copyright page: © 2013 xiaoyu pu all rights reserved
TRANSCRIPT
Copyright page:
© 2013
Xiaoyu Pu
ALL RIGHTS RESERVED
MNC SUBUNIT KNOWLEDGE SOURCING AND COMPETENCE CREATING
ACTIVITIES – A DYNAMIC VIEW OF SUBUNIT EVOLUTION
By
XIAOYU PU
A Dissertation submitted to the
Graduate School-Newark
Rutgers, The State University of New Jersey
in partial fulfillment of the requirements
for the degree of
Doctor of Management
Graduate Program in
Rutgers Business School
written under the direction of
Professor Dr. John A. Cantwell
and approved by
_________________________
_________________________
_________________________
_________________________
Newark, New Jersey
May, 2013
ii
ABSTRACT OF THE DISSERTATION
MNC Subunit Knowledge Sourcing and Competence Creating Activities – A Dynamic
View of Subunit Evolution
By XIAOYU PU
Dissertation Director:
Professor Dr. John Cantwell
The innovative activities of multinational corporation (MNC) operations overseas can be
represented as two types: either competence exploiting (CE) – exploiting the core
competence base of the parent group – or competence creating (CC) – creating new
competencies that were not already among the strengths of the relevant parent company.
To a large extent, the share of these two types of activities determines and reflects a given
subunit’s strategic role within its MNC. This research examines (1) the patterns of MNC
subunits’ knowledge sourcing in terms of the technological and geographical dispersion
of knowledge sources; (2) the extent to which MNC subunits’ technological fields of
expertise are distinct from those of their parent companies, and how this technological
distinctiveness is related to their knowledge sourcing patterns; and (3) how MNC
subunits’ profiles of CC and CE activities (in terms of their overall technological distance
from their parent companies, and the degree to which they are engaged in CC versus CE
activities) evolve over time, reflecting the evolution of their knowledge creating role and
status within their international group. Attention is focused on the heterogeneity of firm-
specific evolutionary paths in the patterns of knowledge accumulation that support CC
iii
activities, controlling for the industry-specific determinants, location-specific factors, and
MNC group structural influences on such technological trajectories.
This study proposes a dynamic model in which the extent to which a subunit is likely to
take up CC activities is influenced by the technological and geographical dispersion of
that subunit’s knowledge sourcing. The results show a consistently positive relationship
between the technological and geographical dispersions of knowledge sourcing, an effect
that is moderated by the extent of subunit specialization in general purpose technology
(GPT) fields, and the geographical proximity between dispersed knowledge sources. We
also find a positive relationship between the technological dispersion of knowledge
sourcing and the technological distinctiveness of subunits. However, the geographical
dispersion of knowledge sources has a negative relationship with subunit technological
distinctiveness. A typology of subunit strategic roles is proposed, based upon the
evolutionary trajectory of a subunit’s share of CC activities and its technological distance
from its parent company.
iv
Preface Acknowledgements
There is no easy word to describe this long journey of PhD studies. In this most important
part of my life, I experienced the ups and downs just like most of my fellow PhD
colleagues – the excitement of pursuing knowledge, the humbleness of encountering
great minds, the stress and struggling of facing challenges, and the indescribable feeling
when six years worth of work is finally ready to be presented. This was indeed the most
fruitful six years, during which I achieved many goals I had set up in the beginning, and
none of these achievements would have come true without the tremendous help, support,
and encouragement from many others who influenced me in one way or another.
I would first like to show the deepest appreciation to my committee chair Dr. John
Cantwell. I studied with Dr. Cantwell since the first year in my PhD program at Rutgers
Business School. His guidance in my pursuit of scholarship, his insights on issues I
encounter in research, as well as his patient and persistent assistant in every aspect of my
study, made this dissertation possible. Even during this last year, when I have started my
job as an assistant professor with a full teaching schedule while working on the
dissertation at the same time, Dr. Cantwell offered to meet and give me guidance on
weekends and even holidays, for which I cannot thank him enough.
My committee member Dr. Farok Contractor gave me insights on the theorization of this
dissertation, with his comments I was able to improve my overall research framework; Dr.
Michelle Gittelman challenged my data construction and research methodology, with her
insights I developed further understanding on patent data; Dr. Lucia Piscitello encouraged
v
me to improve theory building, and to examine further relationship among key variables.
These are just a snapshot of a long list of valuable comments and suggestions my
committee provided to me, for which I greatly thank them.
Aside from my dissertation committee, there are many people helped me along this
journey. Dr. George Farris was one of the first who guided me when I was working as his
teaching assistant. The Technology Management Research Center he directed provided
me funding for research and travels to conferences. I also received generous funding from
Eastern Asia Research Center directed by Dr. Cantwell, as well as Department of
Management and Global Business at Rutgers Business School. Dr. Nancy DiTomaso who
chaired our MGB department, and our secretary Dawn Gist, had always offered their kind
assistance during my years in the program.
In the meanwhile, I need to thank the countless numbers of colleagues and friends, whose
friendship and kindness have made my journey a pleasant one. Among all others, Anke
Pippenbrink, Robert McNamee, Ranfeng Qiu, Jun Li, Hao-hsuan Chiu, and Shengsheng
Huang had offered me incredible help and support.
Finally, I couldn’t have done this without the support from my loving family. My
wonderful husband Ritchie Kim gave me unconditional love and support in all possible
ways. Thank you for staying by my side, your love keeps me going. Mom and dad, thank
you for your support for all these years, hope I have made you proud!
vi
TABLE OF CONTENTS
ABSTRACT OF THE DISSERTATION ........................................................................................ ii
Preface ............................................................................................................................................ iv
TABLE OF CONTENTS ................................................................................................................ vi
LIST OF TABLES ........................................................................................................................ viii
LIST OF FIGURES ........................................................................................................................ ix
1. Introduction ............................................................................................................................. 1
2. Data Overview ........................................................................................................................ 11
2.1 Data Description ............................................................................................................ 11
2.2 Key concepts .................................................................................................................. 19
2.3 Sample Data from Dow Chemical, DuPont, and Novartis .............................................. 26
3. MNC Subunit Knowledge Sourcing ........................................................................................ 35
3.1 Knowledge based view of MNC ..................................................................................... 35
3.2 Local and Remote Knowledge Sourcing ......................................................................... 39
3.3 Knowledge Sourcing Patterns and Subunit Innovation ................................................. 43
3.4 Framework ..................................................................................................................... 51
3.5 Methodology .................................................................................................................. 51
3.6 Results and Discussion ................................................................................................... 53
4. Competence Exploiting and Competence Creating ............................................................... 58
4.1 Competence Exploiting vs. Competence Creating ......................................................... 58
4.2 The Relationship between Knowledge Sourcing Patterns and Competence Creating .. 60
4.3 Framework ..................................................................................................................... 67
4.4 Methodology .................................................................................................................. 68
4.5 Results and Discussion ................................................................................................... 71
5. The Evolution of Subunit Roles .............................................................................................. 79
5.1 Typology Based on Knowledge Sourcing Pattern .......................................................... 79
5.2 Typology Based on Innovation Pattern – CC intensity vs. sub distance ........................ 86
5.3 Evolution of Subunit ....................................................................................................... 91
6. Conclusion .............................................................................................................................. 95
vii
APPENDICES ............................................................................................................................. 100
Appendix A1 – Firm List ........................................................................................................... 100
Appendix A2 - Country and patent distribution list ................................................................. 105
Appendix A3 – Geographic distribution of patents – by parent companies and by foreign subunits .................................................................................................................................... 107
Appendix A4 – 31 Technological Fields and Corresponding Patent Class Code ...................... 108
Appendix B – Table of Correlation for Key Variables in Study 1 .............................................. 111
Appendix C – Table of Correlation for Key Variables in Study 2 .............................................. 112
REFERENCES ............................................................................................................................ 114
CURRICULUM VITAE .............................................................................................................. 121
viii
LIST OF TABLES
Table 1 Geographic distribution of patents ................................................................................... 17Table 2 Technologic distribution of patents .................................................................................. 18Table 3 Distribution of Dow Chemical patents .............................................................................. 29Table 4 DuPont patent distribution ............................................................................................... 30Table 5 Novartis patent distribution .............................................................................................. 32Table 6 Hypothesis 1 Model with all data ...................................................................................... 53Table 7 Hypothesis 1 Model without host country sources .......................................................... 54Table 8 Hypothesis 1 Model without home country sources ........................................................ 54Table 9 Results for H2-H5, Study 1 ................................................................................................ 56Table 10 Results for Study 2 - All Data ........................................................................................... 72Table 11 Results for Study 2 - Large Subunit Sample .................................................................... 73
ix
LIST OF FIGURES
Figure 1 Overall Framework of study ............................................................................................... 9Figure 2 Geographic distribution of parent company locations .................................................... 15Figure 3 Dow Chemical patents invented by subunits ................................................................... 30Figure 4 DuPont patents invented by subunits ............................................................................. 31Figure 5 Novartis patent invented by parent and subunits ........................................................... 33Figure 6 Framework of Study 1 - Knowledge sourcing patterns .................................................... 51Figure 7 Conceptual Framework of Study 2 ................................................................................... 68Figure 8 Subunit typology by knowledge sourcing directions ....................................................... 81Figure 9 Subunit Typology by Competence Creativeness .............................................................. 87Figure 10 Subunit Distance and CC Intensity (All Data) ................................................................. 90Figure 11 Subunit Distance and CC Intensity (Large Subunits) ...................................................... 90Figure 12 Categorization of Subunit Evolution Patterns ................................................................ 91Figure 13 Subunit Evolution Patterns ............................................................................................ 92
1
1. Introduction
Knowledge, or in a lot of cases its application form, technology, is the driving force of
world development. Creation of new knowledge is often times the result of accumulation
and combination of existing knowledge. Because of the “stickiness” of knowledge, the
transfer process could be in fact difficult and time-consuming (Szulanski, 1994).
Researches on the nature of knowledge have shed light on organization’s capability of
creating, transferring, absorbing, and recombination of knowledge (Polanyi, 1966; Winter,
1987; von Hippel, 1988; Kogut and Zander, 1992; Nonaka, 1994; Szulanski, 2000; Gupta
and Govindarajan, 2000).
Traditional views of multinational corporation (MNC) consider the internal (most of the
time top-down) transfer of knowledge to be the dominating type in regards to subunit
knowledge accumulation. Due to the very nature of knowledge – being consist of both
explicit and tacit components, it is considered to be organizationally embedded, which
makes it difficult for inter-firm instead of intra-firm knowledge transfer (Kogut, 1988).
Knowledge transfer of this type relies largely on an intra-firm network. Gupta and
Govindarajan (1991, 2000) constructed typologies of MNC subsidiary roles based on the
transfer of knowledge throughout this inter-firm network. However, later studies have
shown that it is possible and necessary for external knowledge transfer. For instance,
Mowery and Oxley (1996) argued that the forming of strategic alliance and the building
of absorptive capacity could facilitate inter-firm transfer.
2
Although the traditional view of MNC knowledge transfer does shed light on one way
that MNCs tend to organize their innovation systems, it doesn’t fully appreciate the
newer development of subunit roles. Over the past few decades, foreign subunits are no
longer contend to remain in a status of a simple manufacturing center; new knowledge
can be created in subunits, innovations can be initiated in subunits, further along these
innovative technology can be transferred to elsewhere within the MNC, including the
headquarters and other subunits. This process has gained grown importance since it was
first spotted by researchers on MNC knowledge transfer and subsidiary external
embeddedness in the 1990s (Gupta and Govindarajan, 1991; Andersson and Forsgren,
1996).
In this relatively new development of views of MNC knowledge transfer, subsidiaries are
considered to rely, more heavily than before, on an inter-firm network, which involves
actors outside of an MNC’s firm boundary. Subsidiaries draw on external knowledge
sources aside from their own MNC intra-firm networks, expanding their knowledge
horizon, extending their technological trajectories into new sectors, and hence bring in
newly combined knowledge to create its unique capability. Since a company usually has
limited resource to be allocated on strategic development, it is sometimes difficult for
such subsidiaries to give strategic focus on both the intra- and inter-firm knowledge
networks. Moreover, aside from the firm affiliation of knowledge sources, the
geographical boundary plays a role in the decision of knowledge sourcing (Pearce, 1989;
Cantwell, 1995). Although modern development of information and communication
technology has vastly enabled and speeded up ways of communication, giving knowledge
3
seekers greater ability to seek for various knowledge sources from dispersed geographical
locations, it is still widely believed that the transfer of technological knowledge happens
more effectively and efficiently when the physical distance between the originator and
recipient is relatively close. Meanwhile, as the advancement of science and technology,
new development of technology requires more and more comprehensive complex
knowledge that often combines disciplines from previously unrelated fields. Research has
shown that in order to cope with uneven technological development levels and
unpredictable product interdependency, firms tend to extend and diversify their
knowledge to areas more than what’s needed for their production, thus firms tend to have
a larger patent footprint than their actual product footprint (Brusoni, Prencipe and Pavitt,
2001). In such cases knowledge sources may become more dispersed into different
technological fields.
It is known that large MNCs tend to have multiple operational units overseas, sometimes
more than one subsidiary could be set up in a single foreign host country. From a
business standpoint this could make sense when that particular host country provides a
wide range of geographical locations suitable for foreign investment, and these
investments could fall into different divisions of operation, which makes setting up
multiple subsidiary entities necessary; or for certain legal or political reasons MNC needs
to divide up their foreign operation into different functional companies, sometimes these
registered companies may even only exist for the purpose of keeping the right paperwork.
For these various reasons, it has been difficult for research on subsidiary level of MNC to
match the company activities in one particular geographical location to the real
4
subsidiaries of the MNC. In this study, an MNC subunit is defined as an operational unit
of an MNC outside its home country. Instead of looking at the organizational structural
sense of an MNC’s foreign entity, the focus of this study is on the innovative activities of
the MNC that are taking place outside the home country of the parent company. Parent
company, in this study, refers to the part of MNC that conducts innovative activities in
the home country. Thus we have our first research question based on the concepts of
MNC subunit and its knowledge sourcing:
Research Question 1: what are the patterns of knowledge sourcing of an MNC subunit
with respect to its technological dispersion, geographic dispersion, and level of sourcing
concentration in its two main sources – home country and host country?
This question examines the precedents, or input, of a firm’s knowledge accumulation,
which further indicates its own knowledge creation or innovation. One would ask then,
what could be the impact of knowledge sourcing on the outcome of innovation? In the
following study we focus on the outcome of innovation and the relationship between the
input and outcome. Following previous works on competence creating (CC) and
competence exploiting (CE) MNC subsidiaries (Cantwell & Mudambi, 2005), the
outcome of innovation is examined here in the form of either CC or CE types of activities.
CE types of activities indicate that the subunit is following the technological trajectory of
the MNC group and further developing the firm’s strong fields of technology. CE
activities themselves are nevertheless innovative activities, just that they don’t emphasize
the novelty of these innovations from the subunit’s parent company. CC types of
5
activities focus on creating competence in technological areas that are new to the
corporate group, or in technological areas that use to be a relatively weak area for the
group. Subunits focusing on CC types of activities have a better chance of creating future
competence for the MNC group, hence might gain a strategic importance within the
group. But one has to note that heavily investing on CC activities too distant from the
corporate group’s core technological trajectory may result in the subunit or particular
business unit falling out of the loop.
In examining the capability of competence creating, this part of the study focuses on the
ability of a subunit to not just be innovative, but also be innovative in a way that’s
different from its parent company. This distinction could be externally driven, such as a
mandate from the parent company, or a local specialization providing available pool of
knowledge that’s related but different from the MNC group’s area of expertise; it can also
be internally driven – for instance a subunit’s management sees the local opportunity and
has the autonomy to investigate into something new and different. This technological
novelty could mean two things to subunits’ knowledge sourcing: for one thing, the new
threads of technology development would push for a different direction of knowledge
sourcing from the parent company, so that it can provide necessary foundations for
distinctive innovation; for another, wider sourcing of knowledge would in return
reinforce the capability of subunit to create distinct innovations compared to its parent
company. This study defines the term technological distinctiveness as the extent to
which a subunit’s level of technological specialization in a given field is distinct from its
parent company. Hence we have our second research questions:
6
Research Question 2: how does a subunit’s knowledge sourcing pattern influence its
local competence creating capability, hence the technological distinctiveness from its
parent company?
Although we study the impact of knowledge input on competence creating or competence
exploiting activities, we have to note that composition of CC or CE activities would in
turn influence the strategic role of the focal subunit, which may result in a shift of MNC
mandate on this subunit, hence reshape its knowledge sourcing pattern. For instance, a
subunit may start with a competence exploiting mandate, which indicates that its
knowledge sourcing pattern follows those that dedicate on exploiting MNC competency,
with a relatively low level of technological diversification; this in turn reinforces this
subunit’s strength in a concentrated area, making it harder to create competency that’s
new to the corporate group. For this reason, we cannot simply view the relationship
between knowledge input and output as a static one. Over time, their mutual influence
should demonstrate a co-evolutionary pattern that could be captured by examining their
relationship longitudinally.
There are two dimensions of subunits’ innovative activity: CC intensity – the extent to
which a subunit’s innovative activities are composed of CC as opposed to CE, and
subunit technological distance – the extent to which a subunit’s innovations are
technologically distant from its parent company’s innovations. The combination of these
two dimensions demonstrates a subunit’s positioning in competence creativeness in its
international corporation group. Our third research questions tries to look at the
evolutionary pattern of subunit strategic roles based on these two dimensions:
7
Research Question 3: if knowledge sourcing and competence creation are two mutually
interdependent processes, is there a typology of evolutionary trajectories followed by the
paths of technological growth of subunits?
Chemical industry, as one of the traditionally considered knowledge-intensive industries,
rely its growth heavily on the accumulation and recombination of knowledge. This
dissertation has chosen the general chemical industry as the object of research, because of
its fully matured development as an industry over hundreds of years, yet widely spread
into various sub-sectors of technologies with some of them being newly emerging within
the past few decades. Its historically consistent concentration on technology development
has brought in life not only in depth research of fields that are closely related to
traditional chemical industry, but also streams of newly emerged knowledge in fields that
are expanded around the main sectors, such as pharmaceutical, biotechnology, etc.
Research has shown that compared with many other industries, chemical industry
consistently emphasizes on the protection of intellectual property by the means of
patenting; the analysis of patenting behavior in chemical industry can to a large extent
reflect the actual advancement of technology.
Therefore this dissertation is going to focus on large multinational corporations in the
broadly defined chemical industry, including general chemicals, petroleum and refining,
pharmaceuticals, and biotechnology. The studies use USPTO patent data to analyze the
relationship between technology fields, citations, and geographic location of inventors.
Three studies are designed to work out the theoretical framework of subunit knowledge
sourcing pattern and competence creation activities:
8
1) Study one is an explorative study with the purpose of understanding the patterns
of MNC subunit knowledge sourcing, theorizing and examining the relationships
between a subunit’s technological diversification degree, and the technological
diversification and geographic dispersion of their knowledge sources, controlling
for the effect of intra-firm and inter-firm sources – whether the sources are from
different parts of the MNC group or originated external to the firm.
2) Study two further explores the influence of patterns of MNC subunit knowledge
sourcing on competence creativeness of subunits. I examine on technological field
level the extent to which a subunit’s activity is co-specialized compared to its
parent company, then analyze to which extent this co-specialization is influenced
by the knowledge sourcing patterns found in study one. This study looks more
closely at each subdivision of the broadly defined chemical industry,
distinguishing petroleum, pharmaceutical and biotech, from general chemicals.
3) Study three uses time serious analysis to examine the evolutionary pattern
between knowledge sourcing pattern and competence creativeness of MNC
subunit. Meanwhile, this study creates a typology of MNC knowledge sourcing
structure and a typology of strategic roles of MNC subunits.
Together, the three studies in this dissertation create a framework for a better
understanding of MNC subunit knowledge sourcing pattern and competence creating
activities. A comprehensive overview of this framework is demonstrated in Figure 1.
9
Figure 1 Overall Framework of study
Contributions of this dissertation are three fold. Theoretically, the studies contribute in
areas of MNC knowledge accumulation and competence creating by establishing new
connection between the two and further on identifying evolutionary trajectory of MNC
subunit strategic role development. Empirically, the studies created new ways of studying
MNC subunit innovation, as well as a clear typology of subunit role based on its
competence creativeness. It also examines the relationship between subunit knowledge
sourcing and innovative activity, which could lead to further development of study on
these two interdependent characteristics of subunit strategy. Finally, methodologically,
the studies propose a combination of two dimensions – subunit technological distance
and CC activity intensity – to indicate strategic positioning of the subunit within an MNC
international network, this method takes into consideration of the composition of subunit
10
overall innovativeness and its novelty compared to parent company, which gives us a
comprehensive view of subunit competence creativeness.
