- Email: [email protected]
Knowledge acquisition, knowledge loss, and satisfaction in high technology alliances
Journal of Business Research 57 (2004) 610 – 619
Knowledge acquisition, knowledge loss, and satisfaction in high technology alliances Patricia M. Norman* Department of Management, Hankamer School of Business, Baylor University, P.O. Box 98006, Waco, TX 76798-9006, USA Received 5 January 2000; received in revised form 2 January 2002; accepted 22 May 2002
Abstract Drawing on the organizational learning and transaction cost economics (TCE) literature, this study examines how learning intent, opportunities to learn, and a firm’s ability to learn facilitate or hinder three alliance outcomes: knowledge acquisition by the focal firm, knowledge loss to the partner, and alliance satisfaction. The research model proposes that firms attempt to influence learning opportunities based on a partner’s intent and ability to learn as well as on the trust the firm has in the partner. A partner’s learning intent and ability are positively associated with the extent to which a firm protects its own firm-specific knowledge, but they only have significant effects on one alliance outcome, knowledge loss. With more trusted partners, firms are less protective of knowledge and tend to acquire more knowledge, lose less knowledge, and be more satisfied. Equity alliances are associated with lower levels of knowledge loss and higher levels of satisfaction. D 2002 Elsevier Inc. All rights reserved. Keywords: Alliance; High technology; Knowledge; Learning
Competition requires that firms continually acquire and develop new knowledge and skills. A great deal of attention has recently been focused on knowledge acquisition as an important outcome for firms engaged in strategic alliances (Lin and Germain, 1999). Because alliance success has been linked to learning and knowledge sharing (Crossan and Inkpen, 1995), partners have been urged to create an environment conducive to learning. Such an environment, however, can expose a firm’s critical knowledge and capabilities to a partner and may lead to imitation or appropriation. To prevent the loss of firm-specific knowledge, firms may try to prevent such losses by limiting a partner’s learning opportunities. These actions, however, may also reduce the firm’s own opportunities to learn and may affect alliance success. Lyles and Salk (1996) suggest that we need a greater understanding of what facilitates knowledge acquisition and skills development. Despite their important role in knowledge acquisition, however, few studies have empirically examined the learning outcomes of alliances. In one of the
*
Tel.: +1-254-710-6196; fax: +1-254-710-1093. E-mail address: [email protected] (P.M. Norman).
0148-2963/$ – see front matter D 2002 Elsevier Inc. All rights reserved. doi:10.1016/S0148-2963(02)00395-8
few studies to do so, Simonin (1997) investigated how collaborative experience and know-how affected knowledge acquisition. He suggests that ‘‘. . .firm- and alliance-specific variables such as strategic intent, transparency, organizational capabilities, and resource commitments’’ (p. 1170) must also be studied. In line with this recommendation, this paper presents the results of an exploratory study that investigates firm- and alliance-specific factors associated with firm actions to limit partner learning because of the threat of unwanted knowledge acquisition by the partner and, in turn, how these actions and other factors affect the firm’s own knowledge acquisition and its partners’ knowledge acquisition. In high technology product development alliances, partners are interdependent because each partner must contribute resources for the development to succeed (Gulati et al., 1994). Reciprocal information exchange is necessary to complete designs and to enable mutual adjustments (Osborn et al., 1998). Greater frequency and quality of information exchange can increase the innovativeness and quality of products designed, while lowering the costs of development (Larson, 1992). Such exchanges, however, create learning opportunities that enable a firm to appropriate knowledge from partners. Thus, the risk of knowledge appropriation is
P.M. Norman / Journal of Business Research 57 (2004) 610–619
particularly high in development alliances (Park and Kim, 1997) and these alliances are the context used for this study.
1. Theoretical background The model developed in this paper relies heavily on organizational learning, which stresses the importance of knowledge acquisition, and transaction cost economics (TCE), which helps to explain choices about structuring and controlling alliances. 1.1. Organizational learning Huber (1991, p. 89) takes a behavioral perspective on organizational learning and suggests that ‘‘an entity learns if, through its processing of information, the range of potential behaviors is changed.’’ This paper focuses on knowledge acquisition, one of four organizational learning processes, which occurs when any part of a firm gains knowledge that is recognized as potentially useful (Huber, 1991). The ability of firms to acquire and exploit knowledge has been stressed by organizational learning scholars (e.g., Cohen and Levinthal, 1990; Huber, 1991) and has been liked to a firm’s ability to innovate (Fiol, 1996). Alliance partners are a particularly important source of new, external knowledge (Inkpen and Dinur, 1998). Firms have different reasons for and goals associated with alliances. When learning is an explicit and primary goal, acquiring and internalizing skills is a key measure of success (Hamel, 1991). Even when learning is not an explicit goal, however, firms can learn from partners and this learning can enhance their competitive abilities (Crossan and Inkpen, 1995). Therefore, learning is a critical issue regardless of the goals of an alliance. Knowledge is more easily transferred between alliance partners than through market mechanisms (Shenkar and Li, 1999) because learning is a socially embedded process and requires connections through which individuals can share knowledge (von Krogh et al., 1994). With open communication and rich knowledge sharing, alliances become a potentially effective venue for knowledge acquisition (Lincoln et al., 1998). Hamel (1991) suggests that the amount of learning from a partner is influenced by the receiving firm’s intent to learn and its receptivity as well as by the transparency of the partner from whom the knowledge is acquired. Receptivity depends, in part, on exposure to knowledge and on the skills and ability of the acquiring firm. Transparency is influenced by the design of interfaces, the structure of joint tasks, and how protective the partner is of its knowledge. Each firm in an alliance may be simultaneously trying to gain knowledge from its partner, to internalize this knowledge, and then use it to further develop its own competitive advantage (Richter and Vettel, 1995). While some learning is needed to accomplish the alliance goals, too much
611
learning by a partner can strip a firm of critical knowledge. One way to counter this danger is to manage the alliance structure and processes in an effort to control knowledge transfers (Kumar and Seth, 1998). However, limiting communication and information exchange may inhibit learning necessary to meet alliance objectives (Millar et al., 1997). Learning can result in either common or private benefits (Khanna et al., 1998). Common benefits accrue to all partners in an alliance when learning is applied to activities within the alliance. Private benefits, on the other hand, accrue to only one partner when that partner uses what is learned within the alliance or from an alliance partner to generate rents in activities outside of an alliance. While both types of benefits are important, this study examines only private benefits. 1.2. TCE TCE provides a theoretical grounding for the circumstances under which firms will attempt to limit a partner’s learning opportunities. TCE was traditionally concerned with choices about whether activities were performed within a firm’s boundaries or in the marketplace. More recently, TCE has been applied to alliances, specifically how to structure relationships to try to mitigate the transaction hazards that are present in a relationship. Opportunism, which occurs when one party acts in its own self-interest and does so in a deceitful manner (Moschandreas, 1997), is central to a discussion of transaction hazards. While not all parties will to act opportunistically, uncertainty and bounded rationality prevent a decision-maker from knowing a priori which parties will act opportunistically (Williamson, 1991). Thus, mechanisms will be designed to prevent losses from potential opportunism. Recent criticisms of TCE have questioned the assumption that individuals are primarily self-interested and other motives such as altruism have been noted (e.g., Moschandreas, 1997). In this vein, trust has also received increasing attention. The melding of trust with TCE can provide a more complete and accurate picture of how transactions are structured. Trust is the willingness to be vulnerable to another party when their behavior cannot be controlled (Mayer et al., 1995). This definition implies that a trusted party has the opportunity to act opportunistically, but the trustor believes that the party will not act in such a manner. Thus, trust can reduce transaction costs and the need for protective mechanisms (Ring and Van de Ven, 1992) and encourage beneficial behaviors such as more accurate and comprehensive information exchange (Chiles and McMackin, 1996), greater communication (Currall and Judge, 1995), and fewer actions to safeguard knowledge and monitor partner behavior (Inkpen and Li, 1999). In relationships lacking in trust, structural mechanisms are used to either limit the opportunity that a partner has to act opportunistically or to deter opportunistic behavior through pen-
612
P.M. Norman / Journal of Business Research 57 (2004) 610–619
alties that are so severe that they discourage self-interested behavior.
acquisition (Inkpen and Dinur, 1998) and less likely to take actions to appropriate the focal firm’s knowledge. This leads to the following hypothesis:
2. Research model and hypotheses
Hypothesis 1a: The higher the partner’s learning intent, the higher the knowledge protection by the focal firm.
Given the importance of learning and the dangers of knowledge appropriation, this paper develops a model of factors that influence learning outcomes. It is based, in part, on the factors identified by Hamel (1991)—intent to learn, learning opportunities, and ability to learn. Lack of intent, lack of opportunities to learn resulting from a lack of information sharing, and an inability to learn all hinder knowledge acquisition (Larsson et al., 1998). In addition, the model integrates trust as a factor facilitating knowledge transfers. The research model tested examines alliance learning outcomes from the perspective of one party, the focal firm, because the focal firm and its partner enjoy different private benefits. Specifically, the amount of knowledge acquired by the focal firm and the extent to which the focal firm’s competitively valuable firm-specific knowledge is lost to the partner are assessed. While specific hypotheses are not proposed, satisfaction is examined to assess the impact of the model variables on a more general assessment of alliance performance. The model proposes both direct and indirect effects. Specifically, certain factors—partner’s learning intent, partner ability, and trust—are hypothesized to be associated with the amount of knowledge protection by the focal firm and to indirectly affect alliance outcomes through their influence on knowledge protection. Learning opportunities, represented by lower knowledge protection and the use of equity, and trust are expected to directly influence alliance outcomes. 2.1. Factors associated with knowledge protection Firms can structure relationships to protect against the loss of firm-specific knowledge to alliance partners (Kumar and Seth, 1998) by restricting not only the quantity, but also the depth and quality of information that is shared (Park, 1996). This study proposes that firms will attempt to protect knowledge from partners who may learn in ways that are detrimental to the firm. Specifically, firms will be more guarded with partners who have higher learning intents and greater abilities to appropriate knowledge, but will be more open with trusted partners. 2.1.1. Learning intent The perceived partner’s learning intent is the extent to which the focal firm believes that the partner is focused on learning during the alliance. Because not all firms are equally concerned with knowledge acquisition (Richter and Vettel, 1995), learning first requires that a firm have an intent to learn (Hamel, 1991). Without this intent, a partner is less likely to commit resources to knowledge
2.1.2. Ability to learn When trying to appropriate knowledge from an alliance partner, a firm’s absorptive capacity (Cohen and Levinthal, 1990), which is its ability to ‘‘identify, appreciate, and then assimilate such assets’’ (Park and Russo, 1996, p. 878), is critical. Absorptive capacity has been linked to similarity of resources and knowledge bases (Lane and Lubatkin, 1998; Mowery et al., 1996) and similarity of organizational structures and processes (Lane and Lubatkin, 1998) between partner firms. This study uses the similarity of resources and knowledge bases, which allows a partner to more readily understand and appropriate the firm’s knowledge. While some overlap in knowledge is needed to ensure that partners understand and can effectively combine their knowledge, a focal firm is likely to limit the learning opportunities of a partner who has greater ability to take advantage of these opportunities because such a partner is more dangerous than a partner with lower ability. Hypothesis 1b: The higher the partner’s ability to learn, the higher the knowledge protection by the focal firm. 2.1.3. Trust Trust among alliance partners, especially in technology collaborations, is critical (Hausler et al., 1994). Trust reduces the fear of opportunism by partners, which should reduce transaction costs (Williamson, 1991). With trust, the need for monitoring and other control mechanisms is reduced (Inkpen and Li, 1999). In addition, trust encourages behaviors such as open communication and the willingness to share information (Currall and Judge, 1995). Therefore, firms should be less protective with more trusted partners. Hypothesis 1c: The higher the focal firm’s trust in the partner, the lower the knowledge protection by the focal firm. 2.2. Alliance outcomes Hypotheses are now developed for how learning opportunities—lower levels of knowledge protection and the use of equity—and trust affect knowledge acquisition and loss. 2.2.1. Knowledge acquisition Knowledge acquisition refers to skills learned and knowledge acquired by the focal firm from a partner during an alliance.
