A BETTER UNDERSTANDING OF FIRMS
RECOMBINING KNOWLEDGE FROM STRATEGIC
ALLIANCE PARTNERS
Master Thesis
By
Joey van de Burgwal
Rijksuniversiteit Groningen Faculteit Economie en Bedrijfskunde
Msc Business Administration Strategic Innovation Management
First supervisor: P.M.M. de Faria Second supervisor: F. Noseleit
2
ABSTRACT
This study examines the effects of recombining knowledge from alliance partners on the technological performance of a new invention by investigating a total of 91 consolidated firms engaging in R&D alliances for the development of fuel cells in the period of 1990 until 2002. Hypothesis 1, which suggests that recombining knowledge from alliance partners influences the technological performance of a new invention positively, was only partially supported. Finding support for hypothesis 1 depends on a firm’s geographical location. Moreover, support for hypothesis 1 was also found when a firm had started more than four alliances. Hypothesis 2 was not supported as increasing the percentage of knowledge recombined from alliance partners does not further increase the technological performance of a new invention.
Keywords: alliances; alliance partners; inventions; knowledge; knowledge recombination;
patents; patent citations; R&D performance; technological performance
1. INTRODUCTION
The difference between successful and unsuccessful firms is often suggested to depend on the R&D capabilities of a firm to create something new (Rothaermel and Deeds, 2004). A firm that is able to create new inventions can turn around a firm from almost going bankrupt to one of the most successful companies that ever existed (e.g. Apple Inc. with their IPod in 2001). Firms can create such inventions by exploring new knowledge domains (Katila and Ahuja, 2002). However, according to Galunic and Rodan (1998) and Fleming (2001) inventions are often created by recombining knowledge that is already available in the firm. Therefore, a firm’s ability to recombine existing knowledge to develop new and valuable inventions is a key driver of a firm’s R&D performance (Carnabuci and Operti, 2013). While a firm can only recombine its own knowledge, the valuable potential to be recombined again becomes lower when cumulative use increases (Fleming, 2001). Other firms can help to negate this effect as they are a way to gain access to new knowledge.
3
Wassmer and Dussauge, 2012). However, if knowledge is too diverse, it becomes hard to understand (Ahuja and Lampert, 2001). Understanding knowledge well is important, because it helps to reduce the chance of making errors using the knowledge, helps to make the process of knowledge recombination easier and helps to identify the most valuable parts of the knowledge (Katila and Ahuja, 2002).
Hence, the capability of a firm to understand, and therefore, learn from this new and diverse knowledge depends on its absorptive capacity (Cohen and Levinthal, 1990). Zahra and George (2002) describe absorptive capacity as the ability to acquire, assimilate, transform and commercialize new knowledge obtained from others in order to increase their R&D performance. This suggests that firms should be able to not only gain access to new knowledge in order to create new and valuable inventions, but should also have the ability to understand the knowledge from others. Having alliances can help to understand this diverse knowledge well, because working closely together with another firm helps to make understanding another firm’s knowledge easier (Ahuja, 2000).
Yet, existing alliance literature solely analyzes whether R&D performance increases due to the exposure to new knowledge, while neglecting the fact that firms also have to assimilate and recombine that knowledge to increase a firm’s R&D performance. Even some of the most highly mentioned papers in alliance literature disregard this process as these papers are mostly focused on the differences and characteristics of alliance partners (e.g. Sampson, 2007; Phelps, 2010). Neglecting the creation of new inventions, and therefore knowledge recombination, results in an incomplete image of the benefits of having alliances and its effect on the technological performance of newly created inventions (technological performance in short). While alliance literature neglects knowledge recombination, literature that does take into account knowledge recombination ignores the benefits of having alliances. This stream of literature only aims itself on knowledge recombination within intrafirm situations (e.g. Carnabuci and Operti, 2013; Fleming, 2001; Nerkar and Paruchuri, 2005), between different technology domains (e.g. Carnabuci and Bruggeman, 2009; Carnabuci, 2010) or resource combinations (Srivastava and Gnyawali, 2011)
4
partners. Therefore, more research is needed to get a better and deeper understanding of recombining knowledge from strategic alliance partners, which will be the case in this study.
The goal of this study is to analyze the importance of having an alliance for the process of knowledge recombination in order to improve the technological performance of a new invention. Therefore this study tries to answer the following two research questions:
Does recombining knowledge from alliance partners positively influence the technological performance of a new invention?
Are the positive effects of recombining knowledge from alliance partners on the technological performance of a new invention becoming stronger if the percentage of knowledge recombined from alliance partners increases?
Fuel cell patents and firms engaging in alliances for the development of fuel cells are used for the analysis of this study. This data is collected from an online patent database and a large set of news articles published from 1990 until 2002. Combining this data makes it possible to show whether firms have created a new invention that is based on knowledge from an alliance partner or a non-alliance partner. Negative binomial models are used to see if knowledge recombined from alliance partners influences the technological performance positively. Moreover, the models are also used to see if increasing the percentage of knowledge recombined from alliance partners increases the technological performance of a new invention. The results of this study show partial support for the positive influence of knowledge recombined from alliance partners on the technological performance of a new invention. Finding a positive influence depends on the geographical location of a firm. Moreover, support for a positive influence on the technological performance was also found when firms had started more than four alliances. Furthermore, increasing the percentage of knowledge recombined from alliance partners did not increase the technological performance of a new invention.
Besides the theoretical contributions of this study, the results also have managerial implications. The results of this study show that it is not worth the risk, funds and time of every firm to recombine knowledge from alliance partners as it will only result in inventions of higher value under certain conditions.