In the next chapter, I will describe the construction and characteristics of my main dataset,
as well as some key variables that’s been introduced in this chapter. A case example of
Dow Chemical is given to demonstrate the utilization of data. The following three
chapters will be dedicated to studies one, two and three, respectively; then I will conclude
the dissertation with some final discussions in the last chapter.
11
2. Data Overview
In this chapter, I will describe the construction of my dataset, provide definitions of
several key concepts, and set an example with a typical firm in my sample – Dow
Chemical.
2.1 Data Description
Patent dataset construction
The studies in this dissertation are based on innovative activities of multinational
corporations in Chemical industry represented on patent data granted by US Patent and
Trademark Office (USPTO). The reason to choose USPTO patent data to reveal
innovative activities is twofold. For one thing, US patent data is comprehensive, granted
to assignees across the world, which represents the majority of innovative activities
conducted by all types of individuals and organizations in any country. Under the current
trend of fast movement towards globalization, it becomes increasingly important for
companies, especially MNCs, to obtain recognition from a universally accepted
intellectual property protection mechanism, and US patenting system fits this need. As a
result, although the basis of the patent office is in US, but it in fact represents patenting
activities from all over the world. The latest statistical data provided by USPTO shows
that among all of the 3,433,074 patents granted during the period of 01/01/1987 –
12/31/2011, about 52% or so are US originated patents, the other 48% or so are all from
foreign origins. Another reason is that patent data, although some would argue is a post
ante way of examining innovative behavior, and it’s inevitable that there are noises of
12
patenting behavior that are not solely for the purpose of protecting intellectual property
hence being an accurate proxy to innovative activities, studies have shown that as a
objective dataset patent data has the advantage of providing universally comparable
indicators of both innovative activity of firms and connections of these innovative
activities in terms of their accumulation paths (demonstrated by the citations of patents).
Discussion on the using of patent data has led to a specific issue in regards to patent
continuation. Continuation patent application is a unique situation to the US patent
system, it indicates situations when a patent application is a parent to one or more
applications; the continuation applications have the same disclosure as the earlier field
parent application and they claim the benefit of the filing date of the earlier application
(Eisenberg, 2000). One of the key conditions of filing continuation application is that at
least one inventor must be in common between the parent application and the
continuation application. Although it is arguable that continuation helps protecting
“pioneering inventors” in fields with high uncertainty, economists have been long
criticizing the abuse of continuation patents (Lemley & Moore, 2003). In this study, it is
concerned that in a lot of cases that there might be multiple patents in one “patent family”
filed as continuation patents but in fact all belonging to the same invention project.
Therefore, within a dataset of a company’s patent data, there might be ones that belong to
one family, which could result in high interdependence between patents. However,
previous study has shown that the type of continuation mainly used in chemical industry
(which has the character of being R&D intensive) is called Continuation-In-Part (CIP).
CIP patents tend to cover more “valuable” invention compared to the other types of
13
continuations (Hedge, Mowery & Graham, 2007)1
The patent data records in this dissertation come from mainly three sources: NBER US
patent database, patent database developed by Dr. John Cantwell, and some records
pulled directly from the USPTO patent database. The data consists of US patent granted
to 147 large corporations in the general chemical industry between 1976 and 2006.
General chemical industry as the object of research because of its fully matured
development as an industry over hundreds of years, yet widely spread into various sub-
sectors of technologies with some of them being newly emerging within the past few
decades. Its historically consistent concentration on technology development has brought
in life not only in depth research of fields that are closely related to traditional chemical
industry, but also streams of newly emerged knowledge in fields that are expanded
around the main sectors, such as pharmaceutical, biotechnology, etc. Research has shown
that compared with many other industries, chemical industry consistently emphasizes on
the protection of intellectual property by the means of patenting; the analysis of patenting
. In other words, due to the specific
characteristic of chemical industry, continuation patents contain unique new information
that differs from the “parent patent”. Although there might still be interdependence
between continuation patents and their parent patent, to some extent it represents the
continuous natures of innovative activities (along with other efforts besides innovation by
patent attorneys). Further discussion will be brought up again towards the end of this
dissertation in Chapter Six.
1 There are three types of continuations – Continuation Application (CAP), Continuation-In-Part (CIP), and Division. CAP must have the same disclosure as the original application; CIP includes a substantial portion or all of the original application but adds new disclosure to it; Division claims one part of the invention from the original application (Hedge et al., 2007).
14
behavior in chemical industry can to a large extent reflect the actual advancement of
technology.
One of the difficulties of using company patent data from a general patent database is the
attribution of patents to firms. Although each patent has one or multiple assignee
indicating its affiliation with an organization, these assignee codes are not a reliable
source of identifying the actual subunit where the patent is invented. Some firms may
have multiple assignees for their patents due to policy reasons; some may attribute all
patents to their parent company even when the patents are in fact invented in a foreign
location. In order to prevent attributing patent related innovative activities to the wrong
location, and to select the right patents and attribute them to my target firms’ foreign
operations, there are several steps to take. First, I identified the largest firms in the
general chemical industry from Fortune 500 listings and according to records of historical
trends. Only companies with a consistent record of high performance throughout my
research time period are chosen into the sample.2
2 Some firms are later dropped from the analysis sample in studies one and two if they are relatively new (have patenting records for less than 10 years during the researching period of 1976-2006), or have very insignificant patenting activities throughout the period (less than 50 patents granted during the entire thirty-one years).
However, these patents from newer or
smaller firms are counted into the total number for the analysis of chemical industry
innovative activity technology and geographic distribution. The selection of industry
includes firms focusing on general chemical, pharmaceutical and biotechnology,
petroleum refining and related sectors. Countries of origin of these firms include
developed nations like US (71), Germany (8), UK (10), Japan (28), etc., and developing
15
nations like Israel (1), Taiwan (2).3
A list of these companies is available in Appendix A1.
Figure 2 demonstrates the distribution of these 147 firms’ home country (parent company)
locations.
Figure 2 Geographic distribution of parent company locations
Once the target firms are identified, the second step is to map out all affiliates of these
firms. I used the resource of Who Owns Whom to manually input all names of affiliates
for my target firms. Then a match is run between these company names with the standard
company name data on the NBER database, hence extracting out all possible assignee
codes that’s associated with target firms. Each target firm may have various number of
assignee codes, depending on the firm’s policy on patent ownership. However, the
location of these assignees (as provided in the NBER dataset) doesn’t correctly reflect the
location of the invention (innovative activity). The purpose of extracting these assignee
3 The numbers in the parentheses reflects the number of firms in those countries.
Parent Company Distribution
USA
Germany
UK
Italy
France
Japan
Netherlands
Belgium
Switzerland
Denmark
16
codes associated with our target firms is solely to further extract all patents assigned to
these assignees, hence our 147 firms.
Therefore, the third step is to extract all patents assigned to the assignee numbers
identified in step two, which allows us to obtain the full patent dataset of our target firms.
From here on, I was able to extract all 286,667 patents that are assigned to my target 147
firms in the general chemical industry from the NBER patent database.
Then, the fourth step is to identify for each of these patents the first inventor’s country
location. Location of first inventor is a better indicator of the actual location of innovative
activity. This geographical data is extracted from Dr. Cantwell’s proprietary patent
database. The geographic location of first inventors in those patents attributable to
international corporate groups can then be used to examine the geographical distribution
of the technological activity of these firms (Cantwell, 1995). Each patent invented in a
certain country is an indicator of innovative activities by a firm in that country. Subunit,
in this research, is defined as the patent-country combination indicating MNC operation
in a country, which means the country location of the patent’s first inventor indicates the
host country location of the subunit. Table 1 demonstrates a distribution of patents
invented in these countries. Column two shows the overall patent distribution (patents
invented in those countries), column three and four indicates the number of parent
companies in each country and the patents invented by these parent companies; column
five and six indicates the number of foreign subunit in each country and the patent
invented by those subunits. A more comprehensive distribution table of patent numbers
in each country location (including both subunit patents and parent patents distribution)
17
can be found in Appendix A2. Pie chart figures demonstrating the distribution of patents
invented by parent companies and foreign subunits across geographical locations can be
found in Appendix A3.
Table 1 Geographic distribution of patents
Country Patents Parent Patents by
Parent Subunit Patents by
Subunit USA 153,916 71 130,854 62 23,062 Germany 40,341 8 25,768 93 14,573 UK 13,613 10 7,819 88 5,794 Italy 1,445 N/A N/A 65 1,445 France 9,646 7 7,965 70 1,681 Japan 47,776 28 45,240 69 2,536 Netherlands 1,083 2 460 60 623 Belgium 2,727 1 485 57 2,242 Switzerland 8,445 6 7,876 52 569 Denmark 1,167 2 992 30 175 Canada 2,427 2 67 71 2,360 Australia 526 2 79 57 447 South Korea 674 2 151 16 523 Other Countries 2,881 6 506 465 2,375 Total 286,667 147 228,262 1255 58,405
Technological fields of patents
The unit of analysis is technological fields (studies one and two) – innovative activities
conducted by a certain MNC subunit in particular technological fields – and MNC
subunits (study three) – the combination of parent-country indicates an MNC’s operation
unit in a certain country. The primary field of technological activity of each patent is
derived from the US patent class system, and these fields are grouped into 56
18
technological sectors (Cantwell and Andersen, 1996). This grouping is established on the
basis of the class system, providing economic meanings for each of these fields. Due to
the limitation of industry sample, some fields among these 56 are not as highly
represented as others; as a result an analysis across these 56 fields of technological
sectors would somehow demonstrate a skewed result. Therefore this study further
grouped these 56 fields into 31 respective ones in order to have meaningful representative
numbers for each technological field so that there is a comparable structure of data
distribution. Table 2 demonstrates a brief description of each technological field and
patent frequencies of these fields in this research. A detailed description of the
relationship between 31 technological fields and the corresponding patent class and sub-
class codes can be found in Appendix A4.
Table 2 Technologic distribution of patents
tech31 Field Freq. 1 Food and tobacco products 1,735 2 Distillation processes 738 3 Inorganic chemicals 4,139 4 Agricultural chemicals 5,035 5 Chemical processes 13,636 6 Photographic chemistry 9,461 7 Cleaning agents and other compositions 15,082 8 Synthetic resins and fibres 37,678 9 Bleaching and dyeing 3,130
10 Other organic compounds 48,999 11 Pharmaceuticals and biotechnology 57,747 12 Other chemicals and Related - disinfecting, preserving, textiles and explosives 815 13 Metallurgical processes 2,488 14 Miscellaneous metal products 4,080 15 Chemical and allied equipment 10,947 16 Paper making apparatus 1,808 17 Assembly and material handling equipment 1,694
19
18 Other specialised machinery 3,553 19 Other general industrial equipment 2,813 20 Mechanical engineering nes 3,870 21 Electrical devices and systems 2,146 22 Other general electrical equipment 4,427 23 Office equipment and data processing systems 2,431 24 Electrical equipment nes 4,290 25 Transport equipment 896 26 Rubber and plastic products 4,056 27 Non-metallic mineral products 14,143 28 Coal and petroleum products 6,579 29 Photographic equipment 2,417 30 Other instruments and controls 14,019 31 Other manufacturing and non-industrial 1,815
Total
286,667
It is noticeable in Table 2 that the patents used in this study are highly concentrated in
chemical related technological fields, and it also shows a concentration on some non-
chemical fields that are considered general purpose technology (GPT, shown in bold
letters in Table 2). This distribution is as expected with our selection of industry as well
as the utilization of GPT fields.
2.2 Key concepts
Knowledge Sourcing and Patent Citations
The use of patent citation as an indicator of knowledge flow has been questioned by a
serious of research since there is a large portion of citations added by the examiner rather
than by inventors of the patents (Alcacer & Gittelman, 2006; Alcacer, Gittelman &
Sampat, 2009). One cannot argue that those examiner-added citations can be a good
indicator of knowledge “flow”. This study, however, is not trying to measure the actual
“flow” of knowledge. The definition of knowledge sourcing in this dissertation is the
connection of a subunit with direct or indirect sources of knowledge components that
20
helps in the process of recombination and generating new knowledge. This concept of
knowledge sourcing emphasizes the implicit connection between different knowledge
bodies; therefore patent citation - even with examiner added ones – is a good indicator of
knowledge sourcing. If there is a citation linkage between two patents, it means there is a
direct or indirect influence from one to another.
Technology – Diversification, Dispersion, Distinctiveness, and Distance
There is three “D”s about technology – technological diversification, technological
dispersion, and technological distinctiveness – that will be used throughout this
dissertation. It is important to clarify the meanings of each one of these terms before we
move on to the actual analysis.
Technological diversification and technological dispersion have similar constructions
from the point of data. However, the meanings behind them are different. Technological
diversification focuses at the innovative activity itself. By engaging in diversified
technological activity, firms are actively expanding their operations across different types
of productive activity (Cantwell and Piscitello, 2000). Technological diversification is
therefore considered a means of exploring and accumulating corporate competency. From
the data point of view, technological diversification indicates the extent to which a
subunit’s innovative activities (patents) are wide spread across various technological
fields. Technological dispersion, on the other hand, focuses on the dispersed
technological composition of knowledge sources. With respect to the data, technological
dispersion indicates the degree of knowledge sources being widespread across different
21
technological fields. While technological diversification focuses at the subunit itself, it is
an intrinsic measure of the subunit’s strategic intension in regards to fields of
specialization, different fields among a subunit’s expertise could be interrelated to one
another, and there is a meaningful combination of these fields; technological dispersion
focuses at the extrinsic measure of a subunit’s knowledge sources, it is merely a
demonstration of fields that could contribute to a subunit’s technological innovation, and
these fields are not necessarily all interrelated with one another.
That being said, the actual measurement of these two concepts is the same – both are
measured by the reverse of Herfindahl-Hirschman index (HHI) of concentration. HHI
was first used in 1951 to analyze the concentration within the steel industry, calculated by
adding the squares of firm’s market shares in per cent (Weinstock, 1982).
𝐻𝐻𝐻𝐻𝐻𝐻 = �𝑆𝑆𝑖𝑖2𝑛𝑛
𝑖𝑖=1
Where si is the market share of firm i measured in percentage points. This definition
implies that the shares of the larger firms are given greater weight than those of the
smaller companies. The result of this original form of HHI should vary between 0 and 1,
a higher value of HHI indicates higher level of concentration in the market. Taking from
this point of analysis, if the si in the calculation is share of technological innovation
instead of market share, then this index could be used to demonstrate the technological
concentration of given industry. Then we deduct this result from 1, which gives us
reversed result of technological concentration. This could then be used to indicate the
22
diversification/dispersion of technology – depending on whether we are examining the
knowledge bodies themselves or the knowledge sources. Hence, my calculation of
technological diversification/dispersion is:
𝑇𝑇𝑇𝑇𝑇𝑇ℎ𝐷𝐷𝑖𝑖𝐷𝐷𝑖𝑖 = 1 − �(𝑃𝑃𝑖𝑖𝑖𝑖∑ 𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖
)2
31
𝑖𝑖=1
In which TechDivi indicates the technological diversification of subunit i, Pij is the
number of patents a subunit invented in technological field j, and ∑ 𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖 sums up the total
number of patents in all field invented by this subunit.
For technological dispersion, the formula is:
𝑇𝑇𝑇𝑇𝑇𝑇ℎ𝐷𝐷𝑖𝑖𝐷𝐷𝐷𝐷𝑖𝑖 = 1 − �(𝑃𝑃𝑖𝑖𝑖𝑖∑ 𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖
)2
31
𝑖𝑖=1
Since this is for knowledge sourcing data, we want to examine the technological
dispersion of knowledge sources of a subunit’s activity in each field, so the analysis is at
technological field level. Among all the citations by a subunit’s patenting activity in each
field, we construct the similar calculation method as technological diversification, only
this time Pij on the right hand side of the equation indicates number of cited patents in
technological field j.
Technological distinctiveness, the main variable used in study 2, is another concept
that’s completely different from the previous two technological related terms. The notion
23
of technological distinctiveness comes from the idea of creating a variable to examine the
extent to which a subunit’s specialization in certain technological field is different from
its parent company, which can serve as an indicator of that subunit’s competence creating
activities. Before I explain the rationale behind the construction of such variable, I need
to first introduce a concept that is needed in understanding the mechanism behind my
measure – RTA.
Revealed Technological Advantage (RTA) is an index that demonstrates the level of
technological specialization for one unit of observation (a subunit, a firm, or even a
geographic location – country or region) in a given technological field, it was first
pioneered in the work of Soete (1980) and then further developed by Patel and Pavitt
(1987, 1991) and Cantwell (1989, 1995). The RTA index of a firm in a particular
technological field is given by the firm’s share of patenting in that field divided by its
share of patenting in all sectors and is defined as follows:
𝑅𝑅𝑇𝑇𝑅𝑅𝑖𝑖𝑖𝑖 = (𝑃𝑃𝑖𝑖𝑖𝑖 /� 𝑃𝑃𝑖𝑖𝑖𝑖 )/(� 𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖
/��𝑃𝑃𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖
)𝑖𝑖
In which is the total number of patents of firm i in field j. The index varies around unity,
such that values greater than one suggest a firm’s comparative advantage in the field of
activity in question relative to other firms in the same industry, while values less than one
are indicative of a position of comparative disadvantage (Cantwell and Piscitello, 2000).
When this is calculated at a subunit level, the unit of analysis is then changed from firm
to subunit – a firm’s activities within a foreign host country. In the meanwhile, if we
enlarge the unit of analysis to a country level, then this index could be used to reveal
24
comparative advantage or disadvantage of a country’s activities in certain technological
fields.
While RTA index illustrates the technological advantage/disadvantage positions of a
subunit in certain fields, we still need to apply a mechanism to compare the positioning of
both a subunit and its parent company, which then leads to our operationalization of
technological distinctiveness – the extent to which a subunit’s degree of specialization in
a certain technological field is distinct from its parent company. To put it short, we want
a method to effectively compare the RTA between a subunit and its parent company in
each field. Here we took a methods used by Cantwell and Santangelo (2002) that’s
designed to assess the co-presence or absence of a matching degree of specialization
between the two groups of firms in each patent class. This method draws on the measure
of intra-industry trade across sectors in the international trade literature (Grubel and
Lloyd, 1971, 1975). The Grubel-Lloyd index generates a variable GLi by comparing the
export and import of good i:
Where Xi measures the export, and Mi measures the import of a good i. If GLi = 1, there is
only intra-industry trade, whereas if GLi = 0, there is only inter-industry trade. Cantwell
and Santangelo (2002) further developed this model as a measure of co-presence of a
matching degree of specialization between the two groups of firms (k) (where k = 1 or 2)
in a particular ICT patent class (c) in each region (r):
25
Crc = 2 min (RTA1rc, RTA2rc) / (RTA1rc + RTA2rc); 0 ≤ Crc ≤ 1
In my study, however, to examine technological distinctiveness, I use the reverse of co-
specialization, which is:
𝑇𝑇𝐷𝐷𝑖𝑖𝑖𝑖 = 1 −𝑅𝑅𝑇𝑇𝑅𝑅𝑖𝑖𝑖𝑖 + 𝑅𝑅𝑇𝑇𝑅𝑅𝐷𝐷𝑖𝑖 − |𝑅𝑅𝑇𝑇𝑅𝑅𝑖𝑖𝑖𝑖 − 𝑅𝑅𝑇𝑇𝑅𝑅𝐷𝐷𝑖𝑖 |
𝑅𝑅𝑇𝑇𝑅𝑅𝑖𝑖𝑖𝑖 + 𝑅𝑅𝑇𝑇𝑅𝑅𝐷𝐷𝑖𝑖=
|𝑅𝑅𝑇𝑇𝑅𝑅𝑖𝑖𝑖𝑖 − 𝑅𝑅𝑇𝑇𝑅𝑅𝐷𝐷𝑖𝑖 |𝑅𝑅𝑇𝑇𝑅𝑅𝑖𝑖𝑖𝑖 + 𝑅𝑅𝑇𝑇𝑅𝑅𝐷𝐷𝑖𝑖
Where 𝑇𝑇𝐷𝐷𝑖𝑖𝑖𝑖 denotes the technological distinctiveness of subunit i in field j, 𝑅𝑅𝑇𝑇𝑅𝑅𝑖𝑖𝑖𝑖 is the
revealed technological advantage of subunit i in field j, whereas 𝑅𝑅𝑇𝑇𝑅𝑅𝐷𝐷𝑖𝑖 represents the
revealed technological advantage of subunit i’s parent company in field j.
