LISA"

:5

5 v.
4.3. A... s31. .

, . .9.
If»? he a

.
..

.F .
1.30.81 5W n
.. .n... #4... £9.
u
.duvyaJ lit." :5: 1
his...

.3.

‘3-
.1

'1;

.. .2 .5: 3.9.... .......§: 5. .. .. . , . 3...

«it...

 

 

’1'? Michigan state
2 03.6, _ Unwersny

This is to certify that the
dissertation entitled

FINDING THE BALANCE BETWEEN OUTSOURCING AND
INTERNALIZATION: THE KEY TO INNOVATIVE SUCCESS?

presented by

MICHAEL ANDREW STANKO

has been accepted towards fulfillment
of the requirements for the

PhD. degree in Marketing

 

 

“/(Zju962‘iflM/LEQH/LL/

U Ma’jor Professor’s Signature

 

May 16, 2008

 

Date

MSU is an affinnative-action, equal—opportunity employer

- -l---.-n--.-.-.-.-->—c—-.—--

—.-¢.-.-.—.—.— -.-- -

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 K:IProj/Acc&Pres/ClRC/DateDue indd

 

FINDING THE BALANCE BETWEEN OUTSOURCING AND INTERNALIZATION:
THE KEY TO INNOVATIVE SUCCESS?

By

Michael Andrew Stanko

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Marketing

2008

ABSTRACT

FINDING THE BALANCE BETWEEN OUTSOURCING AND INTERNALIZATION:
THE KEY TO INNOVATIVE SUCCESS?

By
Michael Andrew Stanko

This dissertation considers two important questions pertaining to the outsourcing
of innovation. First, how is the innovation outsourcing decision different from more
traditional sourcing decisions? Since Williamson (1991), there has been speculation that
the innovation outsourcing decision is distinct from other governance decisions. This
research examines learning as well as appropriability as determinants of the innovation
outsourcing decision; both of these additional determinants are found to significantly
impact the governance decision.

The second question is: What are the performance implications of outsourcing
innovation activities? The outcomes of outsourcing innovation are not yet well
understood. Here, it is found that firms governing appropriately (i.e., according to
transaction cost predictions) are more innovative than firms governing inappropriately.
However, appropriately governing firms also face higher total innovation costs, though
the increase in costs is proportionately smaller than is the increase in innovativeness.
Further, it is shown that firms which over-contract (i.e., outsource too great a portion of
their innovation activities) face higher total innovation costs than firms which under-

contract, but firms which over-contract tend to be more profitable.

ACKNOWLEDGMENTS

First, I am grateful to my committee for all their efforts during my time in the
doctoral program. Professors Roger Calantone, Cornelia Droge, David Griffith and
Jonathan Bohlmann have been of tremendous assistance both to the completion of this
project as well as to my own development as an emerging scholar.

Thanks also goes to my family, for supporting me through what must have
seemed like an eternity of post-secondary education.

Finally, I would like to thank my fellow students for their friendship, assistance,
and camaraderie during my time at Michigan State. Though there are too many dear
colleagues to list here, if you have also enjoyed too many Owen Cajun Chicken
Sandwiches to count, commiserated over late night econometrics problem sets, shared a

pint at the Peanut Barrel or cheered on the Spartans with me — thanks.

iii

PREFACE

The research in this dissertation was conducted while the author was a Special
Sworn Status researcher of the US. Census Bureau at the Michigan Census Research
Data Center (MCRDC). Research results and conclusions expressed are those of the
author and do not necessarily reflect the views of the Census Bureau. This paper has
been screened to insure that no confidential data are revealed.

This project was supported by Michigan State University’s Vice President for
Research and Graduate Studies through the Health, Institutions and Policy Initiative, and
by the Eli Broad Chair Fund of Roger J. Calantone. Support for this research at the
MCRDC from the National Science Foundation (NSF) (awards no. SES-OOO4322 and
ITR-0427889) is gratefully acknowledged.

Thank you to Clint Carter, Maggie Levenstein, and Stan Sedo of the MCRDC,
Ray Wolfe of the NSF, as well as Arnold Reznek and Lucia Foster of the US. Census

Bureau. Their help with the data involved in this dissertation was tremendously valuable.