P.M. Norman / Journal of Business Research 57 (2004) 610–619
2.2.2. Knowledge protection Regular contact between partner employees facilitates the sharing of information, which increases the chances that knowledge will be spread (Goes and Park, 1997; Lincoln et al., 1998). While limitations on sharing may prevent a partner from acquiring knowledge from the focal firm, it can be self-defeating because it also hinders the ability of the focal firm to learn. Partners often respond to each other’s limiting of information sharing by further reducing their own sharing (Larsson et al., 1998), an action that inhibits knowledge acquisition by the focal firm. Therefore, the following hypothesis proposes that increasing knowledge protection will decrease knowledge acquisition. Hypothesis 2a: The higher the level of knowledge protection, the lower the focal firm’s knowledge acquisition. 2.2.3. Equity Conceptually, equity-based alliances are thought to facilitate the transfer of knowledge between partners (Shenkar and Li, 1999) because partners usually have more frequent interactions and commit greater resources to joint efforts (Uzzi, 1997). Empirically, partners in equity-based alliances have been found to increase their patent cross-citation rates, a proxy for learning, to a greater degree than partners in nonequity alliances (Mowery et al., 1996). Thus, equity alliances are expected to provide greater learning opportunities and greater knowledge acquisition than nonequity alliances, as proposed in the following hypothesis. Hypothesis 2b: The focal firm will experience higher knowledge acquisition in equity alliances than nonequity alliances. 2.2.4. Trust Earlier, it was argued that high trust environments should provide greater learning opportunities because firms are more willing to exchange information with trusted partners. Because of these greater learning opportunities, a firm should be able to learn more from a trusted partner. Therefore, the following hypothesis is proposed: Hypothesis 2c: The higher the focal firm’s trust in the partner firm, the higher the focal firm’s knowledge acquisition. 2.3. Knowledge loss Some, but not all, learning by a partner represents a danger to the focal firm. Partner learning is a danger when it results in the loss of competitively important knowledge. To assess the competitive impact on the focal firm, knowledge loss to a partner that is detrimental to the focal firm and not all forms of partner learning are examined here.
613
2.3.1. Knowledge protection Firm actions to limit knowledge sharing and interactions should reduce the opportunities for partner learning. Thus, knowledge protection should make it harder for a partner to duplicate critical skills and capabilities of the focal firm. In line with this logic, the following hypothesis is proposed: Hypothesis 3a: The higher the level of knowledge protection, the lower the focal firm’s knowledge loss. 2.3.2. Equity Because greater learning opportunities are provided by equity alliances, it may seem that firms engaged in such alliances should experience greater knowledge loss. Yet, equity arrangements may actually reduce the likelihood of knowledge loss to a partner. Equity alliances often rely on more formal mechanisms that may provide some protection against opportunism (Oxley, 1999). In addition, equity investments act as ‘‘hostages,’’ aligning the interests of the parties and reducing the likelihood that a partner will act opportunistically (Gulati, 1995) because such behavior will reduce return on the partner’s investment. Thus, even though greater learning opportunities exist, partners in equity alliances are less likely use knowledge in ways that are detrimental to a firm than partners in nonequity alliances. Hypothesis 3b: The focal firm will experience lower knowledge loss in equity alliances than nonequity alliances. 2.3.3. Trust A trusted partner is not expected to act opportunistically. Thus, even when a trusted partner is presented with more learning opportunities and learns a great deal, it is not likely to use the acquired knowledge in ways that are detrimental to the focal firm. Therefore, knowledge loss should be lower with more trusted partners. Hypothesis 3c: The higher the focal firm’s trust in the partner firm, the lower the focal firm’s knowledge loss. 3. Methods 3.1. Sample Public and private firms specializing in microelectronics, computers, and telecommunications were surveyed. Respondents chose an alliance that involved joint development or modification of a product or service and filled out the survey with reference to a specific partner. Each firm in the sample reported on one or two alliances. Initially, 135 CEOs were sent a letter explaining the project and requesting their firms’ participation. Ten of these firms participated, completing surveys on 15 alliances. Because of the limited success of the initial approach, press releases were examined for firms identified as computer,
614
P.M. Norman / Journal of Business Research 57 (2004) 610–619
telecommunications, or microelectronics firms by Hoover’s online. Surveys were mailed to 252 firms with press releases about joint development alliances. In almost all cases, the points of contact were senior managers involved in the alliance. In an effort to increase the response rate, respondents were offered a summary of the study findings. In addition, follow-up phone calls were made to all potential respondents who had not returned the surveys after three weeks. Thirty-eight firms completed surveys on a total of 47 alliances. The overall response rate was 13.6%, with a rate
of 7.4% for the first method and 15.1% for the second method. This response rate is similar to others who have surveyed top management (e.g., Geletkanycz, 1997) and to those who have studied similar issues in alliances (e.g., Johnson et al., 1996). Four surveys did not contain complete data. In total, 58 useable surveys were received. T-tests compared the data gathered by the two different collection methods. Alliances from the first collection method had greater duration and greater knowledge acquisition. When both alliance duration and a categorical variable
Table 1 Factor analysis of multi-item variablesa Items
Our firm severely restricts partner access to our strategic information. Our firm severely restricts partner access to our marketing plans and information. Our firm severely restricts partner access to our technical information. The alliance has highly structured channels of communication between partners. Working level employees from our firm and the partner never communicate. We trust that our partner’s decisions will be beneficial to the alliance. There is a high level of trust in the working relationship with our partner. We can rely on our partner to abide by the alliance agreement. We trust that our partner’s decisions will be beneficial to our firm. As a result of this alliance, we have improved existing management skills. As a result of this alliance, we have developed new management skills. As a result of this alliance, we have developed new technical skills. As a result of this alliance, we have improved existing technical skills. Our firm is satisfied with the performance of the alliance. Our partner is satisfied with the performance of the alliance. As a result of this alliance, we lost valuable skills and knowledge to our partner. One of our partner’s objectives in forming the alliance was to learn about our management techniques. One of our partner’s objectives in forming the alliance was to learn about an unfamiliar market. One of our partner’s objectives in forming the alliance was to learn about our technical information. Percentage of variance explained Cronbach’s a
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Knowledge protection
Trust
Knowledge acquisition
Satisfaction
Learning intent
0.823
0.069
0.080
0.014
0.077
0.796
0.191
0.082
0.080
0.110
0.763
0.023
0.028
0.214
0.070
0.737
0.308
0.014
0.279
0.038
0.455
0.302
0.235
0.203
0.217
0.098
0.863
0.123
0.318
0.082
0.126
0.825
0.196
0.283
0.005
0.161
0.759
0.061
0.065
0.021
0.342
0.632
0.143
0.191
0.126
0.060
0.041
0.900
0.022
0.243
0.116
0.037
0.801
0.096
0.191
0.064
0.249
0.618
0.122
0.186
0.179
0.251
0.550
0.131
0.082
0.199
0.232
0.098
0.842
0.054
0.101
0.362
0.004
0.787
0.047
0.086
0.168
0.043
0.364
0.333
0.212
0.054
0.137
0.024
0.750
0.155
0.018
0.024
0.128
0.558
0.059
0.014
0.131
0.021
0.435
0.258 .85
0.256 .89
0.200 .82
0.165 .85
0.121 .57
a Items were measured on a seven-point scale, anchored by 1 = strongly disagree and 7 = strongly agree. Items for knowledge protection (Calantone et al., 1993), trust (Inkpen, 1992; Mohr and Spekman, 1994), satisfaction (Glaister and Buckley, 1998) and learning intent (Inkpen, 1992) were modified for the study context. Items for knowledge acquisition and knowledge loss were developed for this study.
P.M. Norman / Journal of Business Research 57 (2004) 610–619
representing collection method were regressed against knowledge acquisition, only alliance duration was significant. Alliance duration is included as a control in the models tested. Data was gathered from Hoover’s online for all firms that were mailed surveys to test for nonresponse bias. T-tests comparing the mean sales, number of employees, and firm age for responding and nonresponding firms revealed no significant differences, indicating that response bias is not a problem. 3.2. Measures A survey instrument was constructed using a panel to provide subjective evaluations of content validity (Rosenthal and Rosnow, 1991). Ten judges reviewed the initial survey items, assessed how well the questions covered each construct, and suggested changes. Once the panel determined that adequate content validity had been achieved, two alliance managers completed and critiqued the survey. Based on their comments, the survey was shortened. Factor analysis, using a varimax rotation, was used to assess the discriminate validity of multiitem measures. Table 1 shows the results of this analysis, which included the item for knowledge loss, along with the Cronbach’s a for each measure. All items loaded on the expected factors with loadings above .40. With the exception of partner learning intent, each measure had an a above .80. This new measure’s a (.57) was within the acceptable range suggested by Nunnally (1967) for early stages of research. The other measures did not use multiple items. Respondents were asked to indicate in which of 15 areas their firm had made contributions to the alliance. For each area, respondents used a seven-point scale to assess the similarity of the focal firm and partner. Partner ability was the average similarity for those areas in which the focal firm made contributions to the alliance. Equity was a dummy variable indicating whether or not the alliance involved equity.