However, this measure of distinctiveness doesn’t specify the direction of distinction – a
large number (closer to one) of 𝑇𝑇𝐷𝐷𝑖𝑖𝑖𝑖 could indicate either the subunit is distinctively
more specialization in the field than its parent company, or the parent company is
distinctively more specialized than this subunit. In order to capture the effect of subunit’s
competence creativeness, one more step is needed to identify subunits that are more
specialized than their parent companies in the focal fields. Hence in actual data
processing, it is necessary to screen out situations when 𝑅𝑅𝑇𝑇𝑅𝑅𝑖𝑖𝑖𝑖 < 𝑅𝑅𝑇𝑇𝑅𝑅𝐷𝐷𝑖𝑖 , and
subsequently set the value of 𝑇𝑇𝐷𝐷𝑖𝑖𝑖𝑖 to zero, indicating that this subunit has no distinct
specialization in this field compared to its parent company.
Finally, the term technological distance between a subunit and its parent company
brings the analysis up to a firm/subunit level again. The calculation of this variable uses
the Jaffe (1986) approach which borrows the concept of Euclidean vector calculation
using (1 – Cosin Similarity) to construct a measure for distance:
26
𝑇𝑇𝑇𝑇𝑇𝑇ℎ 𝐷𝐷𝑖𝑖𝐷𝐷𝐷𝐷𝐷𝐷𝑛𝑛𝑇𝑇𝑇𝑇 = 1 −∑(𝑆𝑆𝑆𝑆𝑆𝑆 × 𝑃𝑃𝐷𝐷𝑃𝑃𝑇𝑇𝑛𝑛𝐷𝐷)
�∑𝑆𝑆𝑆𝑆𝑆𝑆2 × �∑𝑃𝑃𝐷𝐷𝑃𝑃𝑇𝑇𝑛𝑛𝐷𝐷2
Where Sub and Parent represents the number of patents in field j granted to the subunit
and parent, respectively; the ∑ in this equation is calculating the sum of patents
through all technological fields.
2.3 Sample Data from Dow Chemical, DuPont, and Novartis
This section of the paper will depict an overall picture of three selected chemical industry
firms: Dow Chemical (US), DuPont (US), and Novartis (Switzerland), to illustrate the
basic data composition of this research.
Overview of the three firms
The Dow Chemical Company is an international chemical enterprise, whose center of
operations lies in Midland, Michigan, USA. Dow Chemical is known for the
manufacturing of specialty chemicals, advanced materials, agrosciences and plastics.
Dow was founded in 1897 by Canadian chemist Herbert Dow who created a new method
of extracting bromine. Over the next twenty years, Dow was effective a rapidly
diversifying its product line and supplied many of the war materials during World War I,
which were previously imported from Germany. Dow expanded internationally in 1942
with the organization of Dow Chemical of Canada and followed suit in Japan in 1952
with Asahi Dow, it’s first subsidiary overseas and Dow Europe in Zurich, Switzerland.
By 1965, with additional plants created in Latin America, sales outside the U.S
constituted nearly 25 percent of total sales. By 1973, Dow became the first foreign
27
industrial company listed on the Tokyo Stock Exchange and sales outside of the U.S.
reached 47 percent. Currently, 5000 products are manufactured at 188 sites in 36
countries across the globe with a focus on six operating segments: Electronic and
Functional Materials, Coatings and Infrastructure Solutions, Agricultural Sciences,
Performance Materials, Performance Plastics and Feedstocks/Energy.
E. I. du Pont de Nemours and Company, more often referred to as DuPont, started their
business in the explosives industry from the very beginning of 20th century. Its
development experienced six phases: the founding period from 1902 to 1911, the
centralization and diversification from 1911 to 1921, decentralization in the period of
1921 to 1935, Nylon and new Nylon era from 1935 to 1960, the maturation period in the
1960s, and then the rethinking of corporate strategy since the 1970s (Hounshell, 1989,
1992). In the sense of the technological sectors, after starting from explosives, DuPont
soon diversified into other fields such as synthetic fibers and photographic chemicals,
with the driving force of the development of Nitrocellulose technology. In research
organizing, DuPont had a unique strategy of a general experimental laboratory – namely,
the Experimental Station – whose job was to pick up or create opportunities for
innovation in all other divisions in the corporation. This distinction between division labs
and the general lab was DuPont’s strategy of catching up with IG back to the early 20th
century, and which was proved to be successful – at least in the sense of facilitating the
technological diversification of DuPont. In more recent years, the company ranked 66th
28
in the Fortune 500 on the strength of nearly $28 billion in revenues and $1.8 billion in
profits.4
Novartis is a pharmaceutical company that has its center of research and development
based in Basel, Switzerland, with other key research sites located in Horsham, UK;
Vienna, Austria; Tsukuba, Japan; East Hanover, New Jersey, USA. Although the firm
was formed in 1996, its innovative activities dated far back in the 19th century. Merged
from Ciba-Geigy and Sandoz Laboratories, Novartis inherited long innovative history of
J.R. Geigy, CIBA, and Sandoz.
5
Dow Chemical patent description
With an emphasis on innovation, Novartis not only
vigorously develop its in-house R&D capability, but also seeks for global collaboration
possibilities. Their strategy for innovation is “focused diversification”, by which they
mean a concentration on pharmaceutical, biomedical innovation while encouraging
diversification into related areas. Novartis is now ranked as one of the top pharmaceutical
companies in the world, along with Pfizer, GlaxoSmithKline, Sanofi-Aventis, and Merck.
Dow Chemical and its affiliated subunits have been granted a total number of 12193
patents during the years 1976-2006 in the USPTO system, among which a majority
11568 (95%) are invented in the US, the rest 625 (5%) patents are invented by its 24
foreign subunits distributed around the globe. It is obvious that Dow Chemical clearly has
an innovation policy of home base concentration. A distribution of these patents in
different subunits is demonstrated in Table 3. As shown in the table, aside from its parent
4 "Fortune 500: 1955–2005.". CNN. Retrieved September 19, 2011 5 Adapted from "Company history", corporate website (Novartis), novartis.com
29
company, most of the innovative activities of Dow Chemical is concentrated in
developed countries such as Germany, UK, Netherlands, Canada, etc. Figure 3 illustrates
the distribution of patents invented by subunits.
Table 3 Distribution of Dow Chemical patents
Host Country of Subunit Number of Patents Germany 88
UK 86 Italy 13
France 48 Japan 46
Netherlands 81 Belgium 62
Switzerland 58 Sweden 3
Denmark 5 Ireland 1 Spain 6
Greece 2 Canada 89
Australia 5 India 4 Brazil 1 Chile 1
Colombia 2 South Korea 7
Taiwan 2 South Africa 9
Other Latin America 2 Other Asia 4
30
Figure 3 Dow Chemical patents invented by subunits
DuPont Patents
The patents from DuPont also demonstrate a concentrated pattern on its home country the
US. Among the 12969 patents granted during the period 1976-2006, about 90% were
invented by the parent company, with 1229 (about 10%) invented in foreign locations, as
shown in Table 4 and Figure 4.
Table 4 DuPont patent distribution
Host Country of Subunit Number of Patents Germany 283
UK 100 Italy 8
France 45 Japan 149
Netherlands 104 Belgium 53
Switzerland 74 Sweden 2
Denmark 1
Germany
UK
Italy
France
Japan
Netherlands
Belgium
Switzerland
Sweden
Denmark
31
Spain 3 Luxembourg 17
Austria 15 Norway 2
Czechoslovakia 1 USSR 4
Canada 311 Australia 15
New Zealand 1 Brazil 4 Israel 9
Argentina 1 Mexico 1
Venezuela 1
Figure 4 DuPont patents invented by subunits
DuPont has its foreign innovative activities concentrated mainly in Germany and Canada,
followed by other developed countries like Japan, UK, and Netherlands. To some extent,
compared to Dow Chemical, DuPont puts more emphasis on developing centers of
excellence in foreign locations. Dow Chemical, on the other hand, emphasizes a more
Germany
UK
Italy
France
Japan
Netherlands
Belgium
Switzerland
Sweden
Denmark
32
even development across its main foreign R&D locations. It is later shown in the analysis
of competence creativeness, in Chapter Four, that although an equal number of subunits
from both companies has reported competence creating activities, more of these subunits
from DuPont have indicated a continuous strength over the years in competence creating,
and these activities demonstrate a continuous trajectory of DuPont subunit’s areas of
expertise. For example, the subunit of DuPont in Canada has shown over the years a
continuous specialization in technological fields 13, 18 and 19, which represents areas
like metallurgical processes, specialised machinery, and other industry equipments. These are
not DuPont’s key areas of expertise as a corporate group, but their Canadian subunit was
able to develop unique strength in these areas to offer complementary technology
development for the firm.
Novartis
Novartis’s patent distribution is different from Dow Chemical and DuPont. Given that
this is not an US based firm, it is understandable then that Novartis doesn’t have as high a
concentration in US hosted innovation as the other two. Among its total number of 9148
patents in the period of 1976-2006, about 54% (4942) of them are invented in its home
country Switzerland, while the rest are invented in foreign subunits. Table 5 and Figure 5
depicts the numbers and shares of patent distribution.
Table 5 Novartis patent distribution
Host Country of Subunit Number of Patents USA 2160
Germany 894 UK 663
33
Italy 8 France 202 Japan 41
Netherlands 14 Belgium 6
Switzerland 4942 Sweden 7
Denmark 7 Spain 2
Austria 35 Norway 2 Canada 34
Australia 25 India 20 Brazil 2 Israel 1 Chile 1
Mexico 2 South Korea 1
China 1 Other Countries 3
Figure 5 Novartis patent invented by parent and subunits
USA
Germany
UK
Italy
France
Japan
Netherlands
Belgium
Switzerland
Sweden
34
Besides its home country, Novartis has extensive innovative activities recorded by
patents in US, Germany, UK, and France. Aside from regular R&D facilities in these
countries, it is discussed before that Novartis has set up key research sites in countries
outside of its home location. Some newer research centers are not demonstrating a strong
impact from the current dataset, due to the cut up time point of 2006. However, as can be
observed in these charts, in comparison with the previous two firms, Novartis has a more
evenly dispersed geographic composition of innovative activities. One of the research
concentrations of this dissertation is to examine whether this type of structure could
increase the chance of developing CC subunits. Results in later chapters are going to
answer this question. Interestingly, as shown in Chapter Five, Novartis is indeed one of
the few firms that have shown trends of subunit evolution over the years, although
different subunits tend to display distinct patterns of evolution, due to various external
and internal driving forces for innovation.
35
3. MNC Subunit Knowledge Sourcing
3.1 Knowledge based view of MNC
Theories of firm in regards to forms of hierarchy or market to organize production and
related economic behavior were discussed by the school of transaction cost economists
(Coase, 1937; Williamson, 1975). In the resource based view, the organizations are
considered to gain sustainable competitive advantages from the possession of a series of
resources, and hence achieve greater long-term performance (Penrose 1959; Barney 1986;
Grant 1991). Conventional approaches consider general sources of competitive advantage,
namely advantages in cost and differentiation (Porter 1985). Further empirical efforts
clarified the relationship between the organizations’ resources and capabilities, indicating
that when resources are scarce, they become valuable and this leads to temporary
competitive advantages (Schmalensee, 1983; Henderson and Cockburn, 1994; Powell and
Dent-Micallef, 1997; Spanos and Lioukas, 2001; and Powell, Lovallo and Caringal,
2006). If organizations can protect their main resources from imitation, transfer or
substitution, these competitive advantages can become sustainable (Hulland 2004), so
confirming the strategic decisions taken by the firm (Parker and Russell 2004).
Following the lead of Resource Based View, studies in the past few decades have
turned their focus onto a more specific type of resource – knowledge. Scholars have
investigated the concept of knowledge using various perspectives and methods. Polanyi
(1966) brought the notion of “tacit” knowledge versus “explicit” knowledge into
attention of the academia world. Since then the discussion about characteristic of
knowledge has been developed around this notion. Winter (1987) suggested that
36
knowledge could be understood in terms of four dimensions – (a) tacit-articulate, (b)
observable-not observable, (c) complex-simple, and (d) element in a system-independent.
Among these dimensions the one that emphasizes the tacitness nature of knowledge is the
most researched on by different strands of studies. Kogut and Zander distinguished
knowledge as information and know-how (Kogut & Zander, 1992). Information is
“knowledge which can be transmitted without loss of integrity once the syntactical rules
for deciphering it are know”, while know-how is “the accumulated practical skill or
expertise that allows one to do something smoothly and efficiently” (Kogut & Zander,
1992; von Hippel, 1988). Kogut and Zander used the term “combinative capabilities” to
describe the ability of the firm to "generate new combinations of existing knowledge"
and "to exploit its knowledge of the unexplored potential of the technology" (Kogut and
Zander, 1992, P.391). Nonaka’s notion of dynamic interaction between two dimensions
of knowledge transfer extended the thoughts of tacit vs. explicit knowledge, bringing
linkages between the transformations (“knowledge conversion”) of the two types of
knowledge and transfers among different entities that possess knowledge (Nonaka, 1994).
Indeed knowledge is inevitably considered a resource that is crucial for firm
technological capability development (Grant, 1996; Teece et al, 1997). The knowledge
based view of organizations take a “static” perspective in suggesting knowledge as a
strategic asset of an organization (Grant, 1996), where dynamic capabilities discusses the
creation of new knowledge through the interaction of existing tacit and explicit
knowledge (Teece et al, 1997). Upon the distinction between explicit and tacit
knowledge, scholars emphasizes that a globally dispersed R&D operations provide
37
MNCs with competitive advantages not available in single-country centralized R&D
operations (Dunning, 1995), and that such a competitive advantage is based on how
efficiently MNCs share knowledge across HQs and R&D subsidiaries (Gupta &
Govindarajan, 2000). In the context of within a multinational corporation network, Gupta
and Govindarajan (2000) identified five determinants of intra-corporate knowledge flows
(including outflows and inflows), namely the value of knowledge stock, the motivation
disposition to share/acquire knowledge, the existence and richness of transmission
channels, and the absorptive capacity.
In order to avoid confusion caused by the ambiguity of the term “network” and its
wide use across different fields, it is necessary to clarify the definition “network” and
“knowledge sourcing”. In the domain of social science, the origins of the concept of
network refers to a social structure made up of individuals (or organizations) called
“nodes”, which are tied (connected) by one or more specific types of interdependency,
such as friendship, kinship, common interest, financial exchange, or relationships of
beliefs, knowledge or prestige. Clearly, the term itself has a variety of meanings. The
research on organization studies, however, often restrict the term to an extent which
stresses that the formation of a network requires actual connections (most of the time in
the forms of personal communication or interaction) between the nodes. Such definitions
can be found in studies on business network, knowledge flows, technology diffusion, etc.
(Ingram & Roberts, 2000; Owen-Smith & Powell, 2004; Hansen & Lovas, 2004; Bell &
Zaheer, 2007).
38
Following this stream of research, the concept of “knowledge networks” is widely
used in studies on the first level of social networking activities, sometimes
interchangeable with the terms “knowledge flow”, or “knowledge transfer” (Best, 2001;
Britton, 2004; Van Geenhuizen, 2008). However, this study is looking at these network
connections of knowledge reflected on a different level – in this more abstract level of
knowledge network, the “nodes” are entities (usually the organizations) with a stock of
knowledge that’s a combination information and knowhow from different technological
fields, and the “ties” are knowledge connections among different entities in the forms
such as sharing similar knowledge base, applying related technologies, or building upon
one another. This level of connection does not necessarily require actual interpersonal
relationships, but instead it indicates the relatedness of knowledge between organizations,
mainly in the formal or informal manners of acknowledging the related knowledge that a
certain knowledge creation activity is based upon. Therefore, in order to distinguish my
work from that of many social scientific studies of knowledge network, I define this type
of connection as “knowledge sourcing” – the extent to which the knowledge of two
organizations (subunit level) is connected with one another in the process of production
and innovation via the sourcing of knowledge. The sourcing activity itself can be the
result of either direct or indirect knowledge transfer, but the focus here is the cumulative
expression of knowledge relationships. There may or may not be direct actual knowledge
“flows” between a subunit and its knowledge sources, but rather it reflects the inevitable
connection between a subunit’s innovative activities and the sources of knowledge that
accumulated to enable these activities.
39
Within an MNC network, the knowledge sourcing includes connections between
subunits and MNC headquarter, and among subunits in the same MNC group. This type
of knowledge sourcing is by definition international; it reflects the way a multinational
corporation is organized, as well as the pattern of innovative activities in various subunits
within that MNC. Outside of an MNC network, a subunit’s knowledge sourcing could
include connections with local actors, and with any other external sources from
international locations.
3.2 Local and Remote Knowledge Sourcing
The conceptualization of international integration of MNC has been the center of debates
among IB researchers. Earlier findings pioneered by the “product life cycle” model
(Vernon, 1966) suggest that foreign subsidiaries mainly exploit proprietary advantages
abroad generated in home base by the parent firm and thus technological capabilities in
foreign countries were concentrated on the adaptation of products to meet the particular
needs of local market (Ronstadt, 1977; Behrman & Fischer, 1980; Hakanson & Nobel,
1993; Porter, 1990).
More recently, an increasing stream of resource-based and network-based
theoretical and empirical research on multinational firm challenge the traditional view on
MNC value creation. According to them, the MNC is not just an exploiter of home
country knowledge abroad, but a vehicle integrating knowledge from different parts of
word (Ghoshal & Bartlett, 1991; Birkinshaw & Hood, 1998). The emphasis on the
internationally integration strategies at the MNC group level has also shifted from the
40
conventional demand-side potential (as markets) to the supply side potential (innovation
and technology) (Cantwell, 1992, 1993).
In the new context, from Bartlett and Ghoshal (1989)’s suggestion on
“transnational’-form MNCs, to Hedlund (1994)’s proposition of “N-form” corporations,
an increasing strand of literature suggests that MNCs are incrementally sourcing and
generating knowledge from various locations (Pearce, 1989; Cantwell, 1995) and sharing
them across the organization. This knowledge sharing is facilitated by the emerging
ability of multinationals to integrate knowledge across related technologies and across
geographic boundaries (Zander, 2002). In this sense, FDI is interpreted as a mechanism
through which firms seek and develop capabilities on a global basis (Kogut & Chang,
1991; Teece, 1992).
In terms of managing the so called “internal network (intra-firm)” MNCs,
Hedlund and Ridderstrale (1995) observed a closer integration in the international value-
added activities of MNCs in the 1990s and afterwards, and pointed out that the
managerial challenge has shifted from controlling bilateral relations between one
technology creating headquarters and many implementing subsidiaries to a more complex
integrated network in which each units have their own creative capabilities. Zander and
Solvell (2002) also suggested that the integrated but increasingly heterogeneous
innovation structure encourages increasing reciprocity in inter-unit relations. This is
because to the extent that these fields of local knowledge accumulation are themselves
complex or systemic, they may need to draw on wider international sources as well for
41
the complementary branches of expertise required, and these supporting lines of
knowledge may either come from headquarters or from other entities of the parent group.
This integrated network allows the subsidiaries generate innovations based on the
stimuli and resources from host countries (Prahalad & Doz, 1987; Barlett & Ghoshal,
1989; Cantwell, 1992, 1993; Dunning, 1998; Nobel & Birkinshaw, 1998). This
international integration on the one hand can facilitate high subsidiary involvement in the
formulation and implementation of the company’s R&D strategy, and on the other hand
will minimize the “duplication of effort” problems (Hakanson & Zander, 1988).
The embeddedness of MNC subsidiaries in their local environment has been
uncovered and emphasized in the past few years (Uzzi, 1996, 1997; Birkinshaw et al.,
2002). Uzzi (1997) suggested that embeddedness is a logic of exchange that promotes
economies of time, integrative agreements, Pareto improvements in allocate efficiency,
and complex adaptation. Economic behaviors are considered to be embedded in the social
structure (Granovetter, 1985). Yet, the system embeddedness of knowledge, in turn, is
affecting organization structure. System embeddedness is the extent to which the
knowledge is affected by the system or context in which it is embedded. It emphasizes on
the social system (Birkinshaw, Nobel, & Ridderstrale, 2002). It is implied that some
knowledge is much more sensitive to its social and physical context than other
knowledge. For example, Birkinshaw et al. (2002) find out that Ericsson’s software
development is undertaken according to well-established procedures that are relatively
easily replicated in different settings; therefore the sensitivity to physical location and
social context is low. On the other hand, Alfa Laval’s R&D in milking machines and
42
separators has always been undertaken in specialized locations that have dedicated
physical infrastructure with functional activities nearby; hence their underlying
knowledge is highly sensitive to location, and would be very costly to move.