5
analysis will be discussed, resulting in the study’s theoretical- and managerial implications, limitations and suggestions for future research.
2. THEORETICAL BACKGROUND AND HYPOTHESES
Alliances and Technological Performance
6
position of each firm, alliances also help to decrease the risk (e.g. less costs when a technology fails) and uncertainty of both firms that engage in the alliance (Mullins and Sutherland, 1998). While these R&D alliances are studied extensively and can help to gain access to new knowledge in order to improve R&D performance, some of these alliances perform better than other alliances. Some firms for example, do not reap the benefits from knowledge transfers, as: the alliance is discontinued prematurely (Deeds and Rothaermel, 2003), the alliance influences interfirm knowledge transfer negatively (Gomes-Casseres et al., 2006), or as alliance partners may have completely different routines and organizational mechanisms (Srivastava and Gnyawali, 2011).
Having alliance partners in order to increase R&D performance, therefore, depends on how much a firm has to learn and its capability to do so (Sampson, 2007). Current alliance literature describes that the effects of knowledge sharing within alliances on a firm’s ability to create valuable inventions mostly depends on the diversity of knowledge that is shared (e.g. Sampson, 2007; Phelps, 2010), the characteristics of alliance partners and the differences between alliance partners (e.g. Ahuja 2000; Cui and O’Connor, 2012; Duysters et al., 2012; Mowery et al., 1996).
The diversity of knowledge can be described as the variety of knowledge, knowhow and expertise someone has access to through their network (Rodan and Galunic, 2004). This suggests that when a firm’s knowledge diversity increases, it will increase the heterogeneity of knowledge available to that firm and therefore the amount of unique knowledge a firm can gain access to. Sampson (2007) and Phelps (2010) both describe that sharing diverse knowledge within alliances increases R&D performance. It should be added however that having a moderate diversity contributes the most to an increase in the technological performance, because knowledge that is too similar lacks explorative potential. Moreover, sharing knowledge that is too different can result in information overload and confusion that harms the creation of valuable inventions as the respective knowledge is too complex to understand (Ahuja and Lampert, 2001). Therefore, firms commonly avoid having alliances with firms of which knowledge is too different (Mowery et al., 1998).
7
with other firms decrease. Simply put, knowledge from others becomes too similar to a firm’s own knowledge. Moreover, some firms may see the decreasing value of sharing knowledge with external partners when internal knowledge diversity is high becoming even stronger as the risk of knowledge being exposed and used by others also increases (Srivastava and Gnyawali, 2011). Hence, Srivastava and Gnyawali (2011) add that firms with relatively low internal knowledge diversity benefit the most from knowledge that is acquired from alliance partners.
In short, firms that have partners with moderately diverse knowledge or a low internal diversity themselves benefit the most from having alliances.
While the diversity of knowledge and its effects on the technological performance are studied comprehensively in alliance literature, one of the other aspects influencing knowledge sharing between alliance partners and therefore the technological performance are the differences between alliance partners. Hence, the characteristics of firms can make a difference. Some of them may affect the technological performance directly, while other characteristics affect the diversity of knowledge, and therefore, indirectly affect the technological performance.
One of these differences described in literature are the different positions and relationships firms have in alliance networks, which Doz and Hamel (1998) define as a set of linkages between firms that are relatively comparable. There are two types of ties a firm can have in an alliance network: direct and indirect ties (Ahuja, 2000). Direct relationships with other firms are a source of both resources and information, whereas indirect relationships are only a source of information. Although indirect relationships are less costly to maintain, direct relationships, and therefore alliances, are more beneficial as these also include sharing resources between each other. Moreover, having alliance partners that are densely connected with each other benefit the technological performance too (Phelps, 2010). Although R&D alliances are direct relationships with other firms, they can be managed differently. Firms can collaborate with each other by engaging in contract based alliances or creating joint ventures, which are a separate entity owned by both the alliance partners (Mowery et al., 1996). Two firms creating a joint venture proved to be influencing technological performance more positively than a contract based alliance (Mowery et al., 1996; Sampson, 2007). Working more closely together, as is the case in a joint venture, increases the motivation of both partners to share knowledge with each other, and therefore, improves the knowledge flow (Sampson, 2007).
8
diverse relationships with other firms outweigh the positive gains, and therefore, affects the development of new technologies. Managing a firm’s alliance portfolio becomes substantially more difficult when the number of firms in that alliance portfolio increase. Unsuccessfully managing these relations (Cui and O’Connor, 2012), or not having the capability to do so (Duysters et al., 2012), results in conflicts and confusion which could hurt the technological performance. Investments in an alliance structure and alliance portfolio to coordinate alliance activity can help maintaining relations beneficial (Kale et al., 2001; Wassmer and Dussauge, 2012). Moreover, increasing the number of alliance partners from different countries is less beneficial as the geographical distance makes it harder to communicate and coordinate and therefore affects the assimilation of new knowledge (Cui and O’Connor, 2012). Furthermore, having experience working together with different firms can be beneficial as it helps to learn how alliances are managed properly (Duysters et al., 2012). Although other literature suggests that prior experience with the same firm can also be beneficial for technological performance (e.g. Kim and Song, 2007; Sampson, 2005; Sampson, 2007), Li et al. (2008) add that these results should be taken with some caution. Firms that have prior experience working together with the same firm can actually be harmful for the technological performance in certain situations. Alliance partners learn about each other’s routines and core technologies, making it easier to act opportunistically. Hence, it increases the risk of exposing a technology on which a firm’s competitive advantage is based.
In short, current literature suggests that the benefits of having alliances are stronger when firms work more closely together, are positioned densely together, have some prior experience working together with different firms and when firms have the capability to manage close relationships with others.