615
Two control variables, both transformed using log transformations to correct for nonnormal distributions, are included in the models. Alliance duration, measured by the number of months the alliance had been in existence, is controlled because assessments of learning and satisfaction may differ over time. Relative size was measured by the ratio of the number of focal firm employees to the number of partner firm employees. After the transformation, negative values for relative size indicate that the focal firm is smaller than its partner and positive values indicate that the focal firm is larger than its partner. 3.3. Data analysis Because a structural model with both direct and indirect effects on alliance outcomes is proposed, a two-stage least squares analysis (TSLS) is used. While the results for the ordinary least squares (OLS) regressions are shown, the TSLS allows assessment of both direct and indirect effects. TSLS, rather than another method to estimate structural equations, is used because its estimators are the most robust and appropriate for small sample sizes (Hanushek and Jackson, 1977). Each of the alliance outcomes is the dependent variable in separate models.
4. Findings The zero-order correlation matrix is shown in Table 2. Consistent with suggestions that partners do not learn at the same rate (Hamel, 1991), knowledge acquisition and knowledge loss are uncorrelated (r = .05). Thus, alliances can be tools for learning without endangering critical firm knowledge. Table 3 shows the regression results for each outcome variable. The first stage regression for knowledge protection, used to test 1a, 1b, and 1c, is identical for each model. As predicted in Hypothesis 1a, focal firms are more protective
Table 2 Correlation matrix Variable
Mean
S.D.
1
2
3
4
5
6
7
8
9
10
1. Partner learning intent 2. Knowledge protection 3. Equity (0,1) 4. Partner ability 5. Trust 6. Relative size 7. Alliance duration 8. Knowledge acquisition 9. Knowledge loss 10. Satisfaction
3.49 4.64 0.24 3.56 5.29 0.26 2.67 4.28 2.12 5.30
1.27 1.29 0.43 1.11 1.02 3.22 1.09 1.31 1.14 1.16
1.00 .28 * .18 .01 .08 .07 .30 * .17 .25y .14
1.00 .21 .32 * .44** * .05 .14 .05 .24y .34* *
1.00 .11 .07 .17 .23y .30 * .13 .20
1.00 .33 * .38* * .15 .01 .14 .14
1.00 .16 .18 .28 * .28 * .54** *
1.00 .05 .14 .23y .00
1.00 .40* * .26y .18
1.00 .05 .13
1.00 .39* *
1.00
n = 58. * P < .05. ** P < .01. *** P < .001. y P < .10.
616
P.M. Norman / Journal of Business Research 57 (2004) 610–619
Table 3 Regression results Independent variables
First stage for TSLS
Model 1 knowledge acquisition
Knowledge protection OLS Intercept 3.95 * * (1.21) Predictor variables intent Partner learning intent 0.26 * (0.12) Opportunity Knowledge protection Equity Trust 0.44 * * (0.15) Ability Partner ability 0.24 (0.14) Other Alliance duration Relative size F-value 7.243* * * R2 .29 Adjusted R2 .25
0.93 (1.20)
0.17 (0.13) 0.54 (0.35) 0.59* * * (0.16)
TSLS
Model 2 knowledge loss
Model 3 satisfaction
OLS
OLS
2.31y (0.90)
0.10 (2.18)
0.04 (0.31) 0.62 (0.40) 0.51 * (0.23)
0.52* * * (0.14) 0.52* * * (0.14) 0.11 * (0.05) 0.10 * (0.05) 6.83* * * 6.34* * * .40 .38 .34 .32
0.14 (0.12) 0.67y (0.34) 0.22 (0.16)
0.26y (0.13) 0.10 * (0.04) 3.57 * * .26 .18
TSLS 0.72 (2.38)
3.45 * * (1.09)
TSLS 3.14 (1.95)
0.62y (0.34) 0.16 (0.11) 0.12 (0.28) 0.99 * (0.44) 0.69 * (0.32) 0.66y (0.36) 0.06 (0.26) 0.49 * * (0.15) 0.52 * (0.21)
0.26y (0.15) 0.08 (0.05) 3.21 * .24 .16
0.14 (0.12) 0.04 (0.04) 6.17* * * .37 .31
0.14 (0.12) 0.04 (0.04) 5.77* * * .36 .29
* P < .05. ** P < .01. *** P < .001. y P < .10.
when partners have higher learning intents. With more trusted partners, firms tend to be less protective, supporting Hypothesis 1c. Hypothesis 1b is not supported; the coefficient for partner ability is not significant. The results for the OLS ( P < .001) and TSLS ( P < .001) regressions for knowledge acquisition are similar. Hypotheses 2a and 2b are not supported. Knowledge protection has no effect on and equity does not significantly increase knowledge acquisition. Hypothesis 2c, which predicted that higher trust would be associated with knowledge acquisition, is supported in both regressions. Both control variables are significant. Knowledge acquisition increases with alliance age. This seems logical because learning takes time and new skills and competencies develop slowly (LeonardBarton, 1995). As alliances age, a firm becomes more familiar with its partner’s skills, which is likely to ease knowledge acquisition. Over time, the respondents may also become more aware of changes in existing skills and the development of new skills. Firms that are larger than their partners tend to acquire less knowledge than firms that are smaller than their partners. This is an interesting outcome because previous literature has suggested (e.g., Park and Ungson, 1997) that smaller firms should be more concerned with the loss of their skills. While smaller firms may be more worried about such losses, they may actually learn more in alliances because they often start with a limited skill base and are exposed to a much broader skill base when they partner with larger firms. Although smaller partners are often sought because of their technological capabilities (Koh and Venkatraman, 1991), larger firms may not interested in or may be unable to internalize the smaller firm’s capabilities. The indirect effects of learning intent (.01), partner ability (.01), and trust ( .02) on knowledge acquisition
are quite small because of the lack of a relationship between knowledge protection and knowledge acquisition. This may be because this study measured knowledge protection as the focal firm’s sharing of information and communication with the partner. What is more likely to affect knowledge acquisition by the focal firm, however, is what information the partner is willing to share with the focal firm. Similarly, partner ability measured resource overlaps only in areas where the focal firm contributed resources. A more appropriate measure for knowledge acquisition may be resource overlaps in areas of the partner’s contributions. The OLS ( P < .001) and TSLS ( P < .05) models for knowledge loss differ. In the OLS model, knowledge loss tends to be smaller for smaller partners. In line with Hypothesis 3b, knowledge loss is marginally lower in equity than in nonequity alliances. Knowledge loss increases marginally with alliance duration. In the TSLS regression, knowledge loss still increases with alliance duration. Relative size, however, is no longer significant. The coefficient for equity ( P < .05) provides support for Hypothesis 3b. Contrary to the prediction in Hypothesis 3a, knowledge protection marginally increases knowledge loss. Examining the causal model, a partner’s learning intent (.16) and ability (.15) both have substantial positive indirect effects on knowledge loss. While the direct coefficient for trust is insignificant, the total effect of trust on knowledge loss is still overwhelmingly negative ( .21) because of its indirect effect ( .27). Thus, the total negative effect of trust provides support for Hypothesis 3c, which predicted that higher trust in the partner would be associated with lower knowledge loss. The results of the OLS ( P < .001) and TSLS ( P < .001) regressions for satisfaction are similar. Knowledge protection is not significantly associated with satisfaction. It may
P.M. Norman / Journal of Business Research 57 (2004) 610–619
be that in some alliances protection increases satisfaction while in others it decreases satisfaction. Satisfaction is higher for firms involved in equity alliances and when partners are more trusted.