Andersson and Forsgren have completed a series of research on the subsidiary
embeddedness issue (Andersson & Forsgren, 1996; Andersson, Forsgren, & Holm, 2001;
Andersson, Bjorkman, & Forsgren, 2005). They define subsidiary embeddedness as the
extent to which individual relationships in the local market can serve as a source of
knowledge (Andersson et al, 2005). As the level of subsidiary embeddedness increases,
the potential of control from headquarter is higher as it attempts to integrate the
subsidiary better into the overall corporate strategy (Andersson & Forsgren, 1996,
Andersson et al, 2005). They argue that the external technical embeddedness is positively
affecting the subsidiary’s expected market performance, and thus increasing the
subsidiary’s importance for MNC competence development, using absorptive capacity
theory (Andersson et al., 2001).
Strategic alliances also facilitate external knowledge sourcing. Different types of
alliances ranging from “relational” contracts to equity joint venture all have impact on
organizational learning (Contractor & Lorange, 2002). Previous research looked at how
knowledge is managed in strategic alliances (Inkpen, 2002; Martin and Salomon, 2002),
how knowledge is transferred across partners (Mowery et al., 1996; Simonin, 2004), and
how knowledge is acquired from the parents by the joint venture itself (Lyles and Salk,
1996), with an emphasize on how organizations learn from their partners and develop
new competencies through their collaborative efforts (Inkpen, 2002).
43
3.3 Knowledge Sourcing Patterns and Subunit Innovation
When analyzing patterns of knowledge sourcing, there are mainly two dimensions that
can distinguish the composition of knowledge sources – technological areas and
geographical locations. The former looks at different areas of technologies that the
knowledge sources are composed from – whether they are focused on certain particular
fields or dispersed from different areas; the latter examines the geographical locations of
these knowledge sources – whether the sources are geographically concentrated or
scattered across the world. Technically speaking, these two dimensions are not
necessarily correlated to one another. However, on the one hand, the complexity of
technological accumulation usually induces a more proactive search across geographical
boundaries for relative or complementary knowledge sources, whereas on the other hand,
a cross-regional dispersed composition of knowledge sources could more than likely
bring in a wide variety of technological expertise. Therefore I propose that there is a
positive relationship between technological dispersion and geographical dispersion of
knowledge sources for any MNC subunit. The more geographically dispersed their
knowledge sourcing is, the more likely these knowledge sources would be wide spread
across technological fields; the more diverse their technological fields are, the more
likely they are sourcing from a wide range of geographical locations.
The relationship between technological and geographical dispersion of knowledge
sources, however, could be affected by the level of concentration on home country
sourcing or host country sourcing. High concentration on home country sourcing could
result in a relatively low geographical dispersion of knowledge sources, but as a subunit
44
seeking for knowledge sources from its parent company, the technological dispersion of
these sources could still be high if the parent company has a diversified range of
technological activities.
Hypothesis 1: There is a positive relationship between technological dispersion and
geographical dispersion of knowledge sources, but the relationship is moderated by level
of concentration in citation in home or host countries.
Technological dispersion of knowledge sources is determined by the nature of technology
that the focal subunit is working on; usually the more complex the technology is, the
more diverse of knowledge it needs to source from. The development of science and
technology determines that recombination and creation of new knowledge requires an
often cross-sector combination of existing knowledge sources. Although historical study
of technological changes has shown a relatively stable country-specific path dependency
of technology trajectory (Pavitt, 1982; Vertova, 1999), there are empirical evidences
demonstrating an increasingly diversified technological development along with a more
complex and complementary knowledge accumulation (Cantwell and Piscitello, 2000;
Cantwell and Vertova, 2004).
In comparison of the diversification of a subunit’s technological profile and its
knowledge source technological dispersion, I propose that there is likely to be a positive
relationship between those two. Although the causality can be argued in both directions,
this study focuses at the direction from source to outcome. Therefore I suggest:
45
Hypothesis 2: The technological diversification of an MNC subunit’s knowledge
innovation is positively related to the technological dispersion of its knowledge sourcing.
When we are examining the overall effect of technological dispersion of knowledge
sourcing on innovation diversification, one of the areas that we should pay attention to is
the composition of general purpose technology (GPT) in areas of specialization of a
subunit. GPT fields are characterized by the potential for pervasive use in a wide range of
sectors (Rosenberg, 1982; Bresnahan and Trajtenberg, 1992). Some technologies such as
non-electrical machinery (Rosenberg, 1976), instrument and controls, chemical processes,
and computing (Granstrand et, al., 1997) have been found actively mobilized in a wide
range of firms in different industries, and thus to belong to GPTs.
More recently, scholars further characterized the term “general purpose” with not only a
wide range of users, but also the technological cumulativeness, dynamism and
complementarity innovations (Bresnahan and Trajtenberg, 1995; Helpman and
Trajtenberg, 1998). In more recent years, a new ICT-based (Information and
Communication Technologies) technology paradigm has emerged (Granstrand, et al,
1992; Oskarsson, 1993; Patel and Pavitt, 1991). ICTs are then considered an advanced
type of GPTs (Cantwell and Santangelo, 2000). Firms as main actors in these fields tend
to reinforce the development of GPTs to support an even more widely dispersed network
of differentiated creativity; and research has shown a positive relationship between firm’s
involvement in GPT and the technological diversification and geographical
diversification (internationalization) of the firm’s innovation network (Qiu, 2011).
Therefore I hypothesize:
46
Hypothesis 2a: The relationship between technological diversification of an MNC
subunit’s knowledge innovation and technological dispersion of its knowledge sourcing is
positively moderated by the share of general purpose technology fields in the technology
developed by the subunit.
The second dimension to describe knowledge sourcing characteristic is its geographic
dispersion. Geographic dispersion is defined by the level of knowledge sources being
scattered about over a range of geographic locations. When examining the geographic
dispersion of knowledge sources, we are looking at subunits’ sourcing from both their
“internal” and “external” knowledge “networks”. Research shed light on geographic
factors and their influence on knowledge transfer in the past decade (Rosenkopf &
Almeida, 2003; Hansen & Lovas, 2004; Bell & Zaheer, 2007). There is a strong positive
link found between geographic proximity and the sharing of knowledge (Saxenian, 1994;
Allen, 1997); it is difficult to access and transfer knowledge from distant sources
comparing to local knowledge sources (Jaffe et al., 1993; Almeida and Kogut, 1999). In
this perspective, when we are examining the intensity of connections within a certain
knowledge network, the geographic distribution of participants in the network must be
taken into consideration. However, due to the different natures of an MNC subunit’s
“internal” and “external” networks, the geographic dispersion of connections in those two
networks can have different influences on the corresponding network connection
intensities. In the meanwhile, both the connection intensity and geographic dispersion can
be influenced by the industrial characteristics – namely the “pool” of accessible
knowledge in a certain field at a location.
47
Generally speaking, the geographic dispersion of knowledge connectedness
within an MNC largely is supposed to reflect the intensity of connection between
headquarter and subunit, or among different subunits. For a typical MNC subunit, a
geographically concentrated distribution of internal knowledge connections most likely
indicates the strong tie between itself and its headquarter, while a geographically
dispersed internal connection pattern demonstrates the possible even connection between
itself and different subunits in other locations, may or may not include the headquarter.
That being said, as a deduction from hypotheses one and two, one would expect a
geographically dispersed knowledge sourcing could more than likely result in a higher
level of technological diversification of subunit’s innovative activities. Therefore, we
have the following hypothesis:
Hypothesis 3: The technological diversification of an MNC subunit’s knowledge
innovation is positively influenced by the geographic dispersion of its knowledge
sourcing.
Although the sourcing of knowledge is known to be increasingly dispersed worldwide,
geographic proximity still plays an important role in determining the extent of knowledge
spillover. Research has shown that the externality of knowledge has the strongest effect
when both the sender and receiver (either intentionally or unintentionally) are
geographically proximate to one another (Jaffe et al., 1993; Almeida and Kogut, 1999).
Despite the advancement of transportation, communication and information technology,
it is still hard to engage in efficient technological knowledge transfers at a distance.
When geographic locations are treated equal to one another, it is difficult to capture the
48
effect of geographical distance on knowledge sourcing. Therefore we propose a
moderated model which distinguishes the geographic locations of knowledge sourcing by
treating intra-region and inter-region knowledge sources differently. Knowledge sourcing
from a country within the same geographic region indicates high geographical proximity,
while sourcing from a country outside the geographic region indicates low geographical
proximity.
Hypothesis 3a: The relationship between technological diversification of an MNC
subunit’s knowledge innovation and geographic dispersion of its knowledge sourcing
may be moderated by the general geographic proximity of knowledge sources to the
subunit.
The extent of external embeddedness of subunits to its host country (local environment)
could contribute to the composition of fields of technological specialization. As discussed
before, level of local embeddedness is shown to be related to MNC strategies such as
parental control, subunit mandate, subunit autonomy, etc. A highly embedded subunit is
expected to reflect in its technological profile the characteristics of the local technological
expertise. Although it may be intuitive to deduct that the level of technological
diversification of subunit is positively related to the level of sourcing dispersion in the
local environment, it is possible that this effect may be strong enough to suppress the
influence of knowledge sources from outside the host country – in a case where the local
embeddedness is at a high level. By controlling the effect of knowledge sourcing from
host country, we can then observe the international knowledge sourcing effect of a
subunit and its influence on the subunit’s technological diversification.
49
To note here, by excluding the effect of host country knowledge sourcing, we can also
find this as a proxy of the knowledge sourcing from internal sources of a subunit’s own
international multinational group, with the noise of sourcing from external sources that’s
not in the local context.
First, we hypothesize that there is a positive relationship between a subunit’s
technological diversification and its knowledge sourcing technological dispersion outside
its host country. The more dispersed into different fields its knowledge sources are, the
more likely it is to develop diversified technological innovation.
Hypothesis 4a: The technological dispersion of international knowledge sourcing from
outside the host country has a positive impact on the MNC subunit’s technological
diversification.
Then if we take a look at the geographic composition of knowledge sources outside the
host country, following the analysis of geographic dispersion and technological
diversification, it is expected that there is also a positive relationship between these two
variables. Excluding the host country effect, the more geographically dispersed a
subunit’s knowledge sources are, the more likely it is to develop diversified technological
innovation.
Hypothesis 4b: The geographical dispersion of international knowledge sourcing from
outside the host country has a positive impact on the MNC subunit’s technological
diversification.
50
Aside the effect of host country as a geographical location where subunits tend to heavily
source their knowledge from, home country location is another knowledge source that
should be taken into consideration. A subunit sourcing heavily from its home country (in
this case most likely from its parent company, which is the operation of its MNC group in
the home location) could in effect result in an overall technological and geographical
dispersion skewed towards its home country sourcing. It might actually be sourcing from
dispersed technological fields, but if the home country sourcing is mainly drawn from
certain fields, the effect of technological dispersion would be diminished; or, it might be
intentionally concentrating on seeking for sources from certain technological fields, but
due to dispersed sourcing effect from home country (many of the times a result of
sourcing from parent group that establishes the basis of the subunit’s technological
innovation which is intrinsically dispersed) the concentration effect could be diminished.
Therefore by excluding home country sourcing, we may be able to see a different result
showing the knowledge sourcing pattern of subunits. Hence I hypothesize:
Hypothesis 5a: The technological dispersion of external knowledge sourcing from outside
the home country has a positive impact on the MNC subunit’s technological
diversification.
Similarly, if a subunit has a large share of citation from its home country, it’s difficult to
determine whether its knowledge sources through the entire global knowledge base is
geographically concentrated or dispersed. As a lot of subunit tend to rely on its parent
company or home country basis of knowledge stock, it is understandable to observe a
relatively high share of home country originated knowledge sources. By excluding the
51
home country sourcing, we will then be able to capture the subunit’s effort of knowledge
sourcing from the rest of the world. The following hypothesis examines this effect:
Hypothesis 5b: The geographical dispersion of external knowledge sourcing from outside
the home country has a positive impact on the MNC subunit’s technological
diversification.
3.4 Framework
Figure 6 demonstrates a framework for this part of my study.
Figure 6 Framework of Study 1 - Knowledge sourcing patterns
3.5 Methodology
The study uses step-wise regression models to analyze the relationship between
technological and geographical dispersions of knowledge sources, and their impact on
subunit technological diversification. Data is constructed at MNC subunit level, including
US patents of 157 large chemical industry firms granted between 1976 and 2006. In order
to minimize the effect of small number problem, a five-year window is constructed to
aggregate patent numbers into a higher level, creating data panels through the 31-year
period.
52
Key Variables
Technological Diversification – The extent to which a subunit’s innovative activities
(patents) are wide spread across various technological fields (see Chapter 2 for detailed
definition).
Technological Dispersion – The extent to which knowledge sources is composited from
various technological fields (see Chapter 2 for detailed definition).
Geographical Dispersion – The extent to which knowledge sources is composited from
various geographical locations (see Chapter 2 for detailed definition).
Non-host Tech Disp – Technological dispersion of knowledge sources excluding those
from host country.
Non-host Geo Disp – Geographical dispersion of knowledge sources excluding those
from host country.
Non-home Tech Disp – Technological dispersion of knowledge sources excluding those
from home country.
Non-home Geo Disp – Geographical dispersion of knowledge sources excluding those
from home country.
GPT Share – Share of technological innovation that falls into the category of GPT fields.
Inter Region Share – Share of knowledge sources that are from a different geographical
region than the host country of subunit.
53
Moderators and Control Variables
Home Share – Share of knowledge sources from home country of the MNC group.
Host Share – Share of knowledge sources from host country of the subunit.
Host Country Tech Div – Degree of technological diversification of the host country
where the subunit is located.
US-Home – Dummy variable indicating whether the MNC firm is US originated.
US-Host – Dummy variable indicating whether the subunit is located in US.
The table of correlation of key variables is shown in Appendix B.
3.6 Results and Discussion
Hypothesis 1 examines the relationship between technological dispersion and
geographical dispersion of subunit knowledge sources, with the effect being moderated
by shares of home-country or host-country sources. Three separate sets of models are
analyzed for hypothesis 1 to capture the effect of (1) all knowledge sources moderated by
home country sourcing share; (2) knowledge sources outside the host country moderated
by host country sourcing share; (3) knowledge sources outside the home country
moderated by home country sourcing share. The regression results of these three sets of
models are shown in Tables 6, 7 and 8:
Table 6 Hypothesis 1 Model with all data
Tech Dispersion 1
2
3 Geo Dispersion 0.51 *** 0.481 *** 0.408 ***
Home Share
0.064 *** 0.009
54
Geo Dispersion X Home Share
0.115 *** _cons 0.251 *** 0.248 *** 0.289 ***
R-sq 0.254
0.17
0.173 chi2 5741.7 *** 2771.2 *** 2792.7 ***
Table 7 Hypothesis 1 Model without host country sources
Non-host Tech Disp 4
5
6 Non-host Geo Disp 0.45 *** 0.464 *** 0.07 ***
Host Share
-0.139 *** -0.487 *** Non-host Geo Disp X Host Share
0.762 ***
_cons 0.215 *** 0.006 *** 0.46 ***
R-sq 0.19
0.216
0.26 chi2 4462.3 *** 3738.61 *** 4581.2 ***
Table 8 Hypothesis 1 Model without home country sources
Non-home Tech Disp 7
8
9 Non-home Geo Disp 0.468 *** 0.471 *** 0.088 ***
Home Share
-0.176 *** -0.491 *** Non-home Geo Disp X Home Share
0.701 ***
_cons 0.19 *** 0.275 *** 0.45 ***
R-sq 0.2242
0.276
0.306 chi2 4752.2 *** 4309.01 *** 5011.8 ***
From these results we can see that there is indeed a positive relationship between
technological dispersion and geographical dispersion of knowledge sources across all
samples of data. This positive relationship, however, is reduced when a partial sample is
taken – the coefficients shown in model 4 and model 7 as two main effects are lower than
the coefficient in model 1. In the all-data sample model (Table 6 Model 3), a positive
55
effect of share of home country sources is demonstrated. It shows that when there is a
large share of home country sourcing, there is a stronger positive relationship between
technological dispersion and geographical dispersion – which is consistent with the result
in the sample excluding home country sourcing (Table 8 Model 9), the coefficient
becomes smaller in this model, and when the share of home country sourcing is taken
into consideration, it negatively affects the relationship between technological dispersion
and geographical dispersion. The result in Table 7 (Model 6) tells a similar story, except
this time the location excluded from the sample is on host country.
Hypothesis 2 to 5 examines the relationship between subunit’s technological
diversification and its knowledge sourcing patterns. The results (shown in Table 9)
support hypotheses 2, 3, 4a, 4b, 5a, and 5b. However, some interesting results are found
in the test of hypotheses 2a and 3a. There is a negative moderating effect of GPT share on
the relationship between subunit technological diversification and technological
dispersion of knowledge sources. Although GPT share itself is positively influencing
technological diversification (which is consistent with the argument of general purpose
technology), the interaction term of technological dispersion of sourcing and subunit GPT
share shows a negative coefficient. It indicates that subunits with a high concentration on
innovation in GPT fields are themselves intrinsically diversified due to the effect of
investment on GPT, then the effect of overall knowledge sourcing technological
dispersion is not that significant given the case.
56
Table 9 Results for H2-H5, Study 1
Tech Div H2 H2a H2a H3 H3a H3a H4a H4b H5a H5bUS_home -0.74 *** -181 *** -0.08 *** -0.073 *** -0.073 *** -0.079 *** -0.08 *** -0.08 *** -0.073 *** -0.075 *** -0.072 *** -0.076 ***US_host 0.115 *** 0.178 *** 0.101 *** 0.099 *** 0.99 *** 0.11 *** 0.111 *** 0.11 *** 0.11 *** 0.11 *** 0.11 *** 0.11 ***Host Share 0.005 -0.024 0.022 0.026 ** 0.024 ** 0.033 *** 0.004 0.009 0.019 ** 0.01 0.015 0.02Home Share -0.004 0.87 *** 0.009 0.009 0.009 0.028 ** 0.026 * 0.026 * 0.001 0.008 0.007 -0.001_cons 0.685 *** 0.514 0.63 *** 0.56 *** 0.539 *** 0.622 *** 0.656 *** 0.647 *** 0.656 *** 0.669 *** 0.655 *** 0.669 ***Host Tech Div 0.81Tech Disp 0.088 *** 0.088 *** 0.13 ***GPT Share 0.143 *** 0.187 ***Tech Disp X GPT Share -0.087 ***Geo Disp 0.075 *** 0.063 *** 0.079 ***Inter Region Share -0.03 *** 0.001Geo Disp X Inter Region Share -0.06 *Non-host Tech Disp 0.05 ***Non-host Geo Disp 0.025 ***Non-home Tech Disp 0.053 ***Non-home Geo Disp 0.026 ***
R-sq 0.07 0.311 0.11 0.153 0.153 0.078 0.077 0.077 0.091 0.074 0.093 0.075chi2 321.56 *** 490.1 *** 588.4 *** 989.7 *** 1004 *** 394.2 *** 406.9 *** 413.6 *** 447.96 *** 346.5 *** 464.43 *** 349.6 ***
57
For geographical dispersion, the direct effect on technological diversification is a positive
one, just as proposed in hypothesis 3. However, the effect of share of inter-regional
knowledge sourcing is shown to have a negative impact on technological diversification.
The rationale behind this is that when subunits are sourcing their knowledge from a wide
range of geographical locations, they tend to be more concentrated on the targeting
technological fields. In other words, when subunits have a higher level of concentration
on their specialized technological fields, rather than being diversified into various fields,
they tend to source their knowledge more focused and more purposely from the entire
knowledge base around the globe, instead of being geographically bounded. Whereas
when subunits focuses more on knowledge sources from geographically proximate
locations, they tend to draw on all the potential areas of expertise from these locations,
hence more likely to develop a technologically diversified profile of innovation activities.
58
4. Competence Exploiting and Competence Creating
4.1 Competence Exploiting vs. Competence Creating
The theory of organizational capability represents an extension and synthesis of the
contributions from the knowledge based view of organizations, based upon the idea that
the essence of organizational capability is the integration of individuals’ specialized
knowledge (Grant, 1996).
The notion of “exploitation” and “exploration” was first brought into attention in
organizational learning theories (March, 1991). The distinction between the two draws on
March’s (1991, p. 85) view of exploitation as “the refinement and extension of existing
competencies, technologies, and paradigms” and exploration as “experimentation with
new alternatives that have returns that are uncertain, distant, and often negative.” Based
on the distinction between the two concepts, studies of a wide range of management
fields have shown that exploitation and exploration require substantially different
structures, processes, strategies, capabilities, and cultures to pursue and may have
different impacts on firm adaptation and performance (He & Wong, 2004). In general,
exploration is associated with organic structures, loosely coupled systems, path breaking,
improvisation, autonomy and chaos, and emerging markets and technologies.