Recombining Knowledge and Technological Performance
9
in order to be able to exploit it (Zahra and George, 2002). A firm that has the ability to do so will therefore be able to increase its R&D performance (Zahra and George, 2002).
Currently, this study has mostly focused itself on one way to create new inventions. To do so, firms need new and diverse knowledge, which they can obtain by exploring currently unknown knowledge domains with the help of other firms (Levinthal and March, 1993; Katila and Ahuja, 2002). However, there is another way as Katila and Ahuja (2002) suggest that firms can also create new inventions by reusing their existing knowledge. Literature on this knowledge recombination process however, has mostly been studied on a firm level as it describes inventors recombining their knowledge with each other within the firm (e.g. Fleming, 2001).
As described earlier, a firm’s competence to recombine existing knowledge in order to create new innovations can be a key driver of a firm’s R&D performance (Galunic and Rodan, 1998; Carnabuci and Operti, 2013). Literature describes two types of knowledge recombination that have the potential to increase the technological performance. The first type is described as recombinant reuse by Carnabuci and Operti (2013) or refinement of familiar combinations (Fleming, 2001), and can be defined as reusing earlier used combinations of knowledge. The second type can be described as recombinant creation (Carnabuci and Operti, 2013) or recombination of familiar components (Fleming, 2001), and can be defined as recombining existing knowledge to create new knowledge. Recombining knowledge is more beneficial compared to completely new knowledge due to fact that firms have used the knowledge earlier and is therefore better understood (Katila and Ahuja, 2002). This experience helps firms to reduce the chance of making errors using the knowledge, helps them to make the process of using the knowledge more efficient and helps them to easier identify the most valuable parts (Katila and Ahuja, 2002). Sharing and recombining knowledge or components of knowledge is not a simple process, as knowledge recombination has different effects on the technological performance under different circumstances.
10
familiar, can result in a higher chance that knowledge from others is better understood. Fleming and Sorensen (2001), for example, examined the interdependence of an invention¹, which they define as the functional sensitivity of an invention to changes in the constituent components, in relation to the usefulness of an inventor’s efforts. In here, an inventor’s efforts can be described as the result of a process in which inventors search for better recombinations of technologic components. The study of Fleming and Sorensen (2001) shows that the usefulness of an inventor’s efforts increases when inventors within a firm are recombining components that have a moderate degree of interdependence. Basically, this suggests that if the interdependence of components is moderately different, the inventor’s efforts will have a positive effect on a firm’s technological performance as the inventor can find better combinations of knowledge. Additionally, this also suggests that if the interdependence of knowledge components is too different, inventors will not understand the components they share and are overwhelmed by their complexity resulting in a lower technological performance. These findings are similar to alliance literature findings that argue that knowledge shared by firms with a rather different technological background have a hard time understanding each other’s knowledge (Phelps, 2010; Sampson, 2007). Although knowledge recombination has its positive influences on the technological performance, recombining the same knowledge repeatedly will weaken the potential a firm can gain from that knowledge as it becomes less fertile (Fleming, 2001), too complex and more expensive to be recombined again (Katila and Ahuja, 2002).
Recombining Alliance- and Knowledge Recombination Literature
Despite the fact that both literature streams consider knowledge diversity and technological performance, only a subset of both literature streams are interconnected with each other. Where alliance literature focused itself on gaining access to diverse knowledge through partners, knowledge recombination is mostly a concept discussed within literature that focuses on firms increasing the technological performance alone (e.g. Fleming, 2001; Nerkar and Paruchuri, 2005). In short, alliance literature almost completely ignores the steps taken after access to knowledge is gained, while literature on knowledge recombination has not taken into consideration the positive effects of working closely together with other firms. Although both streams are currently quite isolated from each other, similar concepts were tested, which further suggest their strong relatedness.
11
Nerkar and Paruchuri (2005) found that the individual position of an inventor within an intraorganizational network can predict the chance of knowledge being selected by others for recombination. Their findings suggest that the extent of structural holes being spanned by an inventor, increase the chance of their knowledge being selected for knowledge recombination. This is due to the fact that their knowledge is more diverse. Structural holes can be referred to as firms/inventors on either side having access to different flows of knowledge (Hargadon and Sutton, 1997).This means for example, that firm/inventor A has direct access to the knowledge of firms/inventors B, C and D, while firm B, C and D only have direct access to the knowledge of firm/inventor A. Additionally, alliance literature also suggests that firms expanding more structural holes have access to more diverse knowledge (e.g. Ahuja, 2000), meaning that working directly with others has more potential benefits as both sides of the knowledge flow learn directly from each other.
Although current alliance literature mostly ignores knowledge recombination, the study of Srivastava and Gnyawali (2011) is sort of an exception. They studied the relationship between resource quality and resource combinations by alliance partners. Although knowledge is included in resources, it is only a small part of it. They refer to resource quality as the reliability of technological resources from partners within an alliance portfolio. According to Cattini (2005), resources are of high quality if a large set of newly created resource combinations are based on them. If the resource quality of a firm within an alliance portfolio is higher than another firm, it becomes less valuable for that firm to use resources of others. This firm will rather use its own resources to use new combinations in order to create new innovations (Katila, 2002).