5. Discussion Learning intent, ability, and learning opportunities are all distinct, but related constructs. This study suggests that firms vary the amount of knowledge shared with partners in relation to the partners’ learning intent and the level of trust in the partner. Firms are more protective when partners intend to learn more, but reduce attempts to protect knowledge with trusted partners. The findings of this study are in line with previous research suggesting that trust plays a critical role in alliances. With more trusted partners, firms tend to acquire more knowledge and are also more willing to share information. Despite greater information sharing, they actually experienced less knowledge loss to these trusted partners. Finally, partners tend to be more satisfied with an alliance when the focal firm has higher trust in the partner. Interestingly, knowledge protection is not significantly associated with knowledge acquisition or satisfaction and is only marginally associated with knowledge loss. Furthermore, the relationship with knowledge loss is positive rather than negative. Some, but not all, of this may be attributed to the indirect effects of a partner’s learning intent and ability. Rather than acting to limit learning opportunities as has been proposed, it may be that sharing by the focal firm acts as a signal of their willingness to cooperate or commit to the relationship. If interpreted as a signal of cooperation and commitment, a partner may be less willing to appropriate knowledge and use it in ways that hurt the focal firm, even though they have the opportunity to do so. Restricting information sharing can be likened to having a safeguarding orientation (Madhok and Tallman, 1998), which may hinder the development of a mutually beneficial relationship. In the absence of a mutually beneficial relationship, the partner may try to find other ways to derive benefits from the alliance and be more, rather than less, likely to appropriate knowledge from the focal firm. Another possible explanation is that less sharing hinders the fulfillment of the alliance goals, in this case product development. When such goals are not fulfilled, the alliance provides less value to each partner than would otherwise be possible. The presence or absence of equity was used in this study as a proxy for learning opportunities. Firms involved in equity relationships experienced significantly lower levels of knowledge loss and higher levels of satisfaction as well as marginally greater knowledge acquisition. The dynamics of equity relationships may help to explain why knowledge loss is lower, despite the greater learning opportunities usually offered. In previous research, equity relationships have been used as a proxy for other constructs, including
617
trust (e.g., Gulati, 1995). If equity relationships engender more trust, then this would explain why equity relationships are associated with less knowledge loss. In this study, however, trust and equity were both measured and the two were found to be uncorrelated (r=.07). Rather than being associated with trust, equity may change the partners’ assessments of the costs and benefits associated with opportunistic behavior. Madhok and Tallman (1998) argue, for example, that partners may forego opportunities to act opportunistically for economic reasons, rather than because of trust. This study provides empirical evidence consistent with this argument. A firm’s investment in equity alliances is often substantial and potential losses from acting opportunistically may be greater in equity than in nonequity alliances. Thus, firms, even without higher levels of trust, may be more willing to forego opportunistic actions in equity than in nonequity alliances. The empirical tests used here cannot prove causality; thus, the cause and effect relationship between certain variables cannot be verified. The literature has argued both that trust leads to greater information sharing (i.e., less knowledge protection) and that greater information sharing leads to trust. Over time, their relationship is likely to be reciprocal and mutually reinforcing. The relationship between trust and satisfaction may follow similar logic. 5.1. Limitations and future research This study is based on survey responses from a single informant. Data was gathered to assess the respondents’ qualifications for providing this data. Respondents had, on average, interacted with the partner for 3.2 years, been employed by their firms for 8.9 years, and been employed in the industry for 18.7 years. Two actions were taken to reduce the potential problems resulting from common method variance. First, the items used to measure outcomes were placed at the end of the survey (Podsakoff and Organ, 1986). Only demographic information was gathered after the outcome measures. Second, a factor analysis of all variables produced seven factors with eigenvalues greater than one. The first factor accounted for only 24% of the variance, indicating that common method variance is unlikely to be a problem (Harman, 1967). Generalizing from the results of this study must be done with caution. First, this study examined only joint development alliances. Future studies should examine other types of alliances because the factors associated with learning in these alliances may differ. Second, firms may acquire knowledge in areas other than technology and management (e.g., production, distribution, and country-specific knowledge). Third, while firms in less technology intensive industries are likely to face many of the same concerns as high technology firms, differences must be considered. Finally, the sample size is small. Despite these limitations,
618
P.M. Norman / Journal of Business Research 57 (2004) 610–619
this study provides an early empirical test of important factors related to the alliance learning outcomes. Future studies should also consider the effects of different kinds of control mechanisms on learning. For example, studies may consider ways that a firm monitors its partner’s behavior. Research may also draw on the control literature to examine the effectiveness of various formal controls (e.g., output and process controls) and informal controls (e.g., social controls) in controlling partner opportunism. Geringer and Hebert (1989) identify a number of possible mechanisms.
Acknowledgements This research was supported by a research grant from the Cato Center for Applied Business Research at the KenanFlagler Business School, University of North Carolina at Chapel Hill and by a research sabbatical from the Hankamer School of Business, Baylor University. I would like to thank Rich Bettis, Hugh O’Neill, Bill Fischer, Anne Ilinitch, Mohan Tatikonda, the study participants, and three anonymous reviewers for their help.
References Calantone RJ, di Benedetto CA, Divine R. Organisational, technical and marketing antecedents for successful new product development. R&D Manage 1993;23(4):337 – 51. Chiles TH, McMackin JF. Integrating variable risk preferences, trust, and transaction cost economics. Acad Manage Rev 1996;21(1):73 – 99. Cohen WM, Levinthal DA. Absorptive capacity: a new perspective on learning and innovation. Adm Sci Q 1990;35(1):128 – 52. Crossan MM, Inkpen AC. The subtle art of learning through alliances. Bus Q 1995;60(2):68 – 78. Currall SC, Judge TA. Measuring trust between organizational boundary role persons. Org Behav Hum Decis Process 1995;64(2):151 – 70. Fiol CM. Squeezing harder doesn’t always work: continuing the search for consistency in innovation research. Acad Manage Rev 1996;21(4): 1012 – 21. Geletkanycz M. The salience of ‘culture’s consequences’: the effects of cultural values on top executive commitment to the status quo. Strateg Manage J 1997;18(8):615 – 34. Geringer JM, Hebert L. Control and performance in international joint ventures. J Int Bus Stud 1989;20(2):235 – 54. Glaister KW, Buckley PJ. Measures of performance in UK international alliances. Organ Stud 1998;19(1):89 – 118. Goes JB, Park SH. Interorganizational links and innovation: the case of hospital services. Acad Manage J 1997;40(3):673 – 96. Gulati R. Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances. Acad Manage J 1995;38(1):85 – 112. Gulati R, Khanna T, Nohria N. Unilateral commitments and the importance of process in alliances. Sloan Manage Rev 1994;35(3):61 – 9. Hamel G. Competition for competence and inter-partner learning with international strategic alliances. Strateg Manage J 1991;12:83 – 103 (Summer Special Issue). Hanushek EA, Jackson JE. Statistical methods for social scientists. New York: Academic Press, 1977. Harman HH. Modern factor analysis. Chicago: University of Chicago Press, 1967.
Hausler JH, Hohn H-W, Lutz S. Contingencies of innovative networks: a case study of successful R&D collaboration. Res Policy 1994;23(1): 47 – 66. Huber GP. Organizational learning: the contributing processes and the literatures. Organ Sci 1991;2(1):88 – 115. Inkpen, AC. Learning and collaboration: an examination of North American Japanese joint ventures. Doctoral dissertation. The University of Western Ontario; 1992. Inkpen AC, Dinur A. Knowledge management processes and international joint ventures. Organ Sci 1998;9(4):454 – 68. Inkpen AC, Li K-Q. Joint venture formation: planning and knowledgegathering for success. Organ Dyn 1999;27(4):33 – 47. Johnson JL, Cullen JB, Sakano T. Opportunistic tendencies in IJVs with the Japanese: the effects of culture, shared decision making, and relationship age. Int Exec 1996;38(1):79 – 94. Khanna T, Gulati R, Nohria N. The dynamics of learning alliances: competition, cooperation, and relative scope. Strateg Manage J 1998;19(3): 193 – 210. Koh J, Venkatraman N. Joint venture formations and stock market reactions: an assessment in the information technology sector. Acad Manage J 1991;34(4):869 – 92. Kumar S, Seth A. The design of coordination and control mechanisms for managing joint venture – parent relationships. Strateg Manage J 1998; 19(6):579 – 99. Lane PJ, Lubatkin M. Relative absorptive capacity and interorganizational learning. Strateg Manage J 1998;19(5):461 – 77. Larson A. Network dyads in entrepreneurial settings: a study of the governance of exchange relationships. Adm Sci Q 1992;37(1):76 – 104. Larsson R, Bengtsson L, Henriksson K, Sparks J. The interorganizational learning dilemma: collective knowledge development in strategic alliances. Organ Sci 1998;9(3):285 – 305. Leonard-Barton D. Wellsprings of knowledge: building and sustaining the sources of innovation. Boston: Harvard Business School Press; 1995. Lin X, Germain R. Predicting international joint venture interaction frequency in US – Chinese ventures. J Int Mark 1999;7(2):5 – 23. Lincoln JR, Ahmadjian CL, Mason E. Organizational learning and purchase – supply relations in Japan: Hitachi, Matsushita, and Toyota compared. Calif Manage Rev 1998;40(3):241 – 64. Lyles MA, Salk JE. Knowledge acquisition from foreign parents in international joint ventures: an empirical examination in the Hungarian context. J Int Bus Stud 1996;27(5):877 – 903. Madhok A, Tallman SB. Resources, transactions, and rents: managing value through interfirm collaborative relationships. Organ Sci 1998;9(3): 326 – 39. Mayer RC, Davis JH, Schoorman FD. An integrative model of organizational trust. Acad Manage J 1995;20(3):709 – 34. Millar J, Demaid A, Quintas P. Trans-organizational innovation: a framework for research. Technol Anal Strateg Manag 1997;9(4):399 – 418. Mohr J, Spekman R. Characteristics of partnership success: partnership attributes, communication behavior, and conflict resolution techniques. Strateg Manage J 1994;15(2):135 – 52. Moschandreas M. The role of opportunism in transaction cost economics. J Econ Issues 1997;31(1):39 – 57. Mowery DC, Oxley JE, Silverman BS. Strategic alliances and interfirm knowledge transfer. Strateg Manage J 1996;17:77 – 91 (Winter Special Issue). Nunnally J. Psychometric methods. New York: McGraw-Hill, 1967. Osborn RN, Hagedoorn J, Denekamp JG, Duysters G, Baughn CC. Embedded patterns of international alliance formation. Organ Stud 1998; 19(4):617 – 38. Oxley JE. Institutional environment and the mechanisms of governance: the impact of intellectual property protection of the structure of inter-firm alliances. J Econ Behav Organ 1999;38(3):283 – 309. Park SH. Managing an interorganizational network: a framework of the institutional mechanism for network control. Organ Stud 1996;17(5): 795 – 824.
P.M. Norman / Journal of Business Research 57 (2004) 610–619 Park SH, Kim D. Market valuation of joint ventures: joint venture characteristics and wealth gains. J Bus Venturing 1997;12(2):83 – 108. Park SH, Russo MV. When competition eclipses cooperation: an event history analysis of joint venture failure. Manage Sci 1996;42(6): 875 – 90. Park SH, Ungson GR. The effect of national culture, organizational complementarity, and economic motivation on joint venture dissolution. Acad Manage J 1997;40(2):279 – 307. Podsakoff PM, Organ DW. Self-reports in organizational research: problems and prospects. J Manage 1986;12(4):531 – 44. Richter F-J, Vettel K. Successful joint ventures in Japan: transferring knowledge through organizational learning. Long Range Plan 1995; 28(3):37 – 45. Ring PS, Van de Ven AH. Structuring cooperative relationships between organizations. Strateg Manage J 1992;13(7):483 – 98.
619
Rosenthal R, Rosnow RL. Essentials of behavioral research: methods and data. 2nd ed. New York: McGraw-Hill Publishing; 1991. Shenkar O, Li J. Knowledge search in international cooperative ventures. Organ Sci 1999;10(2):134 – 43. Simonin BL. The importance of collaborative know-how: an empirical test of the learning organization. Acad Manage J 1997;40(5):1150 – 74. Uzzi B. Social structure and competition in international networks: the paradox of embeddedness. Adm Sci Q 1997;42(1):35 – 67. von Krogh G, Roos J, Slocum K. An essay on corporate epistemology. Strateg Manage J 1994;15:53 – 71 (Summer Special Issue). Williamson OE. Comparative economic organization: the analysis of discrete structural alternatives. Adm Sci Q 1991;36(2):269 – 96.