iv

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. vi
CHAPTER 1: INTRODUCTION ...................................................................................... 1
CHAPTER 2: LITERATURE REVIEW AND HYPOTHESES ....................................... 9
Transaction Cost Analysis .............................................................................................. 9
Innovation Governance ................................................................................................. ll
Antecedents of External Innovation Governance ......................................................... 12
Non Transaction Cost Determinants of Innovation Governance .................................. 19
Appropriate Governance and Performance ................................................................... 25
Hypotheses .................................................................................................................... 30
CHAPTER 3: MEASURES AND ANALYSIS OVERVIEW ........................................ 42
Data Sources ................................................................................................................. 42
Construct Operationalizations ....................................................................................... 43
Analysis Overview ........................................................................................................ 52
CHAPTER 4: DATA PREPARATION, ANALYSIS AND RESULTS ......................... 56
Preparing Stage One Data ............................................................................................. 56
Preparing Stage Two Data ............................................................................................ 58
Stage One Analyses ...................................................................................................... 62
Stage One Results ......................................................................................................... 64
Stage Two Analyses ...................................................................................................... 66
CHAPTER 5: DISCUSSION, MANAGERIAL IMPLICATIONS, AND FURTHER
RESEARCH DIRECTIONS ............................................................................................. 74
Discussion ..................................................................................................................... 74
Determinants of Governance ........................................................................................ 74
Performance Implications ............................................................................................. 78
Managerial Implications ............................................................................................... 82
Conclusions ................................................................................................................... 84
Limitations .................................................................................................................... 86
Future Research Directions ........................................................................................... 87
Appendix A: Empirical Innovation Governance Research .......................................... 91
Appendix B: Case Study and Managerial Focused Innovation Governance Research
.................................................................................................................................... 100
Appendix C: Empirical Governance Appropriateness Research ............................... 105
Appendix D: SAS Proc NLMixed Code For Stage One Estimation ......................... 108
REFERENCES ............................................................................................................... 109

Table 1:

Table 2:

Table 3:

Table 4:

Table 5 :

Table 6:

Table 7:

Table 8:

Table 9:

Table 10:

Table 11:

Table 12:

Table 13:

Table 14:

Table 15:

LIST OF TABLES

Summary of Contributions .................................................................................. 8
Antecedents of Innovation Governance: Research Consensus .......................... 22
Hypothesized Temporal Outcomes .................................................................... 41
Constructs and Operationalizations ................................................................... 51
Outcome Measures Over Time .......................................................................... 53
Industry Classifications ..................................................................................... 58
First Stage Correlations and Descriptive Statistics ............................................ 60
Second Stage Correlations and Descriptive Statistics ....................................... 61
Stage One - Results of Mixed Models ............................................................... 63
Stage One - Tobit Results ................................................................................ 65
Stage Two - Innovativeness Results ................................................................ 68
Stage Two - Innovation Costs Results ............................................................. 70
Stage Two - Profit Results ............................................................................... 72
Summary of Stage Two Findings .................................................................... 73
Performance Implications ................................................................................ 84

vi

FINDING THE BALANCE BETWEEN OUTSOURCING AND
INTERNALIZATION: THE KEY TO INNOVATIVE SUCCESS?

CHAPTER 1: INTRODUCTION

Innovativeness is central to competitive success. Innovative new products are key
to revenue growth (Sorescu et al., 2003) and future market leadership (Chandy & Tellis,
1998). However, innovation efforts are often unsuccessful and can require tremendous
investments (Chandy et al., 2006). Given this, it is not surprising that innovation has
been a fundamental issue of late in marketing (Carson, 2007; Hauser et al., 2006; Luca &
Atuahene-Gima, 2007; Toubia, 2006). Despite researchers’ interest in innovation, the
problem of managing the firm’s innovation boundaries has not yet been well addressed.
This problem is increasingly relevant because many firms today are supplanting
internally pursued innovation activities with outsourcing arrangements. Innovative firms
in industries as diverse as software, automotive and aerospace, as well as less high tech
industries such as consumer packaged goods, are choosing to outsource innovation efforts
(Carson, 2007). While firms such as Procter & Gamble are making strategic efforts to
develop new products outside their organizational boundaries, other firms have taken
steps to protect and even promote their internal deve10pment competencies; for instance,
upon each iPod is printed “Designed by Apple in California” (Engardio & Einhom,
2005)

Which is the path to innovative success: internal pursuit of innovation activities or
contracting these activities to other firms? Consistent with Transaction Cost Analysis
(TCA), this study proposes that external innovation contracting can improve innovative

performance, but only if it is appropriate to the conditions in which the firm Operates.

TCA has been a major paradigm in marketing, management and other disciplines
in past decades (Rindfleisch & Heide, 1997) and has been the theoretical basis for several
investigations into innovation-related sourcing and alliances (e. g., Pisano, 1990;
Robertson & Gatignon, 1998; Sampson, 2004; Ulset, 1996). TCA has also been used to
investigate governance choice in a wide variety of other applications ranging from
international entry mode (Zhao et al., 2004) to logistics outsourcing (Bienstock &
Mentzer, 1999) to the use of a dedicated salesforce vs. sales agencies (Anderson, 1988).
There are also applications outside of typical business relationships such as Hamilton’s
(1999) investigation of prenuptial contracts.

Why is investigating innovation governance distinct from investigating many of
these other applications? Innovation is “an iterative process initiated by the perception of
a new market and/or service opportunity for a technology based invention which leads to
development, production and marketing tasks striving for the commercial success of the
invention” (Garcia & Calantone, 2002 