12
transform the knowledge (Katila and Ahuja, 2002; Zahra and George, 2002), which will be better achievable when the knowledge from others is moderately diverse (e.g. Miller, 2007; Sampson, 2007). Firms can obtain knowledge from both direct- and indirect ties (Ahuja, 2000). Although not having direct relationships with others costs less funds and effort to manage, having these direct ties with others in the form of alliances adds the benefit of sharing resources such as inventors and equipment (Ahuja, 2000). These extra resources, or input resources as Galunic and Rodan, (1998) describe them, can help to better understand another firm’s knowledge as there is a high chance that technical knowledge obtained from another firm is highly tacit (Ahuja, 2000; Galunic and Rodan, 1998). This suggests that having alliances not only helps to increase the diversity of knowledge as they are a way to gain access to new knowledge. Additionally, having alliances also helps to make the process of knowledge recombination easier than recombining knowledge alone as it includes direct support from another firm (Ahuja, 2000).
Ultimately, it has become clear that knowledge recombination has been neglected by alliance literature as this stream of literature sees knowledge only as an undividable element that can be shared with others, while intrafirm literature already took that step further by suggesting that knowledge can be divided in parts and recombined with each other. Yet, this stream of literature does not take into account that collaborating closely together with other firms can make the process of knowledge recombination easier. Potentially, it could result in even better inventions as alliance partners are also a source of more diverse knowledge. In short, while intrafirm literature acknowledges that recombining parts of knowledge within a firm increases the technological performance, it is unproven whether effects of recombining existing knowledge from one firm with existing knowledge from an alliance partner yields similar results. Moreover, an alliance partner gives a firm access to more diverse knowledge, and an alliance partner also helps a firm to better understand the alliance partner’s knowledge. This suggests that increasing the percentage of knowledge recombined from alliance partners could further increase the technological performance of a new invention.
Therefore, it is hypothesized that:
Hypothesis 1: Recombining knowledge from alliance partners influences the technological performance of a new invention positively
13
3. METHODOLOGY
Research Sample and Data Collection
The firms, and their alliances used in order to test the hypotheses of this study are settled in different industries. Yet, all these firms were developing the same technology from 1990 until 2002, which was the fuel cell. A fuel cell is an electrochemical device that turns chemical energy into electrical energy and can, for example, be used to power vehicles, factories and houses. This period and technology were chosen for two reasons. First, there was almost no commercialization of the fuel cell from 1990 until 2002, due to fact that the technology was still too costly to produce. Hence, firms were still working on prototypes and different types of fuel cells (e.g. SOFC, PEM fuel cell and the MCFC) to lower these costs. To further improve the prototypes and come up with a standard (dominant design) of fuel cells, firms were in need of more knowledge. As explained earlier, firms could obtain this new knowledge by themselves or by engaging in alliances. Secondly, patent data was used to test to test the data. Thus, using an emerging technology was important, because the number of filed patents related to one technology will be higher when a technology is just emerging compared to one that is already emerged.
Three databases were used to obtain the alliance- and patent data needed for the analysis. The first database, which was the fuel cell patent database, was already created with the help of PATSTAT online, also known as the EPO Worldwide Patent Statistical Database. The database of PATSTAT online has most information regarding patents already available, this includes patent categories, making it possible to find the patents related to fuel cell development. Other relevant information needed for the study was also obtainable from this database. This includes the date of application, the number and source of both backward- and forward citations, and the name of the firm that filed the patent.
The second database that was planned to be used aims itself on alliances and is called the Securities Data Company (SDC) database on Joint Ventures and Alliances. However, after some research it became clear that a significant number of alliances related to fuel cell development was missing. Hence, a new alliance database had to be created.
The method used to create the new alliance database is based on a method used by Vasudeva and Anand (2011). With this method alliance data was collected by searching news sources in categories such as major world publications, newspapers, magazines, wire services,
14
tool called Lexis-Nexis. Although the category All news should have included all the news articles, a significant amount of news articles were missing. Hence, the other news sources were also consulted.
A set of search terms, which included words related to alliances and fuel cells², was created to find relevant news articles. This resulted in a total number of around 16.000 news articles spread over almost 40.000 pages. All news articles were searched through, in order to find firms engaging in a fuel cell alliance. If an article mentioned an alliance as illustrated in an example further down³, essential information was documented. This included the names of the firms engaging in an alliance, the start- and end date of the alliance, the name of the article and its publication date, and the type of alliance (e.g. commercial, R&D or licensing).
This resulted in a total of 644 firms and 634 alliances. The alliances suggesting to be related to the development of fuel cells, and therefore, the creation of new inventions were the ones needed to test the hypotheses. However, the firm names of the patent database and the newly created alliance database were not always the same, potentially resulting in errors such as counting one firm as two separate firms. Therefore, another database was created.
This third database contained all firm names as described in both the alliance database and the patent database created with the help of PATSTAT online. Some firm names were spelled differently or wrong, thus corrected. This resulted in a total of 216 firms and 341 fuel cell development alliances, in which each subsidiary was counted as a separate firm. In order to create a more complete sample, subsidiaries and their parents were linked with each other. This resulted in a final sample of 91 consolidated firms (subsidiaries and parents counted as one firm).
³”General motors corp. and Toyota Motor Corp. said Monday they have reached a five-year
agreement to jointly develop cars powered by fuel cells and other non-traditional fuel technologies.” (Xinhua News Agency, 1999)
15
Measures: The Dependent- and Independent Variables
The technological performance of a new invention
The number of patents a firm has filed can be described as a measure of the number of inventions a firm has created (Basberg, 1982). However, simple patent counts are not an indicator of a firm’s R&D performance, because patents vary in their technological significance (Trajtenberg, 1990). Therefore, firms that want to improve their R&D performance should not only aim themselves on increasing the amount of new inventions they create, but also the value of these new inventions. Inventions build upon by others can indicate the value of an invention (Cattani, 2005). Therefore, if the amount of others building upon an invention increases, it also increases the technological importance of an invention. Each patent includes backward citations to another patent, making it also possible to find the amount of forward citations an earlier created patent has received. Hence, the number of forward citations a patent receives from other patents indicates the technological importance of an invention (Carpenter et al., 1981; Trajtenberg, 1990). Further prove of patent citations helping firms to improve their R&D performance was found by Hall et al. (2005) as they found that a firm’s market value increases by 3% for each forward citation a patent they own receives. Therefore, the dependent variable, which is the technological performance of a new invention, was measured by the number of forward patent citations a patent receives in a window of 4 years, adding that these patents have been filed during the period of 1990 until 2002. In short, when the number of forward citations a patent receives increases, it is assumed that the technological performance of that patent increases.