Knowledge acquisition, knowledge loss, and satisfaction in high technology alliances Patricia M. Norman* Department of Management, Hankamer School of Business, Baylor University, P.O. Box 98006, Waco, TX 76798-9006, USA Received 5 January 2000; received in revised form 2 January 2002; accepted 22 May 2002
Abstract Drawing on the organizational learning and transaction cost economics (TCE) literature, this study examines how learning intent, opportunities to learn, and a firm’s ability to learn facilitate or hinder three alliance outcomes: knowledge acquisition by the focal firm, knowledge loss to the partner, and alliance satisfaction. The research model proposes that firms attempt to influence learning opportunities based on a partner’s intent and ability to learn as well as on the trust the firm has in the partner. A partner’s learning intent and ability are positively associated with the extent to which a firm protects its own firm-specific knowledge, but they only have significant effects on one alliance outcome, knowledge loss. With more trusted partners, firms are less protective of knowledge and tend to acquire more knowledge, lose less knowledge, and be more satisfied. Equity alliances are associated with lower levels of knowledge loss and higher levels of satisfaction. D 2002 Elsevier Inc. All rights reserved. Keywords: Alliance; High technology; Knowledge; Learning
Competition requires that firms continually acquire and develop new knowledge and skills. A great deal of attention has recently been focused on knowledge acquisition as an important outcome for firms engaged in strategic alliances (Lin and Germain, 1999). Because alliance success has been linked to learning and knowledge sharing (Crossan and Inkpen, 1995), partners have been urged to create an environment conducive to learning. Such an environment, however, can expose a firm’s critical knowledge and capabilities to a partner and may lead to imitation or appropriation. To prevent the loss of firm-specific knowledge, firms may try to prevent such losses by limiting a partner’s learning opportunities. These actions, however, may also reduce the firm’s own opportunities to learn and may affect alliance success. Lyles and Salk (1996) suggest that we need a greater understanding of what facilitates knowledge acquisition and skills development. Despite their important role in knowledge acquisition, however, few studies have empirically examined the learning outcomes of alliances. In one of the
*
Tel.: +1-254-710-6196; fax: +1-254-710-1093. E-mail address: [email protected] (P.M. Norman).
0148-2963/$ – see front matter D 2002 Elsevier Inc. All rights reserved. doi:10.1016/S0148-2963(02)00395-8
few studies to do so, Simonin (1997) investigated how collaborative experience and know-how affected knowledge acquisition. He suggests that ‘‘. . .firm- and alliance-specific variables such as strategic intent, transparency, organizational capabilities, and resource commitments’’ (p. 1170) must also be studied. In line with this recommendation, this paper presents the results of an exploratory study that investigates firm- and alliance-specific factors associated with firm actions to limit partner learning because of the threat of unwanted knowledge acquisition by the partner and, in turn, how these actions and other factors affect the firm’s own knowledge acquisition and its partners’ knowledge acquisition. In high technology product development alliances, partners are interdependent because each partner must contribute resources for the development to succeed (Gulati et al., 1994). Reciprocal information exchange is necessary to complete designs and to enable mutual adjustments (Osborn et al., 1998). Greater frequency and quality of information exchange can increase the innovativeness and quality of products designed, while lowering the costs of development (Larson, 1992). Such exchanges, however, create learning opportunities that enable a firm to appropriate knowledge from partners. Thus, the risk of knowledge appropriation is
P.M. Norman / Journal of Business Research 57 (2004) 610–619
particularly high in development alliances (Park and Kim, 1997) and these alliances are the context used for this study.
1. Theoretical background The model developed in this paper relies heavily on organizational learning, which stresses the importance of knowledge acquisition, and transaction cost economics (TCE), which helps to explain choices about structuring and controlling alliances. 1.1. Organizational learning Huber (1991, p. 89) takes a behavioral perspective on organizational learning and suggests that ‘‘an entity learns if, through its processing of information, the range of potential behaviors is changed.’’ This paper focuses on knowledge acquisition, one of four organizational learning processes, which occurs when any part of a firm gains knowledge that is recognized as potentially useful (Huber, 1991). The ability of firms to acquire and exploit knowledge has been stressed by organizational learning scholars (e.g., Cohen and Levinthal, 1990; Huber, 1991) and has been liked to a firm’s ability to innovate (Fiol, 1996). Alliance partners are a particularly important source of new, external knowledge (Inkpen and Dinur, 1998). Firms have different reasons for and goals associated with alliances. When learning is an explicit and primary goal, acquiring and internalizing skills is a key measure of success (Hamel, 1991). Even when learning is not an explicit goal, however, firms can learn from partners and this learning can enhance their competitive abilities (Crossan and Inkpen, 1995). Therefore, learning is a critical issue regardless of the goals of an alliance. Knowledge is more easily transferred between alliance partners than through market mechanisms (Shenkar and Li, 1999) because learning is a socially embedded process and requires connections through which individuals can share knowledge (von Krogh et al., 1994). With open communication and rich knowledge sharing, alliances become a potentially effective venue for knowledge acquisition (Lincoln et al., 1998). Hamel (1991) suggests that the amount of learning from a partner is influenced by the receiving firm’s intent to learn and its receptivity as well as by the transparency of the partner from whom the knowledge is acquired. Receptivity depends, in part, on exposure to knowledge and on the skills and ability of the acquiring firm. Transparency is influenced by the design of interfaces, the structure of joint tasks, and how protective the partner is of its knowledge. Each firm in an alliance may be simultaneously trying to gain knowledge from its partner, to internalize this knowledge, and then use it to further develop its own competitive advantage (Richter and Vettel, 1995). While some learning is needed to accomplish the alliance goals, too much
611
learning by a partner can strip a firm of critical knowledge. One way to counter this danger is to manage the alliance structure and processes in an effort to control knowledge transfers (Kumar and Seth, 1998). However, limiting communication and information exchange may inhibit learning necessary to meet alliance objectives (Millar et al., 1997). Learning can result in either common or private benefits (Khanna et al., 1998). Common benefits accrue to all partners in an alliance when learning is applied to activities within the alliance. Private benefits, on the other hand, accrue to only one partner when that partner uses what is learned within the alliance or from an alliance partner to generate rents in activities outside of an alliance. While both types of benefits are important, this study examines only private benefits. 1.2. TCE TCE provides a theoretical grounding for the circumstances under which firms will attempt to limit a partner’s learning opportunities. TCE was traditionally concerned with choices about whether activities were performed within a firm’s boundaries or in the marketplace. More recently, TCE has been applied to alliances, specifically how to structure relationships to try to mitigate the transaction hazards that are present in a relationship. Opportunism, which occurs when one party acts in its own self-interest and does so in a deceitful manner (Moschandreas, 1997), is central to a discussion of transaction hazards. While not all parties will to act opportunistically, uncertainty and bounded rationality prevent a decision-maker from knowing a priori which parties will act opportunistically (Williamson, 1991). Thus, mechanisms will be designed to prevent losses from potential opportunism. Recent criticisms of TCE have questioned the assumption that individuals are primarily self-interested and other motives such as altruism have been noted (e.g., Moschandreas, 1997). In this vein, trust has also received increasing attention. The melding of trust with TCE can provide a more complete and accurate picture of how transactions are structured. Trust is the willingness to be vulnerable to another party when their behavior cannot be controlled (Mayer et al., 1995). This definition implies that a trusted party has the opportunity to act opportunistically, but the trustor believes that the party will not act in such a manner. Thus, trust can reduce transaction costs and the need for protective mechanisms (Ring and Van de Ven, 1992) and encourage beneficial behaviors such as more accurate and comprehensive information exchange (Chiles and McMackin, 1996), greater communication (Currall and Judge, 1995), and fewer actions to safeguard knowledge and monitor partner behavior (Inkpen and Li, 1999). In relationships lacking in trust, structural mechanisms are used to either limit the opportunity that a partner has to act opportunistically or to deter opportunistic behavior through pen-
612
P.M. Norman / Journal of Business Research 57 (2004) 610–619
alties that are so severe that they discourage self-interested behavior.
acquisition (Inkpen and Dinur, 1998) and less likely to take actions to appropriate the focal firm’s knowledge. This leads to the following hypothesis:
2. Research model and hypotheses
Hypothesis 1a: The higher the partner’s learning intent, the higher the knowledge protection by the focal firm.
Given the importance of learning and the dangers of knowledge appropriation, this paper develops a model of factors that influence learning outcomes. It is based, in part, on the factors identified by Hamel (1991)—intent to learn, learning opportunities, and ability to learn. Lack of intent, lack of opportunities to learn resulting from a lack of information sharing, and an inability to learn all hinder knowledge acquisition (Larsson et al., 1998). In addition, the model integrates trust as a factor facilitating knowledge transfers. The research model tested examines alliance learning outcomes from the perspective of one party, the focal firm, because the focal firm and its partner enjoy different private benefits. Specifically, the amount of knowledge acquired by the focal firm and the extent to which the focal firm’s competitively valuable firm-specific knowledge is lost to the partner are assessed. While specific hypotheses are not proposed, satisfaction is examined to assess the impact of the model variables on a more general assessment of alliance performance. The model proposes both direct and indirect effects. Specifically, certain factors—partner’s learning intent, partner ability, and trust—are hypothesized to be associated with the amount of knowledge protection by the focal firm and to indirectly affect alliance outcomes through their influence on knowledge protection. Learning opportunities, represented by lower knowledge protection and the use of equity, and trust are expected to directly influence alliance outcomes. 2.1. Factors associated with knowledge protection Firms can structure relationships to protect against the loss of firm-specific knowledge to alliance partners (Kumar and Seth, 1998) by restricting not only the quantity, but also the depth and quality of information that is shared (Park, 1996). This study proposes that firms will attempt to protect knowledge from partners who may learn in ways that are detrimental to the firm. Specifically, firms will be more guarded with partners who have higher learning intents and greater abilities to appropriate knowledge, but will be more open with trusted partners. 2.1.1. Learning intent The perceived partner’s learning intent is the extent to which the focal firm believes that the partner is focused on learning during the alliance. Because not all firms are equally concerned with knowledge acquisition (Richter and Vettel, 1995), learning first requires that a firm have an intent to learn (Hamel, 1991). Without this intent, a partner is less likely to commit resources to knowledge
2.1.2. Ability to learn When trying to appropriate knowledge from an alliance partner, a firm’s absorptive capacity (Cohen and Levinthal, 1990), which is its ability to ‘‘identify, appreciate, and then assimilate such assets’’ (Park and Russo, 1996, p. 878), is critical. Absorptive capacity has been linked to similarity of resources and knowledge bases (Lane and Lubatkin, 1998; Mowery et al., 1996) and similarity of organizational structures and processes (Lane and Lubatkin, 1998) between partner firms. This study uses the similarity of resources and knowledge bases, which allows a partner to more readily understand and appropriate the firm’s knowledge. While some overlap in knowledge is needed to ensure that partners understand and can effectively combine their knowledge, a focal firm is likely to limit the learning opportunities of a partner who has greater ability to take advantage of these opportunities because such a partner is more dangerous than a partner with lower ability. Hypothesis 1b: The higher the partner’s ability to learn, the higher the knowledge protection by the focal firm. 2.1.3. Trust Trust among alliance partners, especially in technology collaborations, is critical (Hausler et al., 1994). Trust reduces the fear of opportunism by partners, which should reduce transaction costs (Williamson, 1991). With trust, the need for monitoring and other control mechanisms is reduced (Inkpen and Li, 1999). In addition, trust encourages behaviors such as open communication and the willingness to share information (Currall and Judge, 1995). Therefore, firms should be less protective with more trusted partners. Hypothesis 1c: The higher the focal firm’s trust in the partner, the lower the knowledge protection by the focal firm. 2.2. Alliance outcomes Hypotheses are now developed for how learning opportunities—lower levels of knowledge protection and the use of equity—and trust affect knowledge acquisition and loss. 2.2.1. Knowledge acquisition Knowledge acquisition refers to skills learned and knowledge acquired by the focal firm from a partner during an alliance.
