low, medium, and high integration categories based on information concerning: -
product symmetry between the JV and the American parent; cooperation in research and development between the JV and the parent; cooperation in customer relationship management; sharing of top management by the JV and the parent; extensive communication between the JV and the parent.
Results Estimation The results of the PLS estimation are presented in Fig. 2. With PLS, an initial analytical procedure is to examine the indicator loadings to assess indicator contributions to the model. The loading is equivalent to the simple correlation coefficient between an indicator and the latent variable (Fornell et al., 1990). As a general rule, each indicator should have explanatory variance greater than its error variance. In other words, the explained variance in the measure should be greater than 50%, which means the loading should be 0.70 or higher (Carmines and Zeller, 1979). With the exception of 13, the indicators have loadings greater than 0.70. Several explanations are possible for the 13
Fig. 2. Results from the PLS Estimation
211 loading of less than 0.70. The measure could be poor, the construct could be multidimensional, or there could be shared method variance with other indicators. Bearing in mind that 13 (along with 14 and S3) differed significantly in measurement technique from the other measures, the low loading may be indicative of the multidimensionality of integration. The 13 indicator may therefore reinforce the construct validity of integration at the slight expense of reliability. The validity of the construct indicators can also be evaluated. Internal consistency is calculated by taking the sum of the loadings all squared and divided by the sum of the loadings all squared plus the sum of the error variances. The result is analogous to a reliability measure like Cronbach’s Alpha. However, Fornell and Larcker (1981) suggested that because there is no assumption that each indicator is making an equal contribution to the construct, this test of internal consistency may be more relevant than Alpha. Internal consistency was satisfactory for the estimated model, ranging from 0.82 to 0.87 (Table 1). Discriminant validity is assessed by calculating the average variance extracted for each construct (the arithmetic mean of the covariance between construct and its indicators) and comparing it to covariance between the construct and the other constructs (Fomell and Larcker, 1981). For this model, the average variance extracted ranged from 0.69 to 0.92 and in all cases was higher than the covariance with other constructs (Table 1). The structural model converged after five iterations. Although there is no overall test of goodness-of-fit in PLS (Fomell and Barclay, 1983), individual paths can be tested for significance using a jack-knifing technique (Fomell et al., 1990). By temporarily removing observations from the model, a distribution of parameter estimates is constructed and t-values are calculated for each path. The results, shown in Table 2, indicate that seven of the nine hypothesized paths were significant (with two paths significant at P < 0.10). Further verification of the model is obtained from an examination of the variance explained. The dependent or endogenous constructs trust, cooperation and satisfaction had R* values of 0.47-0.67, adding support to the theoretical soundness of the model. The structural model explained nearly
Discriminant
Construct Communication Openness Trust Cooperation Integration Satisfaction
International Joint Ventures and Performance
validity
Internal consistency
Average variance extracted
Maximum covariance
0.87 0.82 0.83 0.84 0.83 0.84
0.92 0.75 0.77 0.88 0.69 0.80
0.07 0.62 0.61 0.62 0.11 0.42
Table 1. Internal Consistency and Discriminant Validity
212 International Business Review 393
50% of the variance in satisfaction suggesting that while there are other factors that influence how JVs will perform, an interorganizational perspective provides useful insights into JV performance. Interpretation of Path Results As predicted, cooperation and relationship openness had a strong direct effect on satisfaction while trust and communication indirectly influenced satisfaction. The predicted relationship between communication and openness was not supported. A possible explanation is that the communication measures were oriented toward operational interactions while openness was aimed at more strategic information. As a result, the extent and quality of operational communication would not necessarily be related to openness in terms of revealing organizational skills and capabilities. The path from cooperation to integration was significant, suggesting that cooperation influences the parent’s commitment to the venture via parent-venture integration. However, the link between integration and satisfaction was weak and not significant. This result was surprising given our perspective that integration is an outcome of the partner exchanges. The results suggest that while a cooperative relationship may increase commitment to the relationship, the assessment of performance may be independent of commitment. The underlying explanation may be that the degree of integration between the parent and its JV is not only evidence of commitment but also evidence of an expectation of future strategic value from the venture. Thus, the relationship between an expectation of future strategic value and satisfaction with historical results may not be strongly related. Discussion Previous research in the area of JV performance has been primarily focused on either relationships between JV control and performance (e.g. Beamish, 1988; Killing, 1983) or performance and JV stability (e.g. Kogut, 1989). By
Path
Table 2. Path Estimates
Communication - openness Communication - trust Openness - trust Openness - cooperation Trust - cooperation Cooperation - integration Openness - satisfaction Cooperation - satisfaction Integration - satisfaction ?P < 0.10. *P < 0.05. **p < 0.01. ***p < 0.001.