Recombining knowledge from alliance partners
16
an alliance partner. The variable share of alliance citations was created to test the second hypothesis and computes the percentage of citations a patent makes to a patent from an alliance partner. When the share of alliance citations increases, it means that the percentage of citations made to patents from alliance partner increases while the percentage of citations made to patents from non-alliance partners decreases.
Measures: Control variables
To make sure alternative explanations are minimized, the analysis controls for several patent- and firm level variables, whose effects on the innovation performance could be confounded with the independent variables.
Older internal technology is often well-known and less fertile (Katila, 2002), and could therefore, be less valuable to recombine. On the other hand, older external knowledge can still be of value when it is not well known (Katila, 2002). This suggests that the age of knowledge has an effect on the technological performance. The age of a patent is calculated as the average age of backward citations.
Using knowledge acquired from different and more diverse knowledge sources helps increasing the chance of creating new and more valuable knowledge (Smith et al., 2005). This can have an effect on the technological performance. This variable is computed as the number of backward citations a patent receives.
Technology becomes more mature when the granted patents are less speculative and closer to the market, which makes it more focused on one patent family (Williams, 2007). This could have an effect on the technological performance of an invention and is therefore controlled for. The share of granted patents in the same patent family is measured as the percentage of patents filed.
If more inventors have worked on a patent, it often increases the diversity of knowledge sourcing from these inventors. This can have an effect on the technological performance (Reagans and Zuckerman, 2001). Team size is computed as the number of inventors that have worked on a particular patent.
17
number of patent authorities can affect the technological performance. The number of patent authorities is calculated as the number of patent authorities a patent is applied to in the patent family.
The diversity of knowledge has a positive effect on the technological performance, although its positive effect on the technological performance will diminish if knowledge becomes too diverse (Fleming and Sorensen, 2001). The technological diversity of backward citations is measured on the basis of herfindahl-index of IPC class concentration in backward citations.
Exploring new technology domains help to gain access to new knowledge and could therefore increase the technological performance (Levinthal and March, 1993). Exploration is measured as the share of citations made to patents that have never been used before.
Using the same knowledge over and over again makes it lose its potential to be successfully used again over time, and therefore, negates the positive effect on the technological performance (Fleming, 2001). Own use in year is calculated as the average number of patents that cite the same patents as the focal patent by the same firm.
Resources build upon by others, in which knowledge is included, indicates that the quality of knowledge is high (Cattani, 2005). High quality knowledge contributes to the technological performance of an invention (Srivastava and Gnyawali, 2011). Other use in year is measured as the average number of patents that cite the same patents as the focal patent by other firms.
Knowledge can be described as highly tacit if it cannot be easily explained, and as a result, becomes hard to understand for others (Galunic and Rodan, 1998). Often, knowledge becomes highly tacit if it resides within the social structures of a firm, such as a great number of different inventors (Hoetker and Agarwal, 2007). This can negatively influence the technological performance of a new invention. Tacitness of knowledge is computed as the average number of inventors per backward citation.
18
Analysis
A theory testing approach is adopted for this research, accordingly a statistical tool was used to test the data. For this study STATA has been used to analyze the data obtained from both the PATSTAT database and the newly created alliance database. The unit of analysis for this study was on a patent level. The technological performance of a new invention, which is the total number of backward citations a patent gets, only includes whole numbers like 0, 1, 2 and 3. This data indicates how often something has happened (that is, how often a patent was cited by someone else), and could therefore, be described as a count variable (Long and Freese, 2001). Linear regression models can be used to test count data. However, these models often result in inefficient, inconsistent and biased estimates (Long and Freese, 2001). Normally researchers make use of poisson regression models as these are specifically made for count data. However, if the variance in data is greater than the mean it is better to make use of negative binomial regression models. The data used to measure the dependent variable is over dispersed (that is, the variance is greater than the mean), and therefore, this study makes use of negative binomial regression models.
4. RESULTS
The descriptive statistics and correlations of the variables discussed in the methodology part are illustrated in respectively Table 1 and 2. Table 1 shows that the dependent variable’s minimum is zero, as it describes
the number of patent citations, and therefore can not be lower than zero. On average, patents were cited almost six times, while the maximum number of citations a patent has received was 85 times. The table also shows that the dependent variable’s variance is indeed higher than the mean indicating that overdispersion is present. The independent variable
any citation to alliance partner
19
that 29% of the patents cited atleast one patent from an alliance partner. The independent variable share of alliance citations shows a minimum share of 0%, indicating that some patents do not cite patents of alliance partners at all. On the other hand, there are also patents that only cite patents from alliance partners illustrated by the maximum share of 100%. On average, 6% of the citations a patent makes are alliance citations.