P.M. Norman / Journal of Business Research 57 (2004) 610–619
2.2.2. Knowledge protection Regular contact between partner employees facilitates the sharing of information, which increases the chances that knowledge will be spread (Goes and Park, 1997; Lincoln et al., 1998). While limitations on sharing may prevent a partner from acquiring knowledge from the focal firm, it can be self-defeating because it also hinders the ability of the focal firm to learn. Partners often respond to each other’s limiting of information sharing by further reducing their own sharing (Larsson et al., 1998), an action that inhibits knowledge acquisition by the focal firm. Therefore, the following hypothesis proposes that increasing knowledge protection will decrease knowledge acquisition. Hypothesis 2a: The higher the level of knowledge protection, the lower the focal firm’s knowledge acquisition. 2.2.3. Equity Conceptually, equity-based alliances are thought to facilitate the transfer of knowledge between partners (Shenkar and Li, 1999) because partners usually have more frequent interactions and commit greater resources to joint efforts (Uzzi, 1997). Empirically, partners in equity-based alliances have been found to increase their patent cross-citation rates, a proxy for learning, to a greater degree than partners in nonequity alliances (Mowery et al., 1996). Thus, equity alliances are expected to provide greater learning opportunities and greater knowledge acquisition than nonequity alliances, as proposed in the following hypothesis. Hypothesis 2b: The focal firm will experience higher knowledge acquisition in equity alliances than nonequity alliances. 2.2.4. Trust Earlier, it was argued that high trust environments should provide greater learning opportunities because firms are more willing to exchange information with trusted partners. Because of these greater learning opportunities, a firm should be able to learn more from a trusted partner. Therefore, the following hypothesis is proposed: Hypothesis 2c: The higher the focal firm’s trust in the partner firm, the higher the focal firm’s knowledge acquisition. 2.3. Knowledge loss Some, but not all, learning by a partner represents a danger to the focal firm. Partner learning is a danger when it results in the loss of competitively important knowledge. To assess the competitive impact on the focal firm, knowledge loss to a partner that is detrimental to the focal firm and not all forms of partner learning are examined here.
613
2.3.1. Knowledge protection Firm actions to limit knowledge sharing and interactions should reduce the opportunities for partner learning. Thus, knowledge protection should make it harder for a partner to duplicate critical skills and capabilities of the focal firm. In line with this logic, the following hypothesis is proposed: Hypothesis 3a: The higher the level of knowledge protection, the lower the focal firm’s knowledge loss. 2.3.2. Equity Because greater learning opportunities are provided by equity alliances, it may seem that firms engaged in such alliances should experience greater knowledge loss. Yet, equity arrangements may actually reduce the likelihood of knowledge loss to a partner. Equity alliances often rely on more formal mechanisms that may provide some protection against opportunism (Oxley, 1999). In addition, equity investments act as ‘‘hostages,’’ aligning the interests of the parties and reducing the likelihood that a partner will act opportunistically (Gulati, 1995) because such behavior will reduce return on the partner’s investment. Thus, even though greater learning opportunities exist, partners in equity alliances are less likely use knowledge in ways that are detrimental to a firm than partners in nonequity alliances. Hypothesis 3b: The focal firm will experience lower knowledge loss in equity alliances than nonequity alliances. 2.3.3. Trust A trusted partner is not expected to act opportunistically. Thus, even when a trusted partner is presented with more learning opportunities and learns a great deal, it is not likely to use the acquired knowledge in ways that are detrimental to the focal firm. Therefore, knowledge loss should be lower with more trusted partners. Hypothesis 3c: The higher the focal firm’s trust in the partner firm, the lower the focal firm’s knowledge loss. 3. Methods 3.1. Sample Public and private firms specializing in microelectronics, computers, and telecommunications were surveyed. Respondents chose an alliance that involved joint development or modification of a product or service and filled out the survey with reference to a specific partner. Each firm in the sample reported on one or two alliances. Initially, 135 CEOs were sent a letter explaining the project and requesting their firms’ participation. Ten of these firms participated, completing surveys on 15 alliances. Because of the limited success of the initial approach, press releases were examined for firms identified as computer,
614
P.M. Norman / Journal of Business Research 57 (2004) 610–619
telecommunications, or microelectronics firms by Hoover’s online. Surveys were mailed to 252 firms with press releases about joint development alliances. In almost all cases, the points of contact were senior managers involved in the alliance. In an effort to increase the response rate, respondents were offered a summary of the study findings. In addition, follow-up phone calls were made to all potential respondents who had not returned the surveys after three weeks. Thirty-eight firms completed surveys on a total of 47 alliances. The overall response rate was 13.6%, with a rate
of 7.4% for the first method and 15.1% for the second method. This response rate is similar to others who have surveyed top management (e.g., Geletkanycz, 1997) and to those who have studied similar issues in alliances (e.g., Johnson et al., 1996). Four surveys did not contain complete data. In total, 58 useable surveys were received. T-tests compared the data gathered by the two different collection methods. Alliances from the first collection method had greater duration and greater knowledge acquisition. When both alliance duration and a categorical variable
Table 1 Factor analysis of multi-item variablesa Items
Our firm severely restricts partner access to our strategic information. Our firm severely restricts partner access to our marketing plans and information. Our firm severely restricts partner access to our technical information. The alliance has highly structured channels of communication between partners. Working level employees from our firm and the partner never communicate. We trust that our partner’s decisions will be beneficial to the alliance. There is a high level of trust in the working relationship with our partner. We can rely on our partner to abide by the alliance agreement. We trust that our partner’s decisions will be beneficial to our firm. As a result of this alliance, we have improved existing management skills. As a result of this alliance, we have developed new management skills. As a result of this alliance, we have developed new technical skills. As a result of this alliance, we have improved existing technical skills. Our firm is satisfied with the performance of the alliance. Our partner is satisfied with the performance of the alliance. As a result of this alliance, we lost valuable skills and knowledge to our partner. One of our partner’s objectives in forming the alliance was to learn about our management techniques. One of our partner’s objectives in forming the alliance was to learn about an unfamiliar market. One of our partner’s objectives in forming the alliance was to learn about our technical information. Percentage of variance explained Cronbach’s a
Factor 1
Factor 2
Factor 3
Factor 4
Factor 5
Knowledge protection
Trust
Knowledge acquisition
Satisfaction
Learning intent
0.823
0.069
0.080
0.014
0.077
0.796
0.191
0.082
0.080
0.110
0.763
0.023
0.028
0.214
0.070
0.737
0.308
0.014
0.279
0.038
0.455
0.302
0.235
0.203
0.217
0.098
0.863
0.123
0.318
0.082
0.126
0.825
0.196
0.283
0.005
0.161
0.759
0.061
0.065
0.021
0.342
0.632
0.143
0.191
0.126
0.060
0.041
0.900
0.022
0.243
0.116
0.037
0.801
0.096
0.191
0.064
0.249
0.618
0.122
0.186
0.179
0.251
0.550
0.131
0.082
0.199
0.232
0.098
0.842
0.054
0.101
0.362
0.004
0.787
0.047
0.086
0.168
0.043
0.364
0.333
0.212
0.054
0.137
0.024
0.750
0.155
0.018
0.024
0.128
0.558
0.059
0.014
0.131
0.021
0.435
0.258 .85
0.256 .89
0.200 .82
0.165 .85
0.121 .57
a Items were measured on a seven-point scale, anchored by 1 = strongly disagree and 7 = strongly agree. Items for knowledge protection (Calantone et al., 1993), trust (Inkpen, 1992; Mohr and Spekman, 1994), satisfaction (Glaister and Buckley, 1998) and learning intent (Inkpen, 1992) were modified for the study context. Items for knowledge acquisition and knowledge loss were developed for this study.
P.M. Norman / Journal of Business Research 57 (2004) 610–619
representing collection method were regressed against knowledge acquisition, only alliance duration was significant. Alliance duration is included as a control in the models tested. Data was gathered from Hoover’s online for all firms that were mailed surveys to test for nonresponse bias. T-tests comparing the mean sales, number of employees, and firm age for responding and nonresponding firms revealed no significant differences, indicating that response bias is not a problem. 3.2. Measures A survey instrument was constructed using a panel to provide subjective evaluations of content validity (Rosenthal and Rosnow, 1991). Ten judges reviewed the initial survey items, assessed how well the questions covered each construct, and suggested changes. Once the panel determined that adequate content validity had been achieved, two alliance managers completed and critiqued the survey. Based on their comments, the survey was shortened. Factor analysis, using a varimax rotation, was used to assess the discriminate validity of multiitem measures. Table 1 shows the results of this analysis, which included the item for knowledge loss, along with the Cronbach’s a for each measure. All items loaded on the expected factors with loadings above .40. With the exception of partner learning intent, each measure had an a above .80. This new measure’s a (.57) was within the acceptable range suggested by Nunnally (1967) for early stages of research. The other measures did not use multiple items. Respondents were asked to indicate in which of 15 areas their firm had made contributions to the alliance. For each area, respondents used a seven-point scale to assess the similarity of the focal firm and partner. Partner ability was the average similarity for those areas in which the focal firm made contributions to the alliance. Equity was a dummy variable indicating whether or not the alliance involved equity.