Path estimates 0.04 0.20** 0.77*** 0.56*** 0.30t 0.32* 0.38* 0.32t 0.08
213 developing a model of performance from an interorganizational exchange theory perspective, this study adds a new dimension to the understanding of JV performance. The study found that the quality of the exchange relationship was an important factor in the parent’s assessment of JV performance. The findings were supported by the interviews with JV managers. Factors such as trust and cooperation were repeatedly identified as ingredients in a successful working JV relationship. The results support the argument that compatibility between the partners is a crucial factor in determining the partners’ relative bargaining power and operating relationship norms. The overall strategic objective of JV parents is the pooling of resources to create value in a way that each of the parents could not achieve by acting alone (Borys and Jemison, 1989). Value creation refers to the process of combining the capabilities and resources of the partners to perform a joint task that has the potential to create monetary or other benefits for the partners. While the perceived value to each of the parents need not be the same, the findings of this study suggest that the partners may place a greater emphasis on complementarity in objectives than has previously been noted in the literature. In looking at the relationships in the model, questions could be raised about several of the path directions. Cooperation is shown as a consequence of trust. Anderson and Nat-us (1990) concluded that cooperation may be antecedent to trust because trust is a present state and cooperation taps past cooperative behavior. While we would argue that trust helps to establish expectations of cooperation, conclusions about causality derived from a static model must be viewed as tentative. The relationship between satisfaction and integration, which was not significant, may be reciprocal. Iteratively, satisfaction could lead to a greater willingness to integrate the JV with the parent which in turn could generate greater satisfaction, and so on. Given the critical nature of these paths in the model, further research is required to make inferences about their causal ordering. For this purpose, perhaps a technique such as LISREL could be used. LISREL allows the researcher to model bi-directional and reciprocal causality. The practical implications in this study lie in the guidance that managers might draw regarding the role played by partner interactions in the JV process. The American firms in this study placed a high premium on what we would refer to as the “quality” of interactions. Although other studies have argued for the importance of trust and cooperative behavior,* explicit linkages with performance have not been examined. Turning to JV stability, it should not be assumed that satisfaction and stability are necessarily related. Uncertainties are the norm in any relationship between firms, and especially ones as complex as international JVs. Despite *For example, Harrigan (I 986, p. 148) asserted that “managers can be as crafty as they please in writing clauses to protect their firm’s technology rights, but the JV’s success depends on trust”.
International Joint Ventures and Performance
214 International Business Review 373
the existence of perceived openness and cooperative partner behavior, partner objectives may diverge and lead to venture instability. Hamel et al. (1989) contended that a high degree of cooperation focused on long-term win/win outcomes may not be productive. They suggested that too much emphasis on the JV’s survival can detract from the competitive goals of the parents. Limitations and Directions for Future Research This is one of the first studies to model JVs from an interorganizational perspective and therefore, has laid the groundwork for understanding the causal sequences of partner interactions. Nevertheless, the findings of the study should be interpreted in light of several limitations. We used a structural equation modeling approach in conjunction with multiple methods of construct measurement. In comparison with previous studies of JV performance, a larger number of constructs was systematically investigated. The trade-off in making the model more comprehensive is the risk associated with reducing theoretically complex phenomena to simple observable indicators. Also, the sample size was constrained by the cost of conducting field research and by the desire to control for industry. Although PLS makes limited demands on sample size, a larger sample would have enhanced the findings. Another issue is the extent to which the results can be generalized. It should be noted that a single industry study was most appropriate for this kind of investigation. Multi-industry data could introduce substantial variation in response characteristics from extraneous factors, such as different industry pressures for internationalization or changes in supplier-manufacturer relationships. By limiting the study to a single industry, variation in response characteristics from extrinsic