Although Table 2 shows mostly negligible, weak, moderately positive and negative correlation between the variables, there is one strong positive correlation. The share of alliance citations shares a moderately positive correlation with the accumulated alliance experience indicating that having more experience with alliances occurs relatively often together with an increase of filed patents that include citations to alliance partners. Moreover, there is a strong correlation between the share of alliance citations and any citation to alliance partner. This indicates that patents citing a patent from an alliance partner often happens together with an increase of the share of alliance citations. Yet, multicollinearity is not an issue as further analyis of the variables shows that the variance inflations factor (VIF) never exceeds 1,38.
20
21
number of patent authorities and own use of patent in year increases. This indicates that the database used in this study is similar as databases used in earlier conducted research, and therefore, can be used to test the hypotheses. Model 2 includes the independent variable any
citation to an alliance partner. It was predicted that recombining knowledge from alliance
22
However this prediction, and therefore hypothesis 1, is not supported as it shows a positive, yet insignificant result (0,059). Model 3 includes the other independent variable share of alliance
citations to test hypothesis 2, which predicted that increasing the percentage of knowledge
recombined from alliance partners would increase the technological performance. However, this hypothesis is also not supported as the results show a positive, yet insignificant result (0,010).
The predicted effects of knowledge recombination on technological performance were not found, thus, further analysis was conducted. Earlier literature describes that having experience working closely together with different firms can be beneficial for performance as it helps to learn how alliances are managed properly (Duysters et al., 2012). Hence, models 4, 5, 6 and 7 illustrate the effects of both independent variables on the technological performance when including a median split of alliance experience (median is four). Model 4 and 5 only includes firms that have started four or less alliances, while model 6 and 7 only takes into account firms that have started more than four alliances. Considering an accumulated alliance experience of four or less does not make any difference, as no significant results are found (model 4 and 5). However, model 6 does include a positive and significant result (0,141). This illustrates that patents citing at least one alliance partner influences the technological performance of an invention positively, but only if the firm that owns the patent has started more than four alliances. Based on this finding, hypothesis 1 can be partially supported. Model 7 shows that no significant effects were found for the share of alliance citations on the technological performance when only including patents from firms that have started more than four alliances.
23
Moreover, finding support or no support for hypothesis 1 seems to depend on the geographical location of the firms engaging in alliances.
5. DISCUSSION AND CONCLUSION
Theoretical- and Managerial Implications
Research on alliance literature has mostly focused its efforts on the benefits of having alliances in order to increase the technological performance of a new invention. Knowledge is seen as one of the most important aspects affecting the relationship between having alliances and the technological performance. However, literature mostly fixated itself on aspects that either positively or negatively influence the access to new knowledge from alliance partners. Aspects such as the characteristics of and differences between alliance partners do have their influences on the diversity of knowledge that is obtained from others (e.g. Ahuja, 2000; Phelps, 2010). Yet, the steps taken after knowledge is obtained are almost completely neglected by current alliance literature. Not taking into account whether knowledge is recombined from alliance partners or non-alliance partners tends to overestimate the amount of newly created knowledge directly resulting from these alliances. Hence, it was unclear if working closely together with firms to recombine knowledge is better than recombining knowledge from non-alliance partners as working closely together with non-alliance partners helps to understand each other’s knowledge better. This study tried to fill the gap of existing literature by examining the differences between knowledge recombined from alliance partners and knowledge recombined from non-alliance partners. However, the empirical results are not supporting the arguments describing that the technological performance will increase when (more) knowledge is recombined from alliance partners.
24
experience helps to better manage relationships with alliance partners that have diverse knowledge (Duysters et al., 2012). Hence, firms can create inventions of higher value with knowledge from alliance partners if they learn how to manage alliances properly. Firms can gain experience by starting more alliances with different firms.
Further investigation of the data showed another interesting insight. Most of the firms with a high number of filed patents were Japanese. Other origins were mostly the U.S.A. and Germany, suggesting that the geographical location of firms could matter. Hence, the analysis took into account these different geographic locations. Japanese alliances are not managed that differently compared to western firms nowadays, which is due to the westernization Japanese firms (Aoki and Lennerfors, 2013). However, alliances used to be managed differently in the nineties (Ahmadjian and Lincoln, 2001; Aoki and Lennerfors, 2013). Most of the large Japanese firms were members of a keiretsu network, which can be described as a Japanese form of business practice associated with a network of long term relationships between firms (Aoki and Lennerfors, 2013). Research in the nineties includes a lot of research regarding the best way to manage relationships with others. As such, the differences between Japanese and western firms were often discussed (e.g. Dyer, 1996; Kotabe, 1998; Putnam and Chan, 1998). These studies argue that the way Japanese firms handled their alliances were competitively much better than their western counterparts due to their higher efficiency. This higher efficiency was a result of governing knowledge exchange based on trust (Dyer, 1996).
25
that trusted each other less. Moreover, increasing the share of alliance citations does not seem to matter, because the analysis only shows positive results for the difference between having- and not having any citation to an alliance partner at all.
Limitations and Future Research
The limitations of this study also indicate directions for future research. First, the news articles consulted to create the alliance database only included English news sources. This has resulted in mostly alliance announcements between U.S. and European, Asian or other U.S. firms. The chance that alliances between two European firms, two Asian firms or an Asian and European firm were neglected is rather high. Future research could take into account news articles from different languages to make a more comprehensive and generizable database of alliances.