Exploitation is associated with mechanistic structures, tightly coupled systems, path
dependence, routinization, control and bureaucracy, and stable markets and technologies
(Ancona et al., 2001; Brown and Eisenhardt, 1998; Lewin et al., 1999).
59
A firm’s competence refers to the knowledge, skills, and related routines that
constitute its ability to create and deliver superior customer value (Day 1994).
Competence exploitation refers to the tendency of a firm to invest resources to refine and
extend its existing product innovation knowledge, skills, and processes, which aims to
greater efficiency and reliability of existing innovation activities; while in contrast,
competence exploration refers to the tendency of a firm to invest resources to acquire
entirely new knowledge, skills, and processes, with the objective of attaining flexibility
and novelty in product innovation through increased variation and experimentation
(March, 1991; Atuahene-Gima, 2005).
When an MNC establishes a foreign subsidiary of a resource-seeking or market-
seeking kind, the initial effort in the traditional view of foreign direct investment (FDI) is
to exploit the competence that’s already established by the parent group in the home
country base. Technological knowledge required in the foreign production of this kind is
mostly created at home, and then transferred via the internal international corporate
network to foreign production locations (Cantwell & Piscitello, 2000; Cantwell &
Mudambi, 2005). As the increasing complexity of technology and internationalization of
production development brought attention to the competence-seeking or knowledge-
seeking activities of foreign investment, the competence-based theory of firm treats firm
as an institution where capabilities are constructed through continuous internal learning;
the major issue of a firm is then no longer the exploitation of established competence, but
rather how to create new competence in a geographically dispersed and technologically
60
diversified system leveraging not only the MNC internal resources but also the
technology base of the host location (Cantwell, 1991; Cantwell & Piscitello, 2000).
4.2 The Relationship between Knowledge Sourcing Patterns and Competence Creating
The types of innovative activities – whether they are competence creating (CC) or
competence exploiting (CE) – that an MNC subunit conducts can be determined mainly
by several factors – the parent group pressure, the locational pressure, or the subunit
autonomy. MNC parent group makes the initial definition of subunit strategic position –
the purpose of establishing a certain subunit in the specific market can be local market
oriented, local resource (natural or labor) oriented or local competence oriented. The first
two types of subunits are designed mainly to exploit the established parent group
competence base in a new location, while the third type seeks to create new competence
that contributes back to the MNC group. Locational characteristics such as industry
expertise, high quality human resource availability, and policy to FDI play an important
part in such choice making of an MNC group. Because the creation of competence
demonstrates the capability of a subunit to innovate into new profitable technological
fields, which in turn earns the subunit better position within the MNC network, and hence
more favorable resource allocation by the parent group, subunit management are
encouraged to seek to achieve a mandate of CC by initiating local knowledge seeking or
new competence creating (Cantwell & Mudambi, 2005).
In order for subunits to achieve a CC mandate, they need to first have a strong
capability of innovation per se; then these innovative activities need to be focused on
61
exploring new inventions that are novel to the parent MNC group. These activities
require knowledge connection and recombination from different sources – from the
parent group to build a strong innovation base, and from the external sources to explore
new areas of innovative capability. Since MNC subunits themselves can be consist of a
wide variety of operations, it is difficult to define a subunit clearly as a CC or CE type.
For instance, a subunit with a CE mandate can establish competence in certain areas that
is new to the parent group – even that isn’t the core competence or that subunit; on the
contrary, a subunit with a CC role can also manage some operations that are generally
based on parent group competence, with little creative inventions in areas new to the
home base. It is more appropriate then to examine the individual innovative activities of a
subunit, instead of treating all things as a whole in one large overseas operation. On the
subunit level, we can then look at the CC activity intensity (the share of competence
creating activities among all innovative activities within a subunit) and CC activity distance
(the technological distance between competence creating activities of a subunit and the closest
areas of expertise of its parent group).
The intensity of CC activity has much to do with the knowledge sourcing and
recombination from all possible sources. High embeddedness in either the MNC
international network or the external local network (or even international) both facilitates
the subunit’s capability of creating new competence. Especially when the embeddedness
in both networks are high, the subunit can have more possible knowledge sources to learn
and recombine, which means higher potential to create new competence. This reflected
62
on the abstract level of network means high knowledge connectedness from either the
internal sources or the external sources can both promote CC activity intensity.
On the level of individual technological fields, the extent to which a subunit’s
specialization in that field is different from the specialization degree of its parent
company can be captured by technological distinctiveness, as introduced in Chapter 2.
This concept takes a further step from simply identifying a subunit’s innovative activities
in one field to be CC or CE, by quantifying how much is the subunit’s degree of
specialization different from that of its parent company. Technological distinctiveness is
an index that demonstrates the composition of CC or CE activities comparing the subunit
and its parent company.
The factors that can influence the degree of technological distinctiveness, and
therefore influence the competence creativeness of a subunit, can be various. Externally,
the mandate from parent company and the technological influence from the host country
environment could have an impact on the subunit’s technological distinctiveness in
certain fields; internally, the management autonomy, the strategy chosen, and the
intellectual capability of human resource could impact a subunit’s technological
distinctiveness in certain fields. Needless to say the parent company’s technological
expertise plays a very important role in defining the subunit’s “distinct” areas of expertise.
Since this concept is a comparison between the levels of specialization of the two entities,
simply demonstrating a strong specialization in a certain field doesn’t necessarily indicate
that the subunit has a high technological distinctiveness in this field. If the level of
63
specialization of subunit and parent are both high, the technological distinctiveness of the
subunit in this field might actually be relatively low.
From the MNC’s point of view, the structure of innovation network for the MNC
group and the positioning of a subunit in this network could be two of the important
factors demonstrating the mandate that’s assigned to the subunit. Structural concentration
of the innovation network usually results in a concentration of innovative activities in the
home bases. This form of MNC structure is consistent with the traditional views of
multinational firms, where research and innovations are mostly initiated from the parent
company, and then transferred to other parts of the international group. Within this type
of structure, foreign subunits are usually conducting incremental innovative activities
around the areas of expertise of the parent company; exploration into new areas of
technology is not supported by the innovation network structure, therefore I propose:
Hypothesis 1a: There is a negative relationship between an MNC’s structural
concentration level of innovative activity and the subunit’s degree of technological
distinctiveness from its parent company.
However, this negative effect could be changed of a subunit has a relatively high share of
innovative activity among all international locations of its MNC group. That is to say,
even if there is a high level of geographical concentration on innovative activities for an
MNC, if the concentration is not limited in the MNC home location (parent company),
but other foreign locations also play an important role in the innovation network, then
there could be a positive relationship between the degree of geographical concentration
64
and the technological distinctiveness of subunit in certain fields. In the meanwhile, if a
subunit is one of the most important innovative locations for its MNC group, it is more
likely that this subunit develops a strong basis of innovation capability, then it is more
likely to diversify its technological expertise into some fields that are not necessarily the
fields of specialization for the parent company. Therefore I propose the following
hypothesis:
Hypothesis 1b: There is a positive relationship between a subunit’s share of innovative
activity in its international corporate group and its degree of technological
distinctiveness from its parent company.
The relationship between a subunit’s knowledge sourcing pattern and technological
distinctiveness could be a mutually interdependent one. It is hard to argue whether it is
because a subunit’s mandate from its parent company that strategized it to perform a
competence creating role, then the subunit focuses its knowledge sourcing into dispersed
technological fields and geographical locations to facilitate its diversification into new
and novel technological fields; or it could be that the subunit has access to dispersed
sources of knowledge due to location specific advantages and human resource factors,
then these readily available knowledge sources could reinforce its specialization in
certain technological fields that are not inherited from specialized field of its parent
company. Although this causality could be argued either way, it is believed that there is a
relationship between a subunit’s knowledge sourcing pattern and the composition of its
technological specialization.
65
As we discussed in Chapter three, the knowledge sourcing pattern of a subunit can be
described by two dimensions – the technological dispersion and geographical dispersion
of its knowledge sources. The former looks at different areas of technologies that the
knowledge sources are composed from – whether they are focused on certain particular
fields or dispersed from different areas; the latter examines the geographical locations of
these knowledge sources – whether the sources are geographically concentrated or
scattered across the world. If a subunit is sourcing from dispersed areas of technological
fields, it is more likely to develop innovative activities in a diversified range of
technology fields (as the result demonstrated in Chapter Three). Among these fields, it is
then more likely to create a new area of expertise that’s different from that of its parent
company. Therefore I hypothesize:
Hypothesis 2a: There is a positive relationship between a subunit’s technological
dispersion of its knowledge sources and its degree of technological distinctiveness from
its parent company.
Similarly, if a subunit has a wide range of knowledge sources across various geographical
locations, it is more likely to develop a higher degree of technological diversification.
Therefore, higher level of geographical dispersion of knowledge sourcing in a certain
field could result in a higher possibility of technological distinctiveness in that field.
Again, I hypothesize a positive relationship between these two:
66
Hypothesis 2b: There is a positive relationship between a subunit’s geographical
dispersion of its knowledge sources and its degree of technological distinctiveness from
its parent company.
The next set of hypotheses is drawn upon the notion of a subunit’s knowledge sourcing
pattern being influenced on its integration with its home country or its embeddedness
with its host country.
It is arguable that high embeddedness in host country could many times result in a
subunit diversifying its technology into areas of expertise of the host country location,
which are not always consistent with its parent company’s areas of expertise. Therefore,
if a subunit’s knowledge sourcing is heavily concentrated on its host country location, it
could dilute the relationship between knowledge sourcing pattern and technological
distinctiveness of the subunit. Therefore I exclude the knowledge sources from subunits’
host countries, and propose the following hypotheses with the expectation of stronger
positive relationships than those in hypotheses 2a and 2b:
Hypothesis 3a: Excluding host country sourcing, there is a positive relationship between
a subunit’s technological dispersion of its knowledge sources and its degree of
technological distinctiveness from its parent company.
Hypothesis 3b: Excluding host country sourcing, there is a positive relationship between
a subunit’s geographical dispersion of its knowledge sources and degree of technological
distinctiveness from its parent company.
67
If a subunit is concentrating its knowledge sourcing heavily from its home country,
especially in the cases where the level of MNC integration is high, it is more likely that
the subunit relies on its home country / parent company technological expertise; hence it
is less likely to develop innovative activities into areas that are distinct from its parent
company. More specifically, the share of subunit concentration on home country sources
should have a negative impact on its technological distinctiveness (tested in model as
control variable). When the sources from home country is excluded from the sample, or
when we focus on knowledge sources outside a subunit’s home country, we should
therefore expect a stronger positive relationship between its technological dispersion of
knowledge sources and degree of technological distinctiveness from its parent company,
and a stronger positive relationship between its geographical dispersion of knowledge
sources and the degree of technological distinctiveness from its parent company. The
corresponding hypotheses are:
Hypothesis 4a: Excluding home country sourcing, there is a positive relationship between
a subunit’s technological dispersion of its knowledge sources and degree of
technological distinctiveness from its parent company.
Hypothesis 4b: Excluding home country sourcing, there is a positive relationship between
a subunit’s geographical dispersion of its knowledge sources and degree of technological
distinctiveness from its parent company.
4.3 Framework
The following framework in Figure 7 summarizes the conceptual model of Study 2.
68
4.4 Methodology
The study uses step-wise GLS regression models. Data is constructed at MNC subunit
level, including US patents of 147 large chemical industry firms granted between 1976
and 2006. In order to minimize the effect of small number problem, a five-year window
is constructed to aggregate patent numbers into a higher level, creating data panels
through the 31-year period. The measurements of key variables are listed as follows:
Dependent Variable:
Technological Distinctiveness
Technological Dispersion
MNC Innovation Concentration
Geographical Dispersion
Subunit Share of Innovation
Firm Structure
Subunit Knowledge Sourcing Pattern
Home Country Sourcing
Host Country Sourcing
H1
H2
H3 H4
Figure 7 Conceptual Framework of Study 2
69
Technological Distinctiveness – The extent to which a subunit’s technological
specialization in a given field is different from that of its parent company (see Chapter
Two for detailed definition and measurement description).
Independent Variables:
Firm Innovation Concentration (FirmCon) – MNC group’s level of concentration of its
innovative activities. This variable is measured with similar mechanism from the HHI
index of concentration, by aggregating the square of share of innovative activities of all
subunits in an MNC group. If 𝑃𝑃𝑖𝑖 indicates the number of patents in subunit i, then the
formula for FirmCon is:
𝐹𝐹𝑖𝑖𝑃𝑃𝐹𝐹𝐹𝐹𝐹𝐹𝑛𝑛 = � �𝑃𝑃𝑖𝑖∑ 𝑃𝑃𝑖𝑖𝑖𝑖
�2
𝑖𝑖
Subunit Share (SubShare) – share of a subunit’s patent within its MNC group.
Technological Dispersion – The extent to which knowledge sources is composited from
various technological fields (see Chapter Two for detailed definition and measurement
description).
Geographical Dispersion – The extent to which knowledge sources is composited from
various geographical locations (see Chapter Two for detailed definition and measurement
description).
Non-host Tech Disp – Technological dispersion of knowledge sources excluding those
from host country.
70
Non-host Geo Disp – Geographical dispersion of knowledge sources excluding those
from host country.
Non-home Tech Disp – Technological dispersion of knowledge sources excluding those
from home country.
Non-home Geo Disp – Geographical dispersion of knowledge sources excluding those
from home country.
Control Variables:
GPT Share – Share of technological innovation that falls into the category of GPT fields.
Inter Region Share – Share of knowledge sources that are from a different geographical
region than the host country of subunit.
Home Share – Share of knowledge sources from home country of the MNC group.
Host Share – Share of knowledge sources from host country of the subunit.
US-Home – Dummy variable indicating whether the MNC firm is US originated.
US-Host – Dummy variable indicating whether the subunit is located in US.
RTA Host – The revealed technological advantage of a subunit’s host country in a given
field. This variable controls for the host country’s technological specialization effect on
subunit technological distinctiveness.
71
Host Type – Dummy variable indicating type of host country: developed country or
developing country.
Firm size – Total number of patents of a given MNC. The variable of firm size uses the
log result of patent number.
Industry – Dummy variables designed to subdivide the general chemical industry into
subcategories based on a MNC group’s main activity: chemical, pharmaceutical and
biology (phar), petroleum and energy (petr).
4.5 Results and Discussion
The results of Study Two are shown in the following Table 10. Most of the models are
shown to be fit, while a few models are indicating a relationship between the independent
variable and dependent variable that’s in the opposite direction as proposed in my
hypotheses. In order to minimize the effect of small number problem, as well as to test
the robustness of the methodology, I then selected the sample with several cut-off points
of subunit patent number. Table 11 reports the results of analysis with a cut-off point of
50, which means all subunits in this sample has more than 50 patents in the
corresponding five-year window.
72
Table 10 Results for Study 2 - All Data
1 2 3 4 5 6 7 8 9
US_home 0.031 0.031 -0.044 * -0.057 ** -0.031 -0.049 ** -0.043 * -0.049 ** -0.046 *
US_host 0.011
0.01
-0.033
-0.06 ** -0.023
-0.044
-0.031
-0.047 * -0.035 hs_share -0.069 * -0.08 * -0.043
0.012
-0.114 ** -0.024
-0.048
-0.031
-0.035
hm_share -0.14 *** -0.125 *** -0.143 *** -0.116 *** -0.207 *** -0.136 *** -0.15 *** -0.121 *** -0.139 ***
RTA_host 0.013 *** 0.013 *** 0.013 *** 0.014 *** 0.013 *** 0.013 *** 0.013 *** 0.013 *** 0.013 ***
host_type -0.065
-0.061
-0.068
-0.073
-0.069
-0.074
-0.068
-0.08
-0.068 log size -0.109 *** -0.108 *** -0.099 *** -0.105 *** -0.1 *** -0.107 *** -0.099 *** -0.106 *** -0.1 ***
phar -0.036 ** -0.029
-0.014
-0.005
-0.015
-0.008
-0.014
-0.008
-0.014 petr 0.038
0.037
0.008
0.012
0.002
0.008
0.008
0.008
0.008
gpt_share
0.083 *** 0.052 ** 0.054 ** 0.052 ** 0.054 ** 0.052 ** 0.05 ** 0.052 **
intreg_share
-0.02
0.01
0.034
-0.008
0.003
0.01
0.001
0.01 _cons 0.878 *** 0.844 *** 0.665 *** 0.574 *** 0.78 *** 0.634 *** 0.671 *** 0.637 *** 0.658 ***
struct
0.299 *** 0.319 *** 0.3 *** 0.327 *** 0.298 *** 0.327 *** 0.302 ***
sub_share
0.003 *** 0.002 *** 0.003 *** 0.002 *** 0.003 *** 0.003 *** 0.003 ***
ced_td
0.138 *** ced_gd
-0.115 **
nhsced_td
0.086 *** nhsced_gd
-0.008
nhmced_td
0.093 *** nhmced_gd
0.01
R sq 0.047
0.052
0.092
0.1
0.093
0.094
0.092
0.095
0.091
Chi2 131.61 *** 146.59 *** 239.91 *** 271.71 *** 245.19 *** 255.51 *** 240.14 *** 258.05 *** 240.01 *** * p < .10
** p < .05 *** p < .01
73
Table 11 Results for Study 2 - Large Subunit Sample
1 2 3 4 5 6 7 8 9
US_home -0.083 * -0.055 -0.138 *** -0.15 *** -0.087 -0.139 *** -0.114 ** -0.139 *** -0.126 **
US_host -0.073
-0.064
-0.087 * -0.102 ** -0.044
-0.087 * -0.053
-0.091 * -0.075 hs_share 0.085
0.138 * 0.161 ** 0.2 *** -0.079
0.161 ** 0.109
0.154 ** 0.104 *
hm_share -0.067
-0.117
-0.116
-0.088 ** -0.299 *** -0.109
-0.224 ** -0.09
-0.155 *
RTA_host 0.024 ** 0.025 *** 0.025 *** 0.027 *** 0.027 *** 0.026 *** 0.026 *** 0.026 *** 0.026 ***
log size -0.129 *** -0.119 *** -0.098 *** -0.105 *** -0.098 *** -0.106 *** -0.092 *** -0.109 *** -0.094 ***
phar 0.074 ** 0.088 *** 0.094 *** 0.097 *** 0.092 *** 0.097 *** 0.093 *** 0.1 *** 0.092 ***
petr 0.096 ** 0.099 ** 0.045
0.052
0.024
0.042
0.058
0.042
0.052 gpt_share
0.171 *** 0.06
0.046
0.076
0.057
0.051
0.052
0.052 *
intreg_share
0.116 * 0.137 ** 0.16 ** 0.094
0.127 * 0.137 ** 0.12 * 0.152 **
_cons 0.788 *** 0.598 *** 0.458 *** 0.371 *** 0.77 *** 0.437 *** 0.563 *** 0.407 *** 0.512 *** struct
0.248 *** 0.27 *** 0.237 *** 0.271 *** 0.217 *** 0.287 *** 0.231 ***
sub_share
0.003 *** 0.003 *** 0.003 *** 0.003 *** 0.003 *** 0.003 *** 0.003 ***
ced_td
0.134 *** ced_gd
-0.32 ***
nhsced_td
0.067 nhsced_gd
-0.145 **
nhmced_td
0.407 *** nhmced_gd
-0.078
R sq 0.052
0.074
0.134
0.139
0.143
0.134
0.149
0.138
0.139 Chi2 63.48 *** 77.48 *** 137.19 *** 144.21 *** 151.1 *** 139.76 *** 146.84 *** 146.23 *** 140.37 ***
* p < .10 ** p < .05 *** p < .01
74
Most of the main effects are shown to be consistent between Table 10 and Table 11.
Hypothesis 1a is not supported by the data analysis in both samples (Model 3 in both
samples). In fact, there is a consistent strong positive relationship between corporate
innovation structural concentration and subunit’s technological distinctiveness, instead of
the negative one proposed in previous analysis. There are two possible explanations to
this result: (1) the firm’s structural concentration could be partially focused on the focal
subunit. In that case the subunit would demonstrate a high share of innovation in the
MNC group. (2) In some cases, if a firm has highly concentrated innovation structure, it
is therefore focusing its core areas of technology development in these concentrated
locations (usually the home country base), whereas the technological development in
other foreign locations could be diversified into various areas that are not the parent
company’s fields of specialization. As a result the technological distinctiveness of those
subunits that are conducting some level of R&D outside of the home country focus their
efforts on developing new areas of technological innovations for the MNC corporation.
The first explanation could be further tested with an examination of interaction effect
between the MNC group’s innovation structural concentration and the subunit’s share of
innovation within the group. The second explanation needs to examine the parent
company’s share of innovation to confirm that the concentration is indeed within the
home country base.