615
Two control variables, both transformed using log transformations to correct for nonnormal distributions, are included in the models. Alliance duration, measured by the number of months the alliance had been in existence, is controlled because assessments of learning and satisfaction may differ over time. Relative size was measured by the ratio of the number of focal firm employees to the number of partner firm employees. After the transformation, negative values for relative size indicate that the focal firm is smaller than its partner and positive values indicate that the focal firm is larger than its partner. 3.3. Data analysis Because a structural model with both direct and indirect effects on alliance outcomes is proposed, a two-stage least squares analysis (TSLS) is used. While the results for the ordinary least squares (OLS) regressions are shown, the TSLS allows assessment of both direct and indirect effects. TSLS, rather than another method to estimate structural equations, is used because its estimators are the most robust and appropriate for small sample sizes (Hanushek and Jackson, 1977). Each of the alliance outcomes is the dependent variable in separate models.
4. Findings The zero-order correlation matrix is shown in Table 2. Consistent with suggestions that partners do not learn at the same rate (Hamel, 1991), knowledge acquisition and knowledge loss are uncorrelated (r = .05). Thus, alliances can be tools for learning without endangering critical firm knowledge. Table 3 shows the regression results for each outcome variable. The first stage regression for knowledge protection, used to test 1a, 1b, and 1c, is identical for each model. As predicted in Hypothesis 1a, focal firms are more protective
Table 2 Correlation matrix Variable
Mean
S.D.
1
2
3
4
5
6
7
8
9
10
1. Partner learning intent 2. Knowledge protection 3. Equity (0,1) 4. Partner ability 5. Trust 6. Relative size 7. Alliance duration 8. Knowledge acquisition 9. Knowledge loss 10. Satisfaction
3.49 4.64 0.24 3.56 5.29 0.26 2.67 4.28 2.12 5.30
1.27 1.29 0.43 1.11 1.02 3.22 1.09 1.31 1.14 1.16
1.00 .28 * .18 .01 .08 .07 .30 * .17 .25y .14
1.00 .21 .32 * .44** * .05 .14 .05 .24y .34* *
1.00 .11 .07 .17 .23y .30 * .13 .20
1.00 .33 * .38* * .15 .01 .14 .14
1.00 .16 .18 .28 * .28 * .54** *
1.00 .05 .14 .23y .00
1.00 .40* * .26y .18
1.00 .05 .13
1.00 .39* *
1.00
n = 58. * P < .05. ** P < .01. *** P < .001. y P < .10.
616
P.M. Norman / Journal of Business Research 57 (2004) 610–619
Table 3 Regression results Independent variables
First stage for TSLS
Model 1 knowledge acquisition
Knowledge protection OLS Intercept 3.95 * * (1.21) Predictor variables intent Partner learning intent 0.26 * (0.12) Opportunity Knowledge protection Equity Trust 0.44 * * (0.15) Ability Partner ability 0.24 (0.14) Other Alliance duration Relative size F-value 7.243* * * R2 .29 Adjusted R2 .25
0.93 (1.20)
0.17 (0.13) 0.54 (0.35) 0.59* * * (0.16)
TSLS
Model 2 knowledge loss
Model 3 satisfaction
OLS
OLS
2.31y (0.90)
0.10 (2.18)
0.04 (0.31) 0.62 (0.40) 0.51 * (0.23)
0.52* * * (0.14) 0.52* * * (0.14) 0.11 * (0.05) 0.10 * (0.05) 6.83* * * 6.34* * * .40 .38 .34 .32
0.14 (0.12) 0.67y (0.34) 0.22 (0.16)
0.26y (0.13) 0.10 * (0.04) 3.57 * * .26 .18
TSLS 0.72 (2.38)
3.45 * * (1.09)
TSLS 3.14 (1.95)
0.62y (0.34) 0.16 (0.11) 0.12 (0.28) 0.99 * (0.44) 0.69 * (0.32) 0.66y (0.36) 0.06 (0.26) 0.49 * * (0.15) 0.52 * (0.21)
0.26y (0.15) 0.08 (0.05) 3.21 * .24 .16
0.14 (0.12) 0.04 (0.04) 6.17* * * .37 .31
0.14 (0.12) 0.04 (0.04) 5.77* * * .36 .29
* P < .05. ** P < .01. *** P < .001. y P < .10.
when partners have higher learning intents. With more trusted partners, firms tend to be less protective, supporting Hypothesis 1c. Hypothesis 1b is not supported; the coefficient for partner ability is not significant. The results for the OLS ( P < .001) and TSLS ( P < .001) regressions for knowledge acquisition are similar. Hypotheses 2a and 2b are not supported. Knowledge protection has no effect on and equity does not significantly increase knowledge acquisition. Hypothesis 2c, which predicted that higher trust would be associated with knowledge acquisition, is supported in both regressions. Both control variables are significant. Knowledge acquisition increases with alliance age. This seems logical because learning takes time and new skills and competencies develop slowly (LeonardBarton, 1995). As alliances age, a firm becomes more familiar with its partner’s skills, which is likely to ease knowledge acquisition. Over time, the respondents may also become more aware of changes in existing skills and the development of new skills. Firms that are larger than their partners tend to acquire less knowledge than firms that are smaller than their partners. This is an interesting outcome because previous literature has suggested (e.g., Park and Ungson, 1997) that smaller firms should be more concerned with the loss of their skills. While smaller firms may be more worried about such losses, they may actually learn more in alliances because they often start with a limited skill base and are exposed to a much broader skill base when they partner with larger firms. Although smaller partners are often sought because of their technological capabilities (Koh and Venkatraman, 1991), larger firms may not interested in or may be unable to internalize the smaller firm’s capabilities. The indirect effects of learning intent (.01), partner ability (.01), and trust ( .02) on knowledge acquisition
are quite small because of the lack of a relationship between knowledge protection and knowledge acquisition. This may be because this study measured knowledge protection as the focal firm’s sharing of information and communication with the partner. What is more likely to affect knowledge acquisition by the focal firm, however, is what information the partner is willing to share with the focal firm. Similarly, partner ability measured resource overlaps only in areas where the focal firm contributed resources. A more appropriate measure for knowledge acquisition may be resource overlaps in areas of the partner’s contributions. The OLS ( P < .001) and TSLS ( P < .05) models for knowledge loss differ. In the OLS model, knowledge loss tends to be smaller for smaller partners. In line with Hypothesis 3b, knowledge loss is marginally lower in equity than in nonequity alliances. Knowledge loss increases marginally with alliance duration. In the TSLS regression, knowledge loss still increases with alliance duration. Relative size, however, is no longer significant. The coefficient for equity ( P < .05) provides support for Hypothesis 3b. Contrary to the prediction in Hypothesis 3a, knowledge protection marginally increases knowledge loss. Examining the causal model, a partner’s learning intent (.16) and ability (.15) both have substantial positive indirect effects on knowledge loss. While the direct coefficient for trust is insignificant, the total effect of trust on knowledge loss is still overwhelmingly negative ( .21) because of its indirect effect ( .27). Thus, the total negative effect of trust provides support for Hypothesis 3c, which predicted that higher trust in the partner would be associated with lower knowledge loss. The results of the OLS ( P < .001) and TSLS ( P < .001) regressions for satisfaction are similar. Knowledge protection is not significantly associated with satisfaction. It may
P.M. Norman / Journal of Business Research 57 (2004) 610–619
be that in some alliances protection increases satisfaction while in others it decreases satisfaction. Satisfaction is higher for firms involved in equity alliances and when partners are more trusted.
5. Discussion Learning intent, ability, and learning opportunities are all distinct, but related constructs. This study suggests that firms vary the amount of knowledge shared with partners in relation to the partners’ learning intent and the level of trust in the partner. Firms are more protective when partners intend to learn more, but reduce attempts to protect knowledge with trusted partners. The findings of this study are in line with previous research suggesting that trust plays a critical role in alliances. With more trusted partners, firms tend to acquire more knowledge and are also more willing to share information. Despite greater information sharing, they actually experienced less knowledge loss to these trusted partners. Finally, partners tend to be more satisfied with an alliance when the focal firm has higher trust in the partner. Interestingly, knowledge protection is not significantly associated with knowledge acquisition or satisfaction and is only marginally associated with knowledge loss. Furthermore, the relationship with knowledge loss is positive rather than negative. Some, but not all, of this may be attributed to the indirect effects of a partner’s learning intent and ability. Rather than acting to limit learning opportunities as has been proposed, it may be that sharing by the focal firm acts as a signal of their willingness to cooperate or commit to the relationship. If interpreted as a signal of cooperation and commitment, a partner may be less willing to appropriate knowledge and use it in ways that hurt the focal firm, even though they have the opportunity to do so. Restricting information sharing can be likened to having a safeguarding orientation (Madhok and Tallman, 1998), which may hinder the development of a mutually beneficial relationship. In the absence of a mutually beneficial relationship, the partner may try to find other ways to derive benefits from the alliance and be more, rather than less, likely to appropriate knowledge from the focal firm. Another possible explanation is that less sharing hinders the fulfillment of the alliance goals, in this case product development. When such goals are not fulfilled, the alliance provides less value to each partner than would otherwise be possible. The presence or absence of equity was used in this study as a proxy for learning opportunities. Firms involved in equity relationships experienced significantly lower levels of knowledge loss and higher levels of satisfaction as well as marginally greater knowledge acquisition. The dynamics of equity relationships may help to explain why knowledge loss is lower, despite the greater learning opportunities usually offered. In previous research, equity relationships have been used as a proxy for other constructs, including
617
trust (e.g., Gulati, 1995). If equity relationships engender more trust, then this would explain why equity relationships are associated with less knowledge loss. In this study, however, trust and equity were both measured and the two were found to be uncorrelated (r=.07). Rather than being associated with trust, equity may change the partners’ assessments of the costs and benefits associated with opportunistic behavior. Madhok and Tallman (1998) argue, for example, that partners may forego opportunities to act opportunistically for economic reasons, rather than because of trust. This study provides empirical evidence consistent with this argument. A firm’s investment in equity alliances is often substantial and potential losses from acting opportunistically may be greater in equity than in nonequity alliances. Thus, firms, even without higher levels of trust, may be more willing to forego opportunistic actions in equity than in nonequity alliances. The empirical tests used here cannot prove causality; thus, the cause and effect relationship between certain variables cannot be verified. The literature has argued both that trust leads to greater information sharing (i.e., less knowledge protection) and that greater information sharing leads to trust. Over time, their relationship is likely to be reciprocal and mutually reinforcing. The relationship between trust and satisfaction may follow similar logic. 5.1. Limitations and future research This study is based on survey responses from a single informant. Data was gathered to assess the respondents’ qualifications for providing this data. Respondents had, on average, interacted with the partner for 3.2 years, been employed by their firms for 8.9 years, and been employed in the industry for 18.7 years. Two actions were taken to reduce the potential problems resulting from common method variance. First, the items used to measure outcomes were placed at the end of the survey (Podsakoff and Organ, 1986). Only demographic information was gathered after the outcome measures. Second, a factor analysis of all variables produced seven factors with eigenvalues greater than one. The first factor accounted for only 24% of the variance, indicating that common method variance is unlikely to be a problem (Harman, 1967). Generalizing from the results of this study must be done with caution. First, this study examined only joint development alliances. Future studies should examine other types of alliances because the factors associated with learning in these alliances may differ. Second, firms may acquire knowledge in areas other than technology and management (e.g., production, distribution, and country-specific knowledge). Third, while firms in less technology intensive industries are likely to face many of the same concerns as high technology firms, differences must be considered. Finally, the sample size is small. Despite these limitations,
618
P.M. Norman / Journal of Business Research 57 (2004) 610–619
this study provides an early empirical test of important factors related to the alliance learning outcomes. Future studies should also consider the effects of different kinds of control mechanisms on learning. For example, studies may consider ways that a firm monitors its partner’s behavior. Research may also draw on the control literature to examine the effectiveness of various formal controls (e.g., output and process controls) and informal controls (e.g., social controls) in controlling partner opportunism. Geringer and Hebert (1989) identify a number of possible mechanisms.