Secondly, the period from 1990 until 2002 was not as fruitful as initially expected. Over 50% of the alliances found were announced in only two out of the 12 years tested, suggesting that a later period could have resulted in a larger sample. Moreover, although there were some positive relationships found between knowledge recombination and the technological performance, the results of the study were mostly not significant. This illustrates another reason for future research to study a later period. The lack of alliances in the first years has probably two reasons. First, it might simply be the result of an increase in the number of alliances as the emergence of the technology became stronger in the late nineties. Secondly, it can be the result of more news articles mentioning alliances focusing on the development of fuel cells, which might be the result of an increase of the popularity of fuel cells. An increasing number of articles from the late nineties was hyping the fuel cell technology as one that could make the petrol engine of cars obsolete. Although it is not really a limitation, it can be interesting to learn whether this hype created by newspapers helps to improve the development of a technology. Moreover, further investigation of Lexis-Nexis shows that 250.000 pages of news articles that were published from 2002 until 2007 mentioned both alliances and fuel cells. This is an increase of 210.000 pages compared to the years that were analyzed in this study.
26
This study only included firms engaging in alliances for the development of fuel cells and as a result care must be taken generalizing these findings. As fuel cells are a high tech product, the pace of development is very swift. This might make the findings less generalizable for other types of technologies.
Patents were used as a measure of knowledge and technological performance. However, not all inventions are filed, which implies that this study does not capture all the inventions created by a firm. Firms decide against patents for two reasons. On the one hand, firms keep their inventions a secret to make sure others will not imitate them. On the other hand, firms might not file patents because they are afraid knowledge might spill over to another firm when engaging in an alliance (Brouwer and Kleinknecht, 1999).
Consolidated firms were used to analyze the data. While this gives a more complete sample, it also means that some firms did not have an alliance with each other directly (e.g. Plug Power is a subsidiary of UTC and has an alliance with another firm, while UTC did not have an alliance with the other firm).
Lastly, care has to be taken when interpreting the results of this study as the independent variable any citation to an alliance partner is a dummy. The variable only shows the difference between patents citing to an alliance patent and patents not citing to an alliance patent, while ignoring the number of alliance citations.
Conclusion
27
6. REFERENCES
Ahuja, G. (2000). Collaboration networks, structural holes and innovation: A longitudinal study. Administrative Science Quarterly. 45(3), 425–457.
Ahuja, G., & Lampert, C.M. (2001). Entrepreneurship in the large corporation: A longitudinal study of how established firms create breakthrough inventions. Strategic
Management Journal, 22(6‐7), 521-543.
Ahmadjian, C. L., & Lincoln, J. R. (2001). Keiretsu, governance, and learning: Case studies in change from the Japanese automotive industry. Organization Science, 12(6), 683-701.
Aoki, K., & Lennerfors, T. T. (2013). Whither Japanese keiretsu? The transformation of vertical keiretsu in Toyota, Nissan and Honda 1991–2011. Asia Pacific Business Review, 19(1), 70-84.
Basberg, B. L. (1982). Technological change in the Norwegian whaling industry: A case-study in the use of patent-statistics as a technology indicator. Research Policy, 11(3), 163-171.
Brouwer, E., & Kleinknecht, A. (1999). Innovative output, and a firm's propensity to patent.: An exploration of CIS micro data. Research Policy, 28(6), 615-624.
Carpenter, M. P., Narin, F., & Woolf, P. (1981). Citation rates to technologically important patents. World Patent Information, 3(4), 160-163.
Carnabuci, G., & Bruggeman, J. (2009). Knowledge specialization, knowledge brokerage and the uneven growth of technology domains. Social Forces, 88(2), 607-641.
28
Carnabuci, G., & Operti, E. (2013). Where do firms' recombinant capabilities come from? Intraorganizational networks, knowledge, and firms' ability to innovate through technological recombination. Strategic Management Journal, 34(13), 1591-1613.
Cattani, G. (2005). Preadaptation, firm heterogeneity, and technological performance: A study on the evolution of fiber optics, 1970–1995. Organization Science, 16(6), 563-580
Cohen, W. M., Levinthal, D.A. (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35, 128-152.
Cui, A.S. & O’Connor, G. (2012) Alliance Portfolio Resource Diversity and Firm Innovation. Journal of Marketing, 76(4), 24-43.
Deeds, D. L., & Rothaermel, F. T. (2003). Honeymoons and liabilities: The relationship between age and performance in research and development alliances. Journal of
Product Innovation Management, 20(6), 468–484.
Doz, Y. L., & Hamel, G. (1998). Alliance advantage: The art of creating value through
partnering. Harvard Business Press.
Duysters, G., Heimeriks, K. H., Lokshin, B., Meijer, E., & Sabidussi, A. (2012). Do Firms Learn to Manage Alliance Portfolio Diversity? The Diversity‐Performance Relationship and the Moderating Effects of Experience and Capability. European Management
Review, 9(3), 139-152.
Dyer, J. H. (1996). Does governance matter? Keiretsu alliances and asset specificity as sources of Japanese competitive advantage. Organization Science, 7(6), 649-666.
Fleming, L. (2001). Recombinant uncertainty in technological search. Management Science, 47(1), 117–132.
29
Galunic, C. & Rodan, S. (1998). Resource recombinations in the firm: knowledge structures and the potential for Schumpeterian innovation. Strategic Management Journal, 19(12), 1193-1201.
Gomes-Casseres, B., Hagedoorn. J., & Jaffe, A. (2006). Do alliances promote knowledge flows? Journal of Financial Economics, 80(1), 5–33.
Hall, B. H., Jaffe, A., & Trajtenberg, M. (2005). Market value and patent citations. RAND
Journal of Economics, 36(3), 16-38.
Hargadon, A., & Sutton, R. I. (1997). Technology brokering and innovation in a product development firm. Administrative Science Quarterly, 42(4), 716-749.
Hoetker, G., & Agarwal, R. (2007). Death hurts, but it isn't fatal: The postexit diffusion of knowledge created by innovative companies. Academy of Management Journal, 50(2), 446-467.
Jiang, L., Tan, J. and & Thursby, M. (2010). Incumbent Firm Invention in Emerging Fields: Evidence Form the Semiconductor Industry. Strategic Management Journal, 32(1), 55–75.
Kale, P., Dyer, J., & Singh, H. (2001). Value creation and success in strategic alliances: alliancing skills and the role of alliance structure and systems. European Management
Journal, 19(5), 463-471.
Katila, R. (2002). New product search over time: Past ideas in their prime. Academy of
Management Journal, 45(5): 995–1010.
Katila, R., & Ahuja, G. (2002). Something old, something new: A longitudinal study of search behavior and new product introduction. Academy of Management Journal, 45(6), 1183-1194.
30
Kotabe, M. (1998). Efficiency vs. effectiveness orientation of global sourcing strategy: A comparison of US and Japanese multinational companies. The Academy of Management
Executive, 12(4), 107-119.
Lane, P.J., & Lubatkin, M. (1998). Relative absorptive capacity and interorganizational learning. Strategic Management Journal, 19(5), 461-477.
Levin, R. C., Klevorick, A. K., Nelson, R. R., Winter, S. G., Gilbert, R., & Griliches, Z. (1987). Appropriating the returns from industrial research and development. Brookings Papers
on Economic Activity, 3, 783-831.
Levinthal, D. A., & March, J. G. (1993). The myopia of learning. Strategic Management
Journal, 14(2), 95-112.
Li, D., Eden, L., Hitt, M. A., & Ireland, R. D. (2008). Friends, acquaintances, or strangers? Partner selection in R&D alliances. Academy of Management Journal, 51(2), 315-334.
Long, J. S., & Freese, J. (2001). Regression models for categorical dependent variables using
Stata. Stata press.
Miller, D. J., Fern, M. J., & Cardinal, L. B. (2007). The use of knowledge for technological innovation within diversified firms. Academy of Management Journal, 50(2), 307-325.
Mockler, R. J., Dologite, D. G., & Carnevali, N. M. (1997). Type and structure of multinational strategic alliances: the airline industry. Strategic Change, 6(5), 249-260.
Mowery, D.C., Oxley, J.E., & Silverman, B.S. (1996). Strategic alliances and interfirm knowledge transfer. Strategic Management Journal, 17, 77-91.
Mowery, D. C., Oxley, J. E., & Silverman, B. S. (1998). Technological overlap and interfirm cooperation: implications for the resource-based view of the firm. Research
31
Mullins, J. W., & Sutherland, D. J. (1998). New product development in rapidly changing markets: an exploratory study. Journal of Product Innovation Management, 15(3), 224-236.
Nelson, R. R. (1990). Capitalism as an engine of progress. Research Policy, 19(3), 193-214.
Nerkar, A., & Paruchuri, S. (2005). Evolution of R&D capabilities: The role of knowledge networks within a firm. Management Science, 51(5), 771-785.
Patel, P., & Pavitt, K. (1997). The technological competencies of the world’s largest firms: Complex and path dependent, but not much variety. Research Policy, 26(2), 141–156.
Phelps, C.C. (2010). A longitudinal study of the influence of alliance network structure and composition on form exploratory innovation. Academy of Management Journal, 53(4), 890-913.
Putnam, L., & Chan, P. S. (1998). The American Keiretsu: America's new competitive advantage. American Business Review, 16(1), 113-118
Reagans, R., & Zuckerman, E. W. (2001). Networks, diversity, and productivity: The social capital of corporate R&D teams. Organization Science, 12(4), 502-517.
Rodan, S., & Galunic, C. (2004). More than network structure: how knowledge heterogeneity influences managerial performance and innovativeness. Strategic Management
Journal, 25(6), 541-562.
Rothaermel, F. T., & Deeds, D. L. (2004). Exploration and exploitation alliances in biotechnology: a system of new product development. Strategic Management
Journal, 25(3), 201-221.
32
Sampson, R. C. (2007). R&D alliances and firm performance: the impact of technological diversity and alliance organization on innovation. Academy of Management Journal, 50(2), 364-386.
Smith, K. G., Collins, C. J., & Clark, K. D. (2005). Existing knowledge, knowledge creation capability, and the rate of new product introduction in high-technology firms. Academy
of Management Journal, 48(2), 346-357.
Spekman, R. E., Forbes III, T. M., Isabella, L. A. & Macavoy, T. C. (1998). Alliance Management: A View from the Past and a look to the Future. Journal of Management
Studies, 35(6), 747-772.
Srivastava, M. K., & Gnyawali, D. R. (2011). When do relational resources matter? Leveraging portfolio technological resources for breakthrough innovation. Academy of
Management Journal, 54(4), 797-810.
Trajtenberg, M. (1990). A penny for your quotes: patent citations and the value of innovations. The Rand Journal of Economics, 21(1), 172-187.
Vasudeva, G., & Anand, J. (2011). Unpacking absorptive capacity: A study of knowledge utilization from alliance portfolios. Academy of Management Journal, 54(3), 611-623.
Wassmer, U., & Dussauge, P. (2012). Network resource stocks and flows: how do alliance portfolios affect the value of new alliance formations? Strategic Management Journal, 33(7), 871-883.
Williams, G. (2007). A maturing industry: Strategic patenting trends, US and academic dominance, and the global biotechnology landscape. Journal of Commercial
Biotechnology, 13(3), 163-175.
Yayavaram, S., & Chen, W. R. (2015). Changes in firm knowledge couplings and firm innovation performance: The moderating role of technological complexity. Strategic
33