Hypotheses 2a, 3a, and 4a examine the relationship between technological dispersion of
knowledge sources and subunit’s technological distinctiveness. All of these hypotheses
are supported in the sample with all data (Models 4, 6, and 8), but the story of the large
75
subunit sample is a different one. Hypotheses 2a and 4a by Models 4 and 8 are supported,
while Model 6 shows insignificant result, which means considering knowledge sources
outside of host country, there isn’t a significant positive relationship between knowledge
source technological dispersion and the subunit’s technological distinctiveness compared
to its parent company. In other words, for large subunits, their knowledge sources’
diversified contributions to their competence creativeness mainly come from the sources
within their host countries. Another implication behind these results is that large subunits
tend to be more diversified in their technological fields than smaller subunits, therefore
the capability of them in developing novel areas of specialization is not that strongly
dependent on their knowledge sources’ technological dispersion. To put it another way,
when the level of technological diversification is constantly high, from the result of study
1 we can expect a similarly high technological dispersion level, and as these levels get
higher their numbers shown in the data gets closer to 1, resulting in very small variation
among observations. Therefore it’s difficult to capture the relationship between
technological dispersion and technological distinctiveness among large subunits. (2) The
insignificant result in Model 6 may indicate that for large subunits, the knowledge
sources from host country location play a significant role in building up their capacity of
developing novel innovations to the MNC group. When these sources are taken out of the
sample, the effect between the international sources’ technological dispersion and the
subunit’s technological distinctiveness becomes insignificant. This means that the
innovative activities that rely on knowledge sourcing around the globe are actually more
76
concentrated on enhancing the areas of expertise that are inherited from the parent
company.
Hypotheses 2b, 3b, and 4b, examine the relationship between a subunit’s knowledge
sourcing geographical dispersion and its technological distinctiveness. It is shown in both
samples that Model 5 is significant with a negative coefficient. This result is inconsistent
with the hypotheses, where I proposed a positive relationship between the two.
Interestingly, this is consistent with the results shown in the previously discussed models
for technological dispersion. Generally speaking, the more geographically dispersed of a
subunit’s knowledge sources, the less likely it is to develop a high level of technological
distinctiveness in that area. This explanation is in fact consistent with our previous
discussion about local embeddedness and competence creating. In the case of a highly
locally embedded subunit, especially in regions with high specialization of technological
fields, the subunit would show a more geographically concentrated pattern of knowledge
sourcing, and is more likely to develop technological specialization in fields that are not
its parent company’s expertise. With this understanding, it is then easier to understand the
unsupported results in Models 7 and 9 for Hypotheses 3b and 4b.
Note that the control variable of firm size shows a consistent negative effect on
technological distinctiveness of subunit. It is because the larger the firms are, the more
likely it is technologically diversified into a wide range of fields, in extreme cases the
number would be very close to 1. There for in such cases it is very difficult for subunits
to specialize in innovative activities that are not in areas of expertise of the MNC group.
Imagine the most extreme case – if the MNC group is so large that it has high level of
77
RTA in all fields that its subunits could possibly develop; then it would be extremely
hard for its subunits to bring in something new to the group!
Another notion is on the control variable of host country RTA. In both samples, there is a
consistently significant positive effect on subunit technological distinctiveness. The
explanation is that if a host country is specialized in a certain technological field, it’s
more likely for the subunit to learn from the host country and develop an expertise in this
field – this again fits the understanding of the importance of local embeddedness.
There are two more different effects between the all-data-sample and large subunit
sample. First, in large subunit sample, those in pharmaceutical industry tend to be more
likely to develop technological distinctiveness compared to others; however the same
effect isn’t found in all-data-sample. Note that our sample size selection is based on
patenting numbers during the given time period. One could therefore explain that for
large subunits, pharmaceutical companies tend to emphasize more on competence
creating, compared to companies in other sectors. However this effect can be substituted
if the sample is allowing smaller subunits from other industry to drive up their average
distinctiveness based on concentrated innovation efforts in areas away from their parent
companies. Second, the share of patents in GPT fields has a positive impact on
technological distinctiveness in the all-data-sample, but such effect seems to be canceled
out in the large subunit sample. The former supports the general argument of GPT fields,
it increases the possibility of firms to develop novel innovation. Therefore, a positive
relationship is shown. However, for large subunits, like discussed in the end of Chapter
Three, because they are likely to be intrinsically diversified due to their sizes, having a
78
high share of patents in GPT fields doesn’t have as much of an impact on its
distinctiveness.
These results in effect demonstrated that the capability of subunit to have fields of
technological distinctiveness is to some extent locally grounded. One of the important
indicators is a contrast between a subunit’s home and host country’s technological
environment – whether or not the subunit’s host country can provide them specialized
technology in some fields that the home country can’t provide. Hence a subunit’s
technological distinctiveness in some ways reflects the local technological distinctiveness
of host country location vis-à-vis the home country location.
79
5. The Evolution of Subunit Roles
This chapter looks at the evolution paths of subunit roles more closely focusing on the
heterogeneity of firm-specific evolutionary paths in the patterns of knowledge accumulation that
support CC activities, controlling for the industry-specific determinants of such technological
trajectories.
5.1 Typology Based on Knowledge Sourcing Pattern
Subunit evolutions are driven by both internal (initiated by subunits themselves) and
external factors (investment from parent company or other external forces) (Birkinshaw
& Hood, 1998). External forces largely shape the options of subunit, while it is the
subunit managers to take the initiative to respond to the external opportunities. The idea
that multinational subsidiaries are differentiated according to their technological
capabilities and roles can be traced back to Ghoshal and Bartlett’s study (1998) on
subunit tasks. They argued that the tasks of affiliates can be classified in three categories:
creation, adoption and diffusion. ‘Creation’ is to use subunit’s own technical and
managerial resources to respond to local circumstances; ‘adoption’ is to adopt innovation
developed by parent company or a central R&D facility, or other national subsidiaries of
the firm; and ‘diffusion’ is to diffuse their local innovations back to the parent company
or to other subsidiaries. Similarly, Pearce (1999) characterized MNC subsidiaries as
world product mandate and regional product mandate by distinguishing their subunit-
level capabilities. Almeida (Almeida, Dokko, & Rosenkopf, 2003) discussed the same
question by unbundling the process of knowledge management, defining this process as
80
search, transfer, and integration, and linking each of these components with subunit
capabilities.
Gupta and Govindarajan (1991) created a new typology of subsidiaries by
applying a knowledge flow based construct. They applied the magnitude and the
direction of knowledge flows of each subunit. (i.e. Subsidiaries are the provider or
receiver of knowledge). Four subunit roles are identified in their model: Global innovator,
integrated player, implementer, and local innovator. Similar categorization can be found
in Ambos and Reitsperger’s study (2004), which distinguishes between technological
mandate subsidiaries and task-related interdependence subsidiaries. Birkinshaw and
Morrison (1995) provided a three-fold typology which classifies subsidiaries as local
implementer, specialized contributor and world mandate. These studies are all based on
the characters and natures of intra-MNC knowledge flow. Cantwell and Janne (1999)’s
research made efforts around the same topic, but from a different approach. Their work
compared the technological interrelatedness of R&D activities in foreign centers with
those in domestic countries. In other words, the research activities by foreign facilities are
distinguished as either ‘replication’ or ‘diversification’.
The analysis of subunit roles in this paper, however, takes a different route. We
distinguish three types of subunit roles according to their relative degrees of involvement
in either the MNC network or the local network in which they operate. Namely the three
types of subsidiaries are international competence exploitation subsidiaries, local
competence exploration subsidiaries, and global competence creation subsidiaries.
Intuitively, there can be a trade-off effect between the subunit’s MNC integration and its
81
local embeddedness given the limited resource of management and investment focus.
However, as suggested by some researchers, in certain circumstances when the MNC
operates within (and sometimes contributes to) a fast changing environment, this trade-
off effect could be eliminated, and instead, a complementary relationship between the
involvements into the two networks could exist. Hence, a subunit can have local
competence creating activities such as knowledge transferring and innovation based on
its relations with local partners and resources, while increasingly relying on competence
exploiting exchanges within its MNC group. Together, they reinforce the subunit’s
position in both local and MNC networks.
A subunit’s involvement in each of the two networks can be demonstrated in a
two-two matrix (Figure 8). Using the degree of MNC integration as one dimension, and
the degree of local embeddedness as the other, each subunit can be allocated to a cell in
this metrics.
ICE
LCE
GCC
MNC Integration
Local Embeddedness
H
H
L
L
Figure 8 Subunit typology by knowledge sourcing directions
82
International Competence Exploitation subunit
A subunit appearing on the upper-left side indicates its higher degree of MNC integration
and lower degree of local embeddedness. It means that the subunit’s technological
activities are largely dependent on the parent group’s technological trajectory, while its
geographical position is simply a strategic choice of the MNC group. The mandate of
these subsidiaries is to exploit the existing competences of the parent group in a new
(foreign) location, which is usually local market oriented. We name this type of subunit
the International Competence Exploitation subunit (ICE). The function of ICE subunit
is therefore closely associated with the product life cycle model (Vernon, 1966), which
indicates the spreading of innovation from home country towards foreign markets.
This is the prevalent case in the 1950s-60s. In this period, large MNCs
internationalize their production in the seeking of foreign markets. In this case, the
linkage between a subunit and its parent group (both headquarter and other subsidiaries)
is very strong. Subsidiaries are under the centralized control of the parent group, and
inter-unit networking is encouraged for the benefit of the whole group’s global strategy.
The more integrated a subunit become into its MNC network, the more likely that the
subunit will be able to appreciate new developments in this intra-firm network, and
therefore the more easier it can acquire and assimilate new knowledge from this network.
Local Competence Exploration subunit
83
A subunit appearing on the lower-right side of the matrix in Figure 1, on the other hand,
indicates a higher degree of embeddedness into the local network, while a lack of
integration within the MNC network. This type of subunit is Local Competence
Exploration subunit (LCE). On the one hand, it actively interacts with its local business
partners, competitors, or even government and other research institutes, in regards to their
technological activities. On the other hand, it maintains only a weak connection with its
MNC group, and its technology developments aren’t necessarily following the path of the
whole MNC network. One typical and extreme example of this type of role is a newly
acquired subunit. This subunit would have an established business network in its local
environment, while its connection with headquarter and other parts of the MNC is very
weak.
In this case, the more embedded the subunit is in its local network, the more likely
it is able to acquire and assimilate new knowledge from this network. However, if the
local network and MNC network have different focuses on technology, the more locally
embedded the subunit is, the less likely it is to have an equally strong capacity to gain
new knowledge from the MNC network.
Global Competence Creation Subunit
Although historically, both ICE and LCE type of subsidiaries have their prevalence in
MNC strategies, neither of these types can serve the purpose of MNC internationalization
in the fast changing global economy nowadays. Given the globalization of technological
activities and knowledge accumulation, subsidiaries are not competitive if they are too
dependent on either network. The conflicts between the alternative networks in previous
84
stages that caused obstacles for knowledge exchange between parties are now less
significant as the opening and interaction of both networks take place in the dual-course
of subunit evolution (Cantwell, 2008).
For an international competence exploitation subunit, focusing on MNC mandates
would no doubt help establishing the advanced technological capability to fulfill the
manufacturing needs for the MNC group, which in turn allows the subunit to achieve a
strategically important position within the MNC group. However, lack of interaction with
local business partners or industrial clusters can result in the loss of emerging technology
advantages. Moreover, cluster economy can provide a local expertise group on certain
advantageous fields, feeding on the information interchange in this type of group could
help the subunit to benefit from local sources of knowledge, which is not necessarily
related closely to the original mandate of the subunit, but could be equally, if not more,
critical to the competition in both the local industry cluster and the expanding global
market.
For a local competence exploration subunit, on the other hand, the benefit of local
business network is fully appreciated. Clustering in specific locations which represent
“centers of excellence” of one or more industries could increase the chances of being
exposed to the most advantageous technology, so that subsidiaries can be focusing on the
local industry or market. However, since the local focus isn’t always consistent with
MNC’s strategic planning. Being overly committed to a local network and building up its
unique competitive advantage will cause a subunit to fall out of the strategy direction of
85
its MNC group, which can result in a weakened position within the MNC. In an extreme
case, this would result in spin-offs.
Hence, a more efficient type of subunit role emerged in the evolution path – the
Global Competence Creation subunit (GCC). Subsidiaries with a GCC role are not only
highly integrated within the MNC’s strategic blueprint, but also tightly embedded in its
local business network. When MNCs move towards a more closely integrated
international network, foreign technological activities carried out by affiliates are not
limited to adopting and developing existing competencies, but rather incrementally
targeting at new ones. Cantwell and Noonan (2007) proposed that the technological
activities of foreign subsidiaries located in Germany are not heavily concentrated in home
base exploiting-type activities, but on exploring-type activities.
Zander (1998) suggests that foreign unit activities are associated with a
significantly higher probability of entry into new and more distantly related fields of
technology, creating a long-term drift into new technological capabilities. In other words,
foreign subsidiaries on average are getting more focused on the ‘R’ in the R&D process
and evolving with the motivations to discover some new capabilities. In this case, an
increasing number of competency-creating subsidiaries will be less reliant upon
knowledge sourcing from their parent company or from other organizations in the home
country of their MNC corporate group. Their strategies are more likely to be locally
accumulative in some specialized fields of knowledge-building that are distinct
(diversified) from the competence base of their parent. In our dynamic model of subunit
role, this evolutionary process can be represented by the arrow from the upper-left side to
86
the upper-right side in Figure 1. It indicates that over time, ICE subsidiaries are shifting
towards GCC subsidiaries. They are combining the local expertise with their MNC
knowledge, creating subunit-specific competence by getting more embedded into their
local network. This type of subunit evolution is largely driven by subunit level initiates,
while enhanced by the support of MNC group and local network.
Meanwhile, the LCE subsidiaries are becoming more integrated into the MNC
network. In order to avoid the problem of repeated R&Ds in geographically dispersed
subsidiaries, and to take full advantages of different locations, MNC reinforce its control
over the LCE subsidiaries and hence increase their contribution to the parent group.
Driven largely by the MNC group, LCE subsidiaries are shifting into the role of GCC,
keeping their advantageous local connection but enhancing their competence creating
capability by closely cooperating with other parts of the MNC network.
Therefore, subsidiaries evolving towards a GCC role tend to achieve both
capability enhancement and charter establishment. They create unique competence by
interacting with partners in both MNC and local network, and hence develop a
strategically important position within the MNC group as an active innovator, while
embracing the benefits of a certain degree of autonomy in its location, which in turn
guarantees its knowledge absorbing capability in each of the networks.
5.2 Typology Based on Innovation Pattern – CC intensity vs. sub distance
Now if we take a look at the innovation patterns of foreign subunits, instead of their
knowledge sourcing compositions, it is understandable if we shift focus to the actual
87
competence creating / competence exploiting activities. As discussed in Chapter Four, the
composition of these two types of activity could determine a subunit’s strategic position
within the MNC group. In this part, I will demonstrate a different typology of subunit
roles based on the extent to which a subunit’s innovative activities are CC as opposed to
CE, and on the technological distance between the subunit and its parent company.
The first dimension is defined as subunit CC intensity. It is measured with the share of
competence creating activities among all innovative activities (CC+CE) within a subunit.
The construction of this variable takes two steps. First, I identified the innovative
activities in fields that are categorized as competence creating vis-à-vis competence
exploiting by comparing the RTA value of subunit i in field j (RTAij) to the RTA value of
subunit i’s parent company in field j (RTApj). If RTAij >1 and RTApj <1, field j is
considered a field of CC for subunit i. This comparison is done by all subunits in all
technological fields. Second, I calculate the number of patents that belong to these CC
fields for each subunit, and then divide this number by the total number of patents for
each corresponding subunit. The result is a percentage number indicating the CC
intensity of a subunit.
The second dimension is subunit distance – the technological distance between
innovative activities of a subunit and its parent company. A detailed definition and
calculation of this variable can be found in Chapter Two.
Combining these two dimensions generates another two-by-two matrix (as shown in
Figure 9).
Subunit Distance
H
Figure 9 Subunit Typology by Competence Creativeness
88
Subunits that are positioned in the lower left corner (Quadrant III of Figure 5) have both
low degree of CC intensity and low distance from its parent company. This type of
subunit is mostly miniature replica of the parent company in a foreign location; its
innovation is mainly incremental ones based upon the technological specialties of the
parent company. This fits the description of traditional subunits, which have a large share
of CE activities as opposed to CC.
Subunits positioned at the upper right corner (Quadrant I of Figure 5), on the other hand,
is the other extreme of the case. These subunits are not just highly innovative, but their
innovation is to a large extent composed of CC activities, and their portfolios of these
activities are significantly different from that of their parent companies’. These subunits
are of high strategic importance to an MNC group as they bring in new potentials to the
entire group, especially when some of these CC areas are matured, the group can then
recognize them as part of its new core competency, and therefore spread these specialties
across the MNC group’s international network. Their strategic role could be categorized
the ultimate competence creating type, and they are considered centers of excellence
within their MNC group’s network.
Subunits that fall under the category of high technological distance from its parent
company while having low CC intensity (Quadrant II of Figure 5) could be the ones that
has a technology portfolio that includes mostly in some certain areas of expertise that are
same as the parent company’s areas of specialization, but their technological field
diversification is highly skewed by a few outliner fields that don’t fall into the categories
of parent company specialization. These subunits are developing some niche applications
89
based on their local host country competency, while their main innovative activity is still
focused on the exploitation of MNC group specialties.
In the lower right quadrant (Quadrant IV of Figure 5), subunits are found to have high
degree of CC intensity with low technological distance compared to their parent
companies. These subunits are the ones of most interest in my typology. They
concentrate their innovative efforts on areas that are not particularly specialized by the
parent company. However, these areas are not completely unrelated or brand new to the
MNC group. Their technology portfolio shows consistency with the parent company, just
that they tend to concentrate more on the fields that are relatively new to the group. The
new knowledge combinations these subunits are making are close to the core competency,
and they conduct a more focused exploration compared to those with a CC role or a
Niche Application Role. We therefore categorize this type of subunit as “Core Base
Extension”. These subunits have the most potential to bring in new competencies to the
MNC group and incorporate these new areas of technology into the development of new
group level core competency.
Figures 10 and 11 show a distribution of subunits among these four categories across the
dimensions of subunit technological distance and subunit CC intensity. Figure 10 samples
the full dataset while Figure 11 focuses on large subunits with more than patents in a five
year window.
90
Figure 10 Subunit Distance and CC Intensity (All Data)
Figure 11 Subunit Distance and CC Intensity (Large Subunits)
91
Comparing these two diagrams, larger subunits seems to demonstrate a stronger positive
relationship between dimensions of subunit technological distance and subunit CC
intensity. Large subunits distribute more closely along the fitted regression line, while
smaller subunits tend to have more various types of positioning as shown on the diagrams.
5.3 Evolution of Subunit
To examine the trajectories of subunit role evolution, I use the typology developed in the
previous part with dimensions of subunit CC intensity and subunit technological distance.
By examining the change in each of these dimensions over different time periods, I am
able to construct a new model that distinguishes subunit evolution paths based on their
directions of movement. Figure 12 shows the categorization of these movements.
Figure 13 shows the scattered distribution of subunits’ evolution patterns. Since this chart
uses change of value of the two dimensions in Figure 5 as the axis, it is intuitive to use
Delta SC
Delta_SD
Figure 12 Categorization of Subunit Evolution Patterns
92
value 0 as the dividing cut-up point to distinguish types of subunit evolution pattern. I’m
using all data sample in this chart. The large subunits sample shows almost the same
results, except with a few less observations. The reason of this is that subunits have to
show a consistency in their overall competence creativeness throughout more than one
time window to be included in this result. Lots of subunits are not showing this
consistency; therefore it’s hard to capture their evolution patterns. Large subunits,
however, are more likely to show up repeatedly in this chart because of the capability of
conducting continuous CC activities.
Figure 13 Subunit Evolution Patterns
In Quadrant I of Figure 13, subunits are demonstrating an increase in both their
technological distance from their parent group and their CC intensity. These subunits are
probably those that are heavily embedded in host country local environment while at the
same time possess a favorable mandate from the MNC group to facilitate their further
development of innovative activities that are new to the MNC group. 42 subunits are
identified in QI, among which the most consistency is shown by the subunit of the Swiss
firm Novartis in the US. Novartis is a pharmaceutical company that has its center of
93
research and development based in Basel, Switzerland, with other key research sites
located in Horsham, UK; Vienna, Austria; Tsukuba, Japan; East Hanover, New Jersey,
USA. Their strategy for innovation is “focused diversification” (from Novartis website),
by which they mean a concentration on pharmaceutical, biomedical innovation while
encouraging diversification into related areas. It is shown in the data that Novartis US
started as a traditional type of subunit, with technological distance of 0.04 and CC
intensity of 0.04; in the following period this pair of numbers were increased to (0.09,
0.30), whereas towards the end of my observation period the numbers reached (0.11,
0.40). It is clear that Novartis US is on the trajectory of evolving into a category of more
emphasize on CC type of activities.