Acknowledgements This research was supported by a research grant from the Cato Center for Applied Business Research at the KenanFlagler Business School, University of North Carolina at Chapel Hill and by a research sabbatical from the Hankamer School of Business, Baylor University. I would like to thank Rich Bettis, Hugh O’Neill, Bill Fischer, Anne Ilinitch, Mohan Tatikonda, the study participants, and three anonymous reviewers for their help.
References Calantone RJ, di Benedetto CA, Divine R. Organisational, technical and marketing antecedents for successful new product development. R&D Manage 1993;23(4):337 – 51. Chiles TH, McMackin JF. Integrating variable risk preferences, trust, and transaction cost economics. Acad Manage Rev 1996;21(1):73 – 99. Cohen WM, Levinthal DA. Absorptive capacity: a new perspective on learning and innovation. Adm Sci Q 1990;35(1):128 – 52. Crossan MM, Inkpen AC. The subtle art of learning through alliances. Bus Q 1995;60(2):68 – 78. Currall SC, Judge TA. Measuring trust between organizational boundary role persons. Org Behav Hum Decis Process 1995;64(2):151 – 70. Fiol CM. Squeezing harder doesn’t always work: continuing the search for consistency in innovation research. Acad Manage Rev 1996;21(4): 1012 – 21. Geletkanycz M. The salience of ‘culture’s consequences’: the effects of cultural values on top executive commitment to the status quo. Strateg Manage J 1997;18(8):615 – 34. Geringer JM, Hebert L. Control and performance in international joint ventures. J Int Bus Stud 1989;20(2):235 – 54. Glaister KW, Buckley PJ. Measures of performance in UK international alliances. Organ Stud 1998;19(1):89 – 118. Goes JB, Park SH. Interorganizational links and innovation: the case of hospital services. Acad Manage J 1997;40(3):673 – 96. Gulati R. Does familiarity breed trust? The implications of repeated ties for contractual choice in alliances. Acad Manage J 1995;38(1):85 – 112. Gulati R, Khanna T, Nohria N. Unilateral commitments and the importance of process in alliances. Sloan Manage Rev 1994;35(3):61 – 9. Hamel G. Competition for competence and inter-partner learning with international strategic alliances. Strateg Manage J 1991;12:83 – 103 (Summer Special Issue). Hanushek EA, Jackson JE. Statistical methods for social scientists. New York: Academic Press, 1977. Harman HH. Modern factor analysis. Chicago: University of Chicago Press, 1967.
Hausler JH, Hohn H-W, Lutz S. Contingencies of innovative networks: a case study of successful R&D collaboration. Res Policy 1994;23(1): 47 – 66. Huber GP. Organizational learning: the contributing processes and the literatures. Organ Sci 1991;2(1):88 – 115. Inkpen, AC. Learning and collaboration: an examination of North American Japanese joint ventures. Doctoral dissertation. The University of Western Ontario; 1992. Inkpen AC, Dinur A. Knowledge management processes and international joint ventures. Organ Sci 1998;9(4):454 – 68. Inkpen AC, Li K-Q. Joint venture formation: planning and knowledgegathering for success. Organ Dyn 1999;27(4):33 – 47. Johnson JL, Cullen JB, Sakano T. Opportunistic tendencies in IJVs with the Japanese: the effects of culture, shared decision making, and relationship age. Int Exec 1996;38(1):79 – 94. Khanna T, Gulati R, Nohria N. The dynamics of learning alliances: competition, cooperation, and relative scope. Strateg Manage J 1998;19(3): 193 – 210. Koh J, Venkatraman N. Joint venture formations and stock market reactions: an assessment in the information technology sector. Acad Manage J 1991;34(4):869 – 92. Kumar S, Seth A. The design of coordination and control mechanisms for managing joint venture – parent relationships. Strateg Manage J 1998; 19(6):579 – 99. Lane PJ, Lubatkin M. Relative absorptive capacity and interorganizational learning. Strateg Manage J 1998;19(5):461 – 77. Larson A. Network dyads in entrepreneurial settings: a study of the governance of exchange relationships. Adm Sci Q 1992;37(1):76 – 104. Larsson R, Bengtsson L, Henriksson K, Sparks J. The interorganizational learning dilemma: collective knowledge development in strategic alliances. Organ Sci 1998;9(3):285 – 305. Leonard-Barton D. Wellsprings of knowledge: building and sustaining the sources of innovation. Boston: Harvard Business School Press; 1995. Lin X, Germain R. Predicting international joint venture interaction frequency in US – Chinese ventures. J Int Mark 1999;7(2):5 – 23. Lincoln JR, Ahmadjian CL, Mason E. Organizational learning and purchase – supply relations in Japan: Hitachi, Matsushita, and Toyota compared. Calif Manage Rev 1998;40(3):241 – 64. Lyles MA, Salk JE. Knowledge acquisition from foreign parents in international joint ventures: an empirical examination in the Hungarian context. J Int Bus Stud 1996;27(5):877 – 903. Madhok A, Tallman SB. Resources, transactions, and rents: managing value through interfirm collaborative relationships. Organ Sci 1998;9(3): 326 – 39. Mayer RC, Davis JH, Schoorman FD. An integrative model of organizational trust. Acad Manage J 1995;20(3):709 – 34. Millar J, Demaid A, Quintas P. Trans-organizational innovation: a framework for research. Technol Anal Strateg Manag 1997;9(4):399 – 418. Mohr J, Spekman R. Characteristics of partnership success: partnership attributes, communication behavior, and conflict resolution techniques. Strateg Manage J 1994;15(2):135 – 52. Moschandreas M. The role of opportunism in transaction cost economics. J Econ Issues 1997;31(1):39 – 57. Mowery DC, Oxley JE, Silverman BS. Strategic alliances and interfirm knowledge transfer. Strateg Manage J 1996;17:77 – 91 (Winter Special Issue). Nunnally J. Psychometric methods. New York: McGraw-Hill, 1967. Osborn RN, Hagedoorn J, Denekamp JG, Duysters G, Baughn CC. Embedded patterns of international alliance formation. Organ Stud 1998; 19(4):617 – 38. Oxley JE. Institutional environment and the mechanisms of governance: the impact of intellectual property protection of the structure of inter-firm alliances. J Econ Behav Organ 1999;38(3):283 – 309. Park SH. Managing an interorganizational network: a framework of the institutional mechanism for network control. Organ Stud 1996;17(5): 795 – 824.
P.M. Norman / Journal of Business Research 57 (2004) 610–619 Park SH, Kim D. Market valuation of joint ventures: joint venture characteristics and wealth gains. J Bus Venturing 1997;12(2):83 – 108. Park SH, Russo MV. When competition eclipses cooperation: an event history analysis of joint venture failure. Manage Sci 1996;42(6): 875 – 90. Park SH, Ungson GR. The effect of national culture, organizational complementarity, and economic motivation on joint venture dissolution. Acad Manage J 1997;40(2):279 – 307. Podsakoff PM, Organ DW. Self-reports in organizational research: problems and prospects. J Manage 1986;12(4):531 – 44. Richter F-J, Vettel K. Successful joint ventures in Japan: transferring knowledge through organizational learning. Long Range Plan 1995; 28(3):37 – 45. Ring PS, Van de Ven AH. Structuring cooperative relationships between organizations. Strateg Manage J 1992;13(7):483 – 98.
619
Rosenthal R, Rosnow RL. Essentials of behavioral research: methods and data. 2nd ed. New York: McGraw-Hill Publishing; 1991. Shenkar O, Li J. Knowledge search in international cooperative ventures. Organ Sci 1999;10(2):134 – 43. Simonin BL. The importance of collaborative know-how: an empirical test of the learning organization. Acad Manage J 1997;40(5):1150 – 74. Uzzi B. Social structure and competition in international networks: the paradox of embeddedness. Adm Sci Q 1997;42(1):35 – 67. von Krogh G, Roos J, Slocum K. An essay on corporate epistemology. Strateg Manage J 1994;15:53 – 71 (Summer Special Issue). Williamson OE. Comparative economic organization: the analysis of discrete structural alternatives. Adm Sci Q 1991;36(2):269 – 96.