factors was reduced. However, the theory and ideas specifically related to JVs should be relevant to different industries. Accordingly, the results of this study need to be extended to other types of industry contexts. Finally, the question of how JV performance should be evaluated remains a complex issue. Anderson (1990), for example, proposed that performance should be assessed based on two dimensions: one, how well managers understand the JV transformation process and two, how thoroughly JV results can be assessed, measured, and judged. In this study, performance was assessed based on the parent’s satisfaction with overall JV performance. To some degree, the structural and strategic homogeneity of the JVs in this study negated the concerns outlined by Anderson (1990). Nevertheless, further research into both the antecedents and the nature of JV performance is encouraged. We hope this study provides a useful framework for examining such issues. References Anderson, E. (1990) Two Firms One Frontier: On Assessing Joint Venture Performance. Sloan Management Review, Vol. 18, Winter, pp. 19-30. Anderson, J.C. and Narus, J.A. (1990) A Model of Distributor Firm and Manufacturer Firm
215 Working Partnerships. Journal of Marketing, Vol. 54, January, pp. 42-58. Beamish, P.W. (1988) Multinational Joint Ventures in Developing Countries. Routledge, London. Beyer, J.M. and Trite, H.M. (1982) The Utilization Process: a Conceptual Framework and Synthesis of Empirical Literature. Administrative Science Quarterly, Vol. 27, pp. 591622. Blau, P.M. (1964) Exchange and Power in Social Life. Wiley, New York. Borys, B. and Jemison, D.B. (1989) Hybrid Arrangements as Strategic Alliances: Theoretical Issues in Organizational Combinations. Academy of Management Review, Vol. 14, No. 1, pp. 234-249. Buckley, P.J. and Casson, M. (1988) A Theory of Cooperation in International Business, in Contractor, F.J. and Lorange, P. (Eds), Cooperative Strategies in International Business, pp. 31-53. Lexington Books, Toronto, Ontario. Carmines, E.G. and Zeller, R.A. (1979) Reliability and Validity Assessment. Sage, Beverley Hills, California. Chowdhury, J. (1989) International Joint Ventures: Some Interfirm Organization Specific Determinants of Success and Failures - A Factor Analytic Exploration. Unpublished Doctoral dissertation, Temple University, Philadelphia, Pennsylvania. Conant, J.S., Mokwa, M.P. and Varadarajan, P.R. (1990) Strategic Types, Distinctive Marketing Competencies, and Organizational Performance: a Multiple Measures-based Study. Strategic Management Journal, Vol. 11, pp. 365-383. Contractor, F.J. and Lorange, P. (1988) Why Should Firms Cooperate: The Strategy and Economics Basis for Cooperative Ventures, in Contractor, F.J. and Lorange, P. (Eds), Cooperative Strategies in International Business, pp. 3-30. Lexington Books, Lexington, Massachusetts. Dwyer, F.R., Schnurr, P.H. and Oh, S. (1987) Developing Buyer-Seller Relationships. Journal ofMarketing, Vol. 51, April, pp. 1 l-27. Fomell, C. (1982) A Second Generation of Multivariate Analysis. Praeger, New York. Fornell, C. and Barclay, D.W. (1983) Jackknifing: a Supplement to Lochmoller’s LVPLS Program, Unpublished paper, Graduate School of Business Administration, University of Michigan. Fornell, C. and Bookstein, F.L. (1982) Two Structural Equation Models: LISREL and PLS Applied to Consumer Exit-Voice Theory. Journal of Marketing Research, Vol. 19, November, pp. 440452. Fomell, C. and Larcker, D.F. (198 1) Evaluating Structural Equation Models with Unobservable Variables and Measurement Error. Journal of Marketing Research, Vol. 18, February, pp. 39-50. Fomell, C., Lorange, P. and Roos, J. (1990) The Cooperative Venture Formation Process: a Latent Variable Structure Modeling Approach. Management Science, Vol. 36, No. 10, pp. 1246-1256. Geringer, J.M. (1988) Joint Venture Partner Selection: Strategies for Developed Countries. Quorum, Westport, Connecticut. Geringer, J.M. and Hebert, L. (1991) Measuring Performance of International Joint Ventures. Journal of International Business Studies, Vol. 22, No. 2, pp. 253-267. Gupta, A.K. (1987) SBU Strategies, Corporate-SBU Relations, and SBU Effectiveness in Strategy Implementation. Academy of Management Journul, Vol, 30, No. 3, pp. 477-500. Hall, R.H., Clark, J.P., Giordano, P.C., Johnson, P.V. and Van Roekel, M. (1977) Patterns of Interorganizational Relationships. Administrative Science Quarterly, Vol. 27, pp. 457474. Hallen, L., Johansen, J. and Seyed-Mohamed, N. (1991) Interfirm Adaptation in Business Relationships. Journal of Marketing, Vol. 55, April, pp. 29-37. Hamel, G. (1991) Competition for Competence and Inter-partner Learning within International Strategic Alliances. Strategic Manugement Journal, Vol. 12, Special issue, pp. 83-104. Hamel G., Doz, Y.L. and Prahalad, C.K. (1989) Collaborate with your Competitors, and Win. Harvard Business Review, Vol. 67, January/February, pp. 133-139. Harrigan, K.R. (1986) Managing for Joint Venture Success. Lexington Books, Lexington,
International Joint Ventures and Performance
216 International Business Review 373
Massachusetts. Harrigan, K.R. (1985) Strategies for Joint Ventures. Lexington Books, Lexington, Massachusetts. Harrigan, K.R. and Newman, W.H. (1990) Bases of Interorganizational Co-operation: Propensity, Power, Persistence. Journal of Management Studies, Vol. 27, pp. 417434. Hennart, J.F. (1988) A Transactions Costs Theory of Equity Joint Ventures. Strategic Management Journal, Vol. 9, pp. 361-374. Ibarra, H. (1993) Network Centrality, Power, and Innovation Involvement: Determinants of Technical and Administrative Roles. Academy of Management Journal, Vol. 36, No. 3, pp. 471-501. Jarillo, J.C. (1988) On Strategic Networks. Strategic Management Journal, Vol. 9, pp. 3141. John, G. and Reve, T. (1982) The Reliability and Validity of Key Informant Data from Dyadic Relationships in Marketing Channels. Journal of Marketing Research, Vol. 19, January, pp. 5 17-524. Kanter, R.M. (1989) When Giants Learn to Dance. Simon & Schuster, New York. Kelley, H.H. and Thibault, J.W. (1978) Interpersonal Relations: a Theory of interdependence. Wiley, New York. Killing, J.P. (1982) How to make a Global Joint Venture Work. Harvard Business Review, Vol. 60, May/June, pp. 120-127. Killing, J.P. (1983) Strategies for Joint Venture Success. Praeger, New York. Kogut, B. (1988) Joint Ventures: Theoretical and Empirical Perspectives. Strategic Management Journal, Vol. 9, pp. 3 19-322. Kogut, B. (1989) The Stability of Joint Ventures: Reciprocity and Competitive Rivalry. The Journal of Industrial Economics, Vol. 38, pp. 183-198. Lee, A.S. (1991) Integrating Positivist and Interpretive Approaches to Organizational Research. Organization Science, Vol. 4, pp. 342-365. Miles, M.B. and Huberman, A.M. (1984) Qualitative Data Analysis: a Sourcebook of New Methods. Sage, Newbury Park, California. Parkhe. A. (1991) Interfirm Diversity, Organizational Learning, and Longevity in Global Strategic Alliances. Journal of International Business Studies, Vol. 22, No. 4, pp. 579-602. Pinder, C.C. and Moore, L.F. (1980) The Inevitability of Multiple Paradigms and the Resultant Need for Middle-range Analysis in Organizational Theory, in Pinder, C.C. and Moore, L.F. (Eds), Middle Range Theory und the Study of Organizations, pp. 87-100. Martinus Nijhof, Boston, Massachusetts. Porter, M.E. and Fuller, M.B. (1986) Coalitions and Global Strategy, in Porter, M.E. (Ed.), Competition in Gobal Industries, pp. 315-343. Harvard Business School Press, Boston, Massachusetts. Schmidt, S.M. and Kochan, T.A. (1977) Interorganizational Relationships, Patterns and Motivations. Administrative Science Quarterly, Vol. 22, pp. 220-234. Seabright, M.A., Levinthal, D.A. and Fichman, M. (1992) Role of Individual Attachments in the Dissolution of Interorganizational Relationships. Academy of Management Journal, Vol. 35, No. I, pp. 122-160. Shortell, S.M. and Zajac, E.J. (1988) Internal Corporate Joint Ventures: Development Processes and Performance Outcomes. Strategic Management Journal, Vol. 9, pp. 527-542. Sullivan, J. and Peterson, R.B. (1982) Factors Associated with Trust in Japanese-American Joint Ventures. Management International Review, Vol. 22, No. 2, pp. 30-40. Toyne, B. (1989) International Exchange, a Foundation for Theory Building in International Business. Journal of International Business Studies, Vol. 20, Spring, pp. l-18. Westney, D.E. (1988) Domestic and Foreign Learning Curves in Managing International Cooperative Strategies, in Contractor, F.J. and Lorange, P. (Eds), Cooperative Strategies in Internutional Business, pp. 339-346. Lexington Books, Lexington, Massachusetts. Zand, D.E. (1972) Trust and Managerial Problem Solving. Administrative Science Quarterly, Vol. 17, pp. 299-239. Received January 1994
217 Indicator*
Measure
Cl
North American joint venture managers regularly communicate partner managers
c2
North American joint venture managers partner in joint venture decisions
involve managers from the Japanese
01
Both partners willingly share information
with each other
02
The relationship
03
The Japanese partner willingly shares information
Tl
The Japanese partner is a firm that stands by its word
with Japanese
between the joint venture partners is open and informal
T2
There is a high level of trust in the working relationship
between the partners
T3
The North American partner can rely on its partner venture management agreement
to abide by the joint
co2
The efforts of the partners are complementary objectives of the joint venture
co1
The partners venture
Sl
The North American venture
s2
The joint venture has met the objectives for which it was established
11
Regular communication between joint venture managers and North American partner head office managers
I2
Involvement of managers from the head office of the North American partner in joint venture decisions
actively
*All indicators use seven-point
International Joint Ventures and Performance
work together
and consistent
with the
so that both can benefit from the joint
parent is satisfied
with the performance
scales. Revised March 1994
of the joint
Appendix. Summary of Model Indicators Measured by Questionnaire