Quadrant II of Figure 13 includes subunits that are increasing in their technological
distance from the parent company, yet the level of CC activity intensity decreases
overtime. These subunits are moving into the direction of concentrating more on CE
activities, but their areas of niche application stays strong and even draws further away
from the parent company. A typical example of subunit in this quadrant is ExxonMobil’s
operation subunit in Canada.
Quadrant III is consisting of subunits that are decreasing on both technological distance
and CC intensity. Depending on the actually type of subunit, falling into this type of
evolutionary trajectory could mean two things: (1) the subunit is already well advanced in
both technological distance and CC intensity as a center of excellence, now it is time for
the competence created by this subunit to be transferred back to the parent company and
then further to the international network of MNC group – as a result the subunit itself
94
demonstrates a relatively lower level of CC intensity and technological distance overtime;
(2) the subunit is less of a center of excellence, and the field of specialization which is
new to the parent company is either no longer a specialty of the subunit itself, or
somehow adapted by the parent company. One typical example is US company Schering
Plough’s operation in Germany. Its subunit technological distance dropped from 0.033 to
0.004, while its CC intensity dropped from an outlining 0.949 all the way down to 0.387.
This is an extreme case due to a merger between the US company Plough Inc. and
Germany company Schering Corporation. In this case, the reduction of CC activity
presence is showing the effect of two companies corresponding and transmission
technological capabilities into one another.
Quadrant IV indicates an increase in CC intensity while the technological distance
decreases. This could be caused by a more focused effort on CC areas that are to some
extent related to the parent company’s fields of specialization – just like apples don’t fall
far from the tree. An example would be Novartis’s operation in Germany, which follows
perfectly to the company’s strategy – “focused diversification” as explained in previous
discussion.
95
6. Conclusion
The innovative activities of multinational corporation (MNC) operations overseas can be
represented as two types: either competence exploiting (CE) – exploiting the core
competence base of the parent group – or competence creating (CC) – creating new
competencies that were not already among the strengths of the relevant parent company.
To a large extent, the share of these two types of activities determines and reflects a given
subunit’s strategic role within its MNC. This research examines (1) the patterns of MNC
subunits’ knowledge sourcing in terms of the technological and geographical dispersion
of knowledge sources; (2) the extent to which MNC subunits’ technological fields of
expertise are distinct from those of their parent companies, and how this technological
distinctiveness is related to their knowledge sourcing patterns; and (3) how MNC
subunits’ profiles of CC and CE activities (in terms of their overall technological distance
from their parent companies, and the degree to which they are engaged in CC versus CE
activities) evolve over time, reflecting the evolution of their knowledge creating role and
status within their international group. Attention is focused on the heterogeneity of firm-
specific evolutionary paths in the patterns of knowledge accumulation that support CC
activities, controlling for the industry-specific determinants, location-specific factors, and
MNC group structural influences on such technological trajectories.
Three research questions were brought up in the Introduction chapter, now let’s revisit
every one of them:
96
Research Question 1: what are the patterns of knowledge sourcing of an MNC subunit
with respect to its technological dispersion, geographic dispersion, and level of sourcing
concentration in its two main sources – home country and host country?
Three studies were designed around patent and patent citation data of large multinational
corporation groups in the chemical industry to examine these three questions. For
Research Question 1, we find out that subunits tend to have a knowledge sourcing pattern
of increasing dispersion both in technological fields and in geographical locations. There
is a positive relationship between technological dispersion and geographical dispersion,
but the effect is moderated by the effect of share of home country sourcing and share of
host country sourcing. The patterns of knowledge sourcing have positive relationship
with the technological diversification of subunit, which in turn has an implication on the
influencing patterns discussed in Study Two.
Research Question 2: how does a subunit’s knowledge sourcing pattern influence its
local competence creating capability, hence the technological distinctiveness from its
parent company?
This part of study focuses on the technological distinctiveness as an indicator of
competency creativeness of subunits. We found out that the technological dispersion of a
subunit’s knowledge sourcing for activities in a certain field has a positive relationship
with the subunit’s technological distinctiveness of that field, while the geographical
dispersion of sourcing in a given field has a negative relationship with technological
distinctiveness. These findings confirms the relationship between knowledge sourcing
97
patterns and subunit technological distinctiveness, while supporting the streams of
current literature on subunit technological diversification, internationalization, and local
embeddedness.
Research Question 3: if knowledge sourcing and competence creation are two mutually
interdependent processes, is there a typology of evolutionary trajectories followed by the
paths of technological growth of subunits?
Study 3 examines this last question with exploratory data analysis. Two typologies of
subunit roles are identified, with the first one categorizing subunits based on their
knowledge sourcing patterns, and the second one categorizing subunits based on their
competence creativeness patterns. Based on the results from studies one and two, these
two typologies could be interconnected due to the interdependency between these two
sets of concepts. Four types of evolutionary patterns were recognized in the last section,
with regards to the second typology of subunit role proposed in that chapter – subunit
technological distance and subunit CC intensity.
Contributions of this dissertation are three fold. Theoretically, the studies contribute in
areas of MNC knowledge accumulation and competence creating by establishing new
connection between the two and further on identifying evolutionary trajectory of MNC
subunit strategic role development. Empirically, the studies created new ways of studying
MNC subunit innovation, as well as a clear typology of subunit role based on its
competence creativeness. It also examines the relationship between subunit knowledge
sourcing and innovative activity, which could lead to further development of study on
98
these two interdependent characteristics of subunit strategy. Finally, methodologically,
the studies propose a combination of two dimensions – subunit technological distance
and CC activity intensity – to indicate strategic positioning of the subunit within an MNC
international network, this method takes into consideration of the composition of subunit
overall innovativeness and its novelty compared to parent company, which gives us a
comprehensive view of subunit competence creativeness.
However, due to the nature of data (patent database) used in this study, it is very difficult
to identify real subsidiaries entities for analysis. Instead I’m using the firm-country
combination to indicate an MNC’s innovative activity in a foreign country. This may
cause some problems especially for a large country like the US, since there is huge
variation across regions in this country. A potential solution is to divide the country into
different regions, so that these regions can be comparable to other countries.
Another limitation of the data is the identification of patent affiliations of citations. The
current dataset doesn’t provide this information. If further research on be based on a
complete data of patent and corporate affiliation information, we can further examine the
relationship between inter- firm and intra- firm knowledge sourcing patterns, and their
impact on subunit innovation.
As discussed in Chapter Two, patent continuation could result in multiple count of single
invention, as well as high interdependence of different patents within one firm. If
invention level data could be acquired, it will avoid having this problem in research. One
of the suggestions for future improvement of this study is to rule out non-independent
99
patents by the same inventors within each firm. This process requires using the original
USPTO dataset to identify all inventor information, as well as patent continuation
information.
Future study could concentrate more on a comprehensive empirical analysis of the
evolutionary trajectory for subunits. Location industry conditions could be taken into
consideration when analyzing these evolutionary trends.
100
APPENDICES
Appendix A1 – Firm List
Firm Name Home Country 3M US Abbott Laboratories US Air Liquide France Air Products and Chemicals US Akzo Nobel Netherlands Albright and Wilson UK Allergan US Allied US American Cyanamid US American Home Products US Amgen US Asahi Kasei Japan Ashland US Astellas Pharma Japan AstraZeneca UK Avery Dennison US Avon Products US Basell Netherlands BASF Germany Bausch Lomb US Baxter International US Bayer Ag Germany Beecham Group UK Biogen Idec US BOC UK Boehringer Ingelheim Germany BP UK Bristol-Myers Squibb US Cabot US Celanese US Cephalon US CF Industries US Charbonnages de France France Chemische Werke Huls Germany Chemtura US Chesebrough Ponds US
101
Chevron Phillips Chemical US Ciba Specialty Chemicals Switzerland Clariant Switzerland Clorox US Cognis Germany Colgate-Palmolive US ConocoPhillips US CSL Australia Cytec Industries US Daicel Chemical Industries Japan Daiichi Sankyo Japan Dainippon Inks & Chemicals Japan Dainippon Sumitomo Pharma Japan Denki Kagaku Kogyo Japan Dow Chemical US Dow Corning Corporation US E.I.Du Pont De Nemours and Company US Eastman Chemical US Ecolab US Eisai Japan Eli Lilly US Enterprise Miniere et Chemique France Ethyl US ExxonMobil US Firm_name home_country Forest Laboratories US Formosa Chemicals & Fibre Taiwan Formosa Plastics Corporation Taiwan Fuji Photo Film Japan Genetech US Genzyme US Gilead Sciences US Glaxo UK GlaxoSmithKline UK H. Lundbeck Denmark Hanwha Chemical Korea Henkel Germany Hercules US Hexion US Huntsman US ICI UK
102
Int Minerals & Chemicals US Johnson Johnson US Johnson Matthey UK Kaneka Japan Kao Corporation Japan Kemira Finland King Pharmaceuticals US Koppers US Kuraray Japan Kyowa Hakko Kogyo Japan Lanxess Germany LG Chem Korea Lion Japan Lonza Switzerland Loreal France Lyondell Chemical US Marathon Oil US Merck US Mitsubishi Chemical Holdings Corporation Japan Mitsubishi Gas Chemical Japan Mitsubishi Petrochemicals Japan Mitsubishi Pharma Japan Mitsui Chemicals Japan Mitsui Petrochemical Japan Monsanto US Morton Norwich Products US Nalco US Norsk Hydro Norway Nova Chemicals Canada Novartis Switzerland Novo Nordisk Denmark Olin US Orica Australia PennWalt US Pfizer Inc US Potash Corp. of Saskatchewan Canada PPG Industries US Praxair US Procter & Gamble US Reckitt Colman UK Reichhold Chemicals US
103
Revlon US Rhodia France Rhone-Poulenc France Richardson Vicks US Roche/Sapac Switzerland Rohm and Haas US Rutgerswerke Germany Sanofi Aventis France Schering Plough US SCM US Sekisui Chemical Japan Sherwin Williams US Shin-Etsu Japan Shionogi Japan Shiseido Japan Showa Denko Japan Smith Kline Beckman US Solutia US Solvay Belgium Squibb US Stauffer Chemicals US Sterling Drug US Sumitomo Chemical Japan Syngenta Switzerland Taiho Pharmaceutical Co Japan Takeda Pharmaceutical Co Japan Teva Pharmaceutical Industries Israel Tosoh Japan Transammonia US Ube Industries Japan Union Carbide US Upjohn US Valero Energy US Valspar US W.R. Grace US Wacker-Chemie US Warner Lambert US Watson Pharmaceuticals US Wyeth US Yara International Norway
104
105
Appendix A2 - Country and patent distribution list
Country Patents Patents
by Parent Patents by
Subunit USA 153,916 130,854 23,062 Germany 40,341 25,768 14,573 UK 13,613 7,819 5,794 Italy 1,445
1,445
France 9,646 7,965 1,681 Japan 47,776 45,240 2,536 Netherlands 1,083 460 623 Belgium 2,727 485 2,242 Switzerland 8,445 7,876 569 Sweden 856
856
Denmark 1,167 992 175 Ireland 151
151
Spain 193
193 Portugal 3
3
Luxembourg 30
30 Greece 8
8
Austria 303
303 Norway 318 278 40 Finland 106 82 24 German Democratic Republic (1945-1989) 1
1
Hungary 37
37 Poland 10
10
Czechoslovakia 5
5 Yugoslavia 4
4
USSR 71
71 Canada 2,427 67 2,360 Australia 526 79 447 New Zealand 37
37
India 117
117 Brazil 96
96
Israel 178 140 38 Argentina 8
8
Chile 12
12 Colombia 6
6
Mexico 30
30 Panama 1
1
Peru 1
1
106
Venezuela 12
12 Philippines 17
17
South Korea 674 151 523 Taiwan 48 6 42 Turkey 1
1
South Africa 53
53 West Indies and Guianas 8
8
Bulgaria 1
1 East Indies 14
14
Other Latin America 9
9 Other Africa 11
11
China 38
38 Uruguay 1
1
Ecuador 1
1 Other Asia 11
11
Other Middle East 13
13 Other Europe (Monaco, Liechtestein, Iceland, Malta, Andorra, Albania, Cyprus, Greenland) 7
7
Hong Kong 31
31 Singapore 23
23
Total 286,667 228,262 58,405
107
Appendix A3 – Geographic distribution of patents – by parent companies and by foreign subunits Geographic distribution of patents invented by parent companies:
Geographic distribution of patents invented by subunits:
Patents by Parent
USA
Germany
UK
Italy
France
Japan
Netherlands
Belgium
Switzerland
Denmark
Patents by Subunit
USA
Germany
UK
Italy
France
Japan
Netherlands
Belgium
Switzerland
Denmark
108
Appendix A4 – 31 Technological Fields and Corresponding Patent Class Code
Tech31 Field Name Patent class (sub-class) 1 Food and tobacco products 127 (29-71), 131 (1-226, 291-999), 426 2 Distillation processes 201, 203 3 Inorganic chemicals 423 4 Agricultural chemicals 71, 504 5 Chemical processes 23, 51 (293-328), 55 (1-99), 62 (1-122), 95,
117, 134 (1-42), 156 (1-479, 600-668), 204 (1-192, 900-914), 205, 210 (501-982), 216, 260 (95, 684-708), 427, 432 (1-53), 518
6 Photographic chemistry 430 7 Cleaning agents and other
compositions 106, 252, 508, 510, 512, 516, 588
8 Synthetic resins and fibres 260 (1-94, 666-683, 709-999), 520,521, 522,523, 524, 525, 526, 527, 528
9 Bleaching and dyeing 8 10 Other organic compounds 260 (96-665), 530, 534, 536, 540, 544, 546,
548, 549, 552, 554, 556, 558, 560, 562, 564, 568, 570, 930, 987
11 Pharmaceuticals and biotechnology 424, 435, 436, 514, 800, 935 12 Other chemicals and Related -
disinfecting, preserving, textiles and explosives
422, 2, 36,245,289,450, 149
13 Metallurgical processes 29, 75, 148, 164 (1-148, 101-265), 419, 420 14 Miscellaneous metal products 3,4,7, 10, 16,24, 27, 30 (1-165, 167, 395-499,
501-999), 49, 63, 70, 108, 109, 124, 132, 135, 138, 150, 160, 182, 190, 206, 211, 215 (100-367), 220, 232, 248, 256, 267, 272, 279, 285, 292, 312, 383, 403, 411, 464, 623
15 Chemical and allied equipment 34, 51 (1-292, 329-999), 55 (100-999), 68, 96, 118, 134 (43-999), 156 (480-599, 699-999), 159, 196, 202, 209, 210 (1-500), 261, 366, 422 (44-999), 494, 502, 503
16 Paper making apparatus 53, 162, 229, 493 17 Assembly and material handling
equipment 186, 187 (1-28, 30-999), 193, 198, 212,224, 226, 242, 254 (134-999), 258, 271, 294, 402, 406, 410, 414, 901
18 Other specialised machinery 15, 30 (166, 168-394, 500), 79, 98, 100, 116, 133, 140, 141, 147, 157, 169, 194, 221, 222, 227, 254 (1-133), 277, 291, 300, 401, 425, 452
19 Other general industrial equipment 48, 91, 92, 110, 122, 126, 137, 165, 184, 185, 188, 192, 237, 239, 251, 303, 415, 416, 417, 418, 431, 432 (54-999)
109
20 Mechanical engineering nes 99, 127 (1-28), 131 (227-900), 59, 72, 76, 81, 82, 83, 163, 164 (149-999), 173, 225, 228 (1-100), 234, 266, 269, 308 (1-9, 11-245), 384, 407, 408, 409, 413, 474, 65 (138-999), 241 (132-999), 249, 56, 111, 130, 172, 278, 460, 37, 171, 404 (83-133), 166, 175, 299, 445, 12, 19, 26, 28, 38, 57, 66, 69, 87, 112, 139, 223, 101, 199, 270, 276, 281, 282, 283, 412, 462, 142, 144, 145, 235 (61-89; 419-434), 400
21 Electrical devices and systems 174, 200, 307 (1-199, 586-999), 308, 323, 327, 328, 330, 331, 333, 334, 335, 336, 337, 338, 339, 361, 363, 372, 439, 505
22 Other general electrical equipment 62 (123-999), 136, 204 (193-499), 219, 236, 290, 310, 318, 320, 322, 361 (433-436), 373, 388, 392, 429, 437, 438
23 Office equipment and data processing systems
235 (375-386, 400-418, 435-457), 360, 364, 365, 369, 371, 377, 700, 701, 702, 703, 704, 705, 706, 707, 708, 709, 710, 711, 712, 713, 714, 715, 716, 717, 718, 719, 720, 725, 726, 902
24 Electrical equipment nes 178, 179, 329, 332, 367, 370, 375, 379, 455, 340, 341, 382, 342, 343, 84, 181, 348, 358, 381, 386, 725, 313, 314, 315, 362, 357
25 Transport equipment 123, 180, 296, 244 (1-13, 15-999), 114 (1-19, 26-999), 440, 441, 104, 105, 213, 238, 246,191, 280, 293, 295, 298, 301, 305
26 Rubber and plastic products 152, 264 27 Non-metallic mineral products 52, 65 (1-137), 125, 215 (1-99), 241 (1-131),
428, 442, 501 28 Coal and petroleum products 44, 208, 585 29 Photographic equipment 354, 355, 396, 399 30 Other instruments and controls 33, 73, 74, 128, 177, 187 (29), 235 (1-60, 90-
374, 387-399, 458-999), 250, 324, 346, 347, 349, 350, 351, 352, 353, 356, 359, 368, 374, 378, 385, 398, 433, 475, 600, 601, 602, 604, 606, 607
31 Other manufacturing and non-industrial
60, 376, 976, 5, 217, 297, 6, 14, 17, 40, 42, 43, 47, 54, 86, 89, 102, 114 (20-25), 119, 168, 231, 244 (14), 273, 380, 404 (1-82, 134-999), 405, 434, 446, 449, 452, 463
110
111
Appendix B – Table of Correlation for Key Variables in Study 1
cng_td ced_td ced_gd hsced_td nhsce~td nhsce~gd hmced_td nhmce~td nhmce~gdhost_td US_home US_host hs_share hm_share gpt_sh~e ced_gp~e intreg~ecng_td 1.000ced_td 0.3019* 1.000ced_gd 0.1250* 0.4720* 1.000hsced_td 0.3045* 0.6020* -0.010 1.000nhsced_td0.2474* 0.6956* 0.2203* 0.4201* 1.000nhsced_gd0.1782* 0.2731* 0.4923* 0.3222* 0.4491* 1.000hmced_td 0.2547* 0.6154* 0.016 0.7595* 0.5653* 0.2415* 1.000nhmced_t 0.2847* 0.6801* 0.2190* 0.5470* 0.8566* 0.4200* 0.4170* 1.000nhmced_g0.1520* 0.2600* 0.4813* 0.2121* 0.3999* 0.8243* 0.2856* 0.4530* 1.000host_td 0.5127* 0.1912* 0.1536* 0.2127* 0.1909* 0.1600* 0.1600* 0.1980* 0.1446* 1.000US_home -0.009 0.0469* -0.1616* 0.1305* 0.008 0.0755* 0.2428* -0.0797* 0.1954* -0.0715* 1.000US_host 0.2693* 0.1911* -0.1248* 0.3869* 0.0605* 0.3127* 0.2670* 0.1417* 0.1712* 0.1680* 0.3598* 1.000hs_share 0.1216* -0.1080* -0.6862* 0.3488* -0.0763* 0.1450* 0.2062* -0.0314* -0.1383* 0.2042* 0.3026* 0.7044* 1.000hm_share -0.0551* -0.1984* -0.7135* 0.1422* -0.1323* -0.2388* 0.2481* -0.1648* 0.0525* -0.1281* 0.6662* 0.2887* 0.6159* 1.000gpt_share 0.1924* 0.0515* 0.0190* 0.0634* 0.0418* 0.0154* 0.0304* 0.0502* -0.0227* 0.025 -0.0465* 0.0327* 0.013 -0.0430* 1.000ced_gpt_s 0.0722* 0.1116* -0.0324* 0.1299* 0.1417* 0.0166* 0.1164* 0.1383* 0.008 0.020 0.008 0.014 0.002 0.008 0.0696* 1.000intreg_sha-0.2120* -0.2212* -0.0288* -0.2436* -0.0258* -0.2703* -0.1781* -0.0694* -0.1224* -0.1944* -0.2068* -0.5728* -0.7815* -0.2089* -0.0289* 0.009 1.000
112
Appendix C – Table of Correlation for Key Variables in Study 2
Dissim1 RTA_host logsize struct sub_sh~e ced_td ced_gd nhsce~td nhsce~gd nhmce~td nhmce~gd
T. Distinct 1
RTA_host 0.1249* 1 logsize -0.1620* 0.0075 1
FirmCon 0.0588* 0.0448* 0.1736* 1 sub_share 0.1499* -0.0253* -0.1153* 0.1046* 1
Tech Disp 0.0579* -0.0535* 0.0544* -0.0514* 0.0766* 1 Geo Disp 0.0113 -0.0157 0.0061 -0.0279* 0.0427* 0.4066* 1
nhsced_td -0.0173 -0.0441* 0.1173* -0.0704* 0.0645* 0.7421* 0.2016* 1 nhsced_gd -0.0545* -0.0027 0.0315* -0.1029* 0.0319* 0.2041* 0.5901* 0.3648* 1
nhmced_td 0.0063 -0.0641* 0.0386* -0.1851* 0.0904* 0.7019* 0.2233* 0.7020* 0.3306* 1 nhmced_gd -0.0431* -0.006 0.0611* -0.1061* 0.0303* 0.1969* 0.6105* 0.2909* 0.5996* 0.3858* 1
*: p < .05 Sample: All data
113
Distin RTA_host logsize FirmCon sub_sh~e Tech Disp Geo Disp nhsce~td nhsce~gd nhmce~td nhmce~gd
T. Distinct 1
RTA_host 0.0524* 1 logsize -0.1890* -0.0345 1
FirmCon 0.0410* 0.0137 0.1480* 1 sub_share 0.2544* -0.0224 -0.3173* 0.3736* 1
Tech Disp 0.0337 -0.0567* 0.0507* -0.0627* 0.0162 1 Geo Disp 0.0114 0.0381 0.0479* -0.0356 0.0226 0.3069* 1
nhsced_td -0.0707* -0.0283 0.0918* -0.0620* 0.025 0.6484* 0.1453* 1 nhsced_gd -0.1316* 0.0301 -0.0656* -0.1111* -0.0264 0.1557* 0.3483* 0.3964* 1
nhmced_td 0.0245 -0.0282 -0.0026 -0.1285* 0.018 0.7585* 0.0875* 0.6634* 0.2992* 1 nhmced_gd -0.0659* 0.035 0.1158* -0.0630* 0.0005 0.0804* 0.6169* 0.2434* 0.4214* 0.2030* 1
*: p < .05 Sample: Large subunits
114
REFERENCES
Almeida, P., Dokko, G., & Rosenkopf, L. 2003. Startup size and the mechanisms of external learning: Increasing opportunity but declining usefulness? Research Policy, 23(2): 301–315.
Ambos, B., & Reitsperger, W. 2004. Offshore centers of excellence: social control and success. Management International Review, 44(Special Issue 2): 51–65.
Andersson, U. 2003. Managing the transfer of capabilities within multinational corporations: the dual role of the subsidiary. Scandinavian Journal of Management, 19: 425-442.
Andersson, U., Bjorkman, I., & Forsgren, M. 2005. Managing subsidiary knowledge creation: the effect of control mechanisms on subsidiary local embeddedness. International Business Review, 14: 521-538.
Andersson, U., & Forsgren, M. 1996. Subsidiary embeddedness and control in the multinational corporation. International Business Review, 5(5): 487-508.
Andersson, U., Forsgren, M. & Holm, U. 2001. Subsidiary embeddedness and competence development in MNCs – a multi-level analysis. Organization Studies, 22(6): 1013-1034.
Andersson, U., Forsgren, M., & Holm, U. 2002. The strategic impact of external networks – subsidiary performance and competence development in the multinational corporation. Strategic Management Journal, 23:979-996.
Andersson, U., Forsgren, M., & Holm, U. 2007. Balancing subsidiary influence in the federative MNC: a business network view. Journal of International Business Studies, 38: 802-818.,
Andersson, U., Forsgren, M., & Pedersen, T. 2001. Subsidiary performance in multinational corporations: the importance of technology embeddedness. International Business Review, 10:
Alcacer, J. & Gittelman, M. 2006. Patent citations as a measure of knowledge flows: the Influence of examiner citations. The Review of Economics and Statistics, 88(4): 774–779.
Alcacer, J., Gittelman, M. & Sampat, B. 2009. Applicant and examiner citations in U.S. patents: An overview and analysis. Research Policy, 38(2): 415-427.
115
Asmussen, C.G., Pedersen, T. & Dhanaraj, C. 2009. Host-country environment and subsidiary competence: Extending the diamond network model. Journal of International Business Studies, 40: 42-57.
Atuahene-Gima, K. 2005. Resolving the capability-rigidity paradox in new product innovation. Journal of Marketing, 69:61-83.
Ahuja, G. 2000. Collaboration Networks, Structural Holes, and Innovation: A Longitudinal Study. Administrative Science Quarterly, 45(3): 25-455
Bell, G. & Zaheer, A. 2007. Geography, Networks, and Knowledge Flow. Organization Science, 18(6): 955-972
Birkinshaw, J., Nobel, R. and Ridderstrale, J., 2002. Knowledge as a contingency variable: do the characteristics of knowledge predict organization structure?, Organization Science, Vol.13, No.3, pp.274-289.
Brusoni, S., Prencipe A. and Pavitt, K. 2001. Knowledge Specialisation, Organizational Coupling and the Boundaries of the Firm: Why Firms Know More Than They Make?, Administrative Science Quarterly, 46 (4): 597-621.
Cantwell, J.A. 1991. The theory of technological competence and its application to international production, in D. McFetridge (ed.), Foreign Investment, Technology and Economic Growth, University of Calgary Press: Calgary.
Cantwell J.A. 1995. The globalisation of technology: what remains of the product cycle model? Cambridge Journal of Economics, 19(1): 155-174.
Cantwell J. A. and H. B. Andersen. 1996. A statistical analysis of corporate technological leadership historically, Economics of Innovation and New Technology, 4:211-234.
Cantwell J. A. 2009. Location and the multinational enterprise. Journal of International Business Studies, 40:35-41.
Cantwell J.A, & Mudambi R. 2005. MNC competence-creating subsidiary mandates. Strategic Management Journal 26(12): 1109-1128.
Cantwell J.A., Piscitello L. 2000. Accumulating technological competence - its changing impact on corporate diversification and internationalisation, 2000, Industrial and Corporate Change, 9(1): 21-51.
Cantwell J.A., Piscitello L. 2005. The recent location of R&D activities by large MNCs in the European regions: The role of spillovers and externalities, Regional Studies, 39: 1-16.
116
Cantwell J.A., Piscitello L. 2007. Attraction and deterrence in the location of foreign-owned R&D activities: the role of positive and negative spillovers, International Journal of Technological Learning, Innovation and Development, 1: 1-30.
Contractor, F. J. & Lorange, P. 2002. The growth of alliances in the knowledge-based economy. International Business Review. 11(4): 485-502.
Day, G. 1994. The capabilities of market-driven organizations. Journal of Marketing, 58 (October): 37–52.
Dunning, J. 1995. Reappraising the eclectic paradigm in an age of alliance capitalism, Journal of International Business Studies, 26(3): 461–492
Eisenberg, H. 2000. Patent law you can use: continuation patent applications. From website of Yale University Office of Cooperative Research.x
Forsgren, M., Holm, U., & Johanson, J. 2005. Transfer of knowledge in the embedded multinational - the role of shared values and business networks. in Managing the Embedded Multinational: A Business Network View, Northampton: Edward Elgar.
Gittelman, M., 2006, “National institutions, public-private knowledge flows, and innovation performance: a comparative study of the biotechnology industry in the US and France”, Research Policy, 35:1052-1068.
Ghoshal, S., & Bartlett, C. 1988. Creation, adoption, and diffusion of innovations by subsidiaries of multinational corporations. Journal of International business studies, 19(3): 365-388.
Ghoshal, S., & Bartlett, C. 1991. The multinational corporation as an interorganizational network. Academy of Management Review, 15(4): 603–625.
Granovetter, M. 2005. The impact of social structure on economic outcomes. Journal of Economic Perspectives, 19(1): 33-50
Grubel, Herbert G.; Lloyd, Peter J. (1971). "The Empirical Measurement of Intra-Industry Trade". Economic Record 47(4): 494–517.
Grubel, Herbert G.; Lloyd, Peter J. (1975). Intra-industry trade: the theory and measurement of international trade in differentiated products. New York: Wiley.
Gulati,R. 1999. Network location and learning: the influence of network resources and firm capanbilities on alliance formation. Strategic Management Journal, 20:397-420
Gupta, A. and Govindarajan, V., 1991. Knowledge flows and the structure of control within multinational corporations, The Academy of Management Review, 16(4):768-792
117
Gupta, A. and Govindarajan, V., 2000. Knowledge flows within multinational corporations, Strategic Management Journal, 21(4):473-496
He, Z. & Wong, P. 2004. Exploration vs. exploitation: an empirical test of the ambidexterity hypothesis. Organization Science, 15(4): 481-494
Hedge, D. Mowery, D., & Graham, S. 2007. Pioneers, submariners, or thicket-builders: which firm use continuations in patenting? NBER Working Paper No. 13153.
Hedlund, G. 1994. A model of knowledge management and the N-form corporation’, Strategic Management Journal, Summer Special Issue, 15: 73–90.
Hedlund, G., & Ridderstråle, J. 1995. International development projects: Key to competitiveness, impossible, or mismanaged. International Studies of Management and Organization, 25(1–2): 121–157.
Hounshell, D.A. (1992), "Continuity and change in the management of industrial research: the Du Pont Company, 1902-1980", chapter 8 in G. Dosi, R. Giannetti and P.A. Toninelli (eds.), Technology and Enterprise in a Historical Perspective, New York: Oxford University Press.
Inkpen, A.C. 2002. Learning, Knowledge Management, and Strategic Alliances: So Many Studies, So Many Unanswered Questions, in F. J. Contractor and P. Lorange ( eds.) Cooperative Strategies and Alliances, Elsevier Science: Amsterdam, pp: 267-289.
Jaffe, A. 1986. Technological Opportunity and Spillovers of R&D: Evidence from Firms' Patents, Profits and Market Value, American Economic Review, LXXVI (1986): 984-1001.
Jaffe, A., Trajtenberg, M., & Henderson, R. 1993. Geographic Localization of Knowledge Spillovers as Evidenced by Patent Citations, The Quarterly Journal of Economics, 108(3): 577-598.
Kim, L. 1998. Crisis construction and organizational learning: capability building in catching-up at Hyundai Motor. Organization Science, 9: 506-201.
Kogut, B. 1988. Joint ventures: Theoretical and empirical perspectives, Strategic Management Journal, 9(4): 3 19-332.
Kogut, B. 1991. Joint ventures and the option to expand and acquire, Management Science, 37(1):19-33.
Kogut, B. & Chang, S. 1991. Technological Capabilities and Japanese Foreign Direct Investment in the United States. The Review of Economics and Statistics, 73: 400-413.
118
Kogut, B. & Zander, U., 1992, Knowledge of the firm, combinative capabilities, and the replication of technology, Organization Science, 3(3): 383-397.
Kogut, B. & Zander, U., 1993, Knowledge of the firm and the evolutionary theory of the multinational corporation, Journal of International Business Studies, 24(4): 625-645.
Lemley, M. and Moore, K., 2004. Ending abuse of patent continuations. Boston University Law Review, 84.
Lyles, M.A. & Salk, J.E. 1996. Knowledge acquisition from foreign parents in international joint ventures: an empirical examination in the Hungarian context, Journal of International Business Studies. 27(5): 877-903.
Macher, J., 2006. Technological Development and the Boundaries of the Firm: A Knowledge-Based Examination in Semiconductor Manufacturing. Management Science, 52(6): 826-843
March, J. G. 1991. Exploration and exploitation in organizational learning. Organization Science. 2(1): 71–87
Martin, X. and Salomon, R. 2002. Knowledge transfer capacity and cooperative strategies: when should firms leverage tacit knowledge through licensing and alliances? In F. J. Contractor and P. Lorange (eds.) Cooperative Strategies and Alliances, Elsevier Science: Amsterdam, pp: 729-748.
Martin, X. and Salomon, R., 2003. Tacitness, learning, and international expansion: a study of foreign direct investment in a knowledge-intensive industry. Organization Science, 14(3): 297-311.
McEvily, B. & Zaheer, A. 1999. Briging Ties: a source of firm heterogeneity in competitive capabilities. Strategic Management Journal, 20: 1133-1156.
McEvily, B. & Marcus, A. 2005. Embedded Ties and the Acquisition of Competitve capabilities. Strategic Management Journal, 26: 1033-1055.
Mowery, D. C., & Oxley, J. E. 1995. Inward technology transfer and competitiveness: The role of national innovation systems. Cambridge Journal of Economics, 19: 67-93.
Mowery, D. C., Oxley, JE. & Silverman, BS. 1996. Strategic Alliance and Interfirm Knowledge Transfer. Strategic Management Journal, 17(winter special issue): 77-91
119
Mudambi, R. and Navarra, P., 2004. Is knowledge power? Knowledge flows, subsidiary power and rent-seeking within MNCs, Journal of International Business Studies, 35(5): 385-406.
Nobel, R., & Birkinshaw, J. 1998. Innovation in multinational corporations: control and communication patterns in international R&D operations. Strategic Management Journal, 19(5): 479-496.
Simonin, B. L. 2004. An Empirical Investigation of the Process of Knowledge Transfer in International Strategic Alliances. Journal of International Business Studies, 35(5): 407-427.
Pearce, R. 1999. The evolution of technology in multinational enterprises: the role of creative subsidiaries. International Business Review, 8: 125-148.
Phene, A., & Almeida, P. 2008. Innovation in multinational subsidiaries: the role of knowledge assimilation and subsidiary capabilities. Journal of International Business Studies, 39: 901-919.
Porter, M. 1990. The Competitive Advantage of Nations. Free Press, New York.
Prahalad, C., & Doz, Y. 1981. An approach to strategic control in MNCs. Sloan Management Review, 23(Summer): 5–13.
Ronstadt, R. C. 1977. Research and Development abroad by U.S. Multinationals. Praeger, New York.
Rowley, T., Behrens, D. & Krackhardt, D. 2000. Redundant Governance Structures: An analysis of structural and relational embeddedness in the steel and semiconductor industries. Strategic Management Journal, 21: 369-386
Szulanski, G. (1994). Intra-firm transfer of best practice project: Executive summary of the findings (Report) APQC.
Szulanski, G. 2000. The Process of Knowledge Transfer: A Diachronic Analysis of Stickiness. Organizational Behavior and Human Decision Processes. 82(1), May: 9-27
Uzzi, B. 1996. The sources and consequences of embeddedness for the economic performance of organizations: the network effect. American Sociological Review, 61(4): 674-698.
Uzzi, B. 1997. Social structure and competition in interfirm networks: the paradox of embeddedness. Administrative Science Quarterly, 42: 35-67.
120
Teece, D., Pisano, G., & Shuen, A. 1997. Dynamic capabilities and strategic management. Strategic Management Journal, 18(7): 509-533.
Tilton, J. E. 1971. International diffusion of technology: the case of semiconductors. Washington DC: Brookings Institution Press.
Uzzi, B. 1996. The sources and consequences of embeddedness for the economic performance of organizations: the network effect. American Sociological Review, 61(4): 674-698.
Uzzi, B. 1997. Social structure and competition in interfirm networks: the paradox of embeddedness. Administrative Science Quarterly, 42: 35-67.
Vernon, R. 1966. International investment and international trade in the product cycle. Quarterly Journal of Economics, 81: 190-207.
Williamson, O., 1985, The Economic Institutions of Capitalism, Chapter 4, pp.85-102.
Winter, S., 1987, “Knowledge and competence as strategic assets”, D. Teece, ed. The Competitive Challenges – Strategic for Industrial Innovation and Renewal. Ballinger, Cambridge, MA, pp.159-184.
Zander, I. 2002. The formation of international innovation networks in the multinational corporation: an evolutionary perspective. Industrial and Corporate Change, 11(2): 327-353.
Zander, I., & Solvell, O. 2000. Cross-border innovation in the multinational corporation. International Studies of Management and Organization, 30(2): 44-68.
121
CURRICULUM VITAE
EMPLOYMENT
Assistant Professor at SUNY Oswego 2012.8– Present Department of Marketing and Management, School of Business
EDUCATION
Rutgers Business School, PhD in Management: International Business 2007.9 – 2013.5
School of Management, Zhejiang University, Doctoral program: Management 2004.9 – 2007.6 Chu Kechen Honors College, Zhejiang University, Bachelor: Computer Science 2000.9 – 2004.6
DISSERTATION
MNC subunit knowledge sourcing and competence creating activities – a dynamic view of subunit evolution
Dissertation Committee: John Cantwell (chair), Farok Contractor, Michelle Gittelman, Lucia Piscitello
Dissertation Abstract:
This dissertation suggests a framework to examine the intra- and inter- firm knowledge sourcing behaviors of an MNC subunit, and how this influences the extent to which the subunit is being competence creative (CC) rather than competence exploitative (CE), and the study looks at how in turn the composition of CC and CE activities in a location influences the knowledge creating role and status of a subunit within its international group. I developed a dynamic model which proposes that the extent to which a subunit is likely to take up CC activities is influenced by the intensity and spread of that subunit’s connectedness of their processes of knowledge accumulation with various sources - whether in other parts of the focal MNC group or external to the firm. The focus of attention is placed on the heterogeneity of firm-specific evolutionary paths in the patterns of knowledge accumulation that support CC activities, controlling for the industry-specific determinants of such technological trajectories.
122
AWARDS
• Nominated for Best Dissertation Proposal Award in AIB 2011, Nagoya • Dissertation Fellowship 2011-2012 by Rutgers Graduate School • Technology Management Research Center fund for Summer Research Assistants,
2011 • Dean's Fund for Summer Ph.D. Research Assistants, 2010 • Competitive summer research scholarship, 2008
RESEARCH INTERESTS
• Intra- and Inter- firm knowledge networks of MNC • MNC subsidiary strategy, knowledge flow, subsidiary evolution, organizational
culture • Technological innovation: exploration versus exploitation, technology diffusion,
learning
PUBLICATIONS
Slone, R., Becker, S., Penton, P., Pu, X., McNamee, R., (Nov-Dec 2011). Managing Global R&D Networks. Research Technology Management, 54( 6 ), 59-61
Chen, J., Pu, X., & Shen, H. 2010. A Comprehensive Model of Technological Learning: Empirical Research on Chinese Manufacturing Sector. In The Rise of Technological Power in the South, Edt. Xiaolan Fu and Luc Soete. Palgrave Macmillan 2010.
WORKING PAPERS
Pu X and Cantwell J, Dual network patterns of MNC subunit knowledge sourcing – the legend, evolution, and new trend
Cantwell J, Piscitello L and Pu X, MNC subunit evolution – the role of technological relatedness and location openness
McNamee R and Pu X, Insights into MNC’s global knowledge networks: strategies and implementations (Research on Research project of IRI)
123
Huang S, McNamee R, Piepenbrink A and Pu X, Marshal vs. Jacobs: an intervention via multi-dimensional, dynamic characterization of clusters
CONFERENCE PRESENTATIONS
Pu X, (AIB 2011, Nagoya) MNC subunit knowledge sourcing and CC activities - a dynamic view of subunit evolution
Huang S, McNamee R and Pu X (AIB 2011, Nagoya) Marshall vs. Jacobs: an intervention via multi-dimensional, dynamic characterization of clusters
Pu X, (AIB 2010, Rio de Janeiro) Organization-culture fit: and insight into MNC subsidiary’s culture dimensions
Pu X and Qiu R, (AOM 2009, Chicago) The impact of network involvement on absorptive capacity: a dynamic model of subsidiary role evolution
Pu X, (CICALICS Workshop 2006, Chengdu) Research on the Mechanisms of Technology Diffusion in CoPS Innovation – A Review of Current Studies
Pu X, (GLOBELICS Academy 2006, Lisbon) Research on the Mechanisms of Technology Diffusion in CoPS Innovation: An IPR Protection Perspective.
TEACHING
• International Business (Summer 2009, Fall 2009, Spring 2010, Winter 2011, Summer 2011, Fall 2012)
• Principle of Management (Summer 2011, Fall 2012, Spring 2013) • Strategic Management (Spring 2013)
PROFESSIONAL SERVICES
• Reviewer for AOM (2007, 2008, 2009, 2011), AIB (2009, 2010, 2011, 2013) • Co-organizer of departmental seminar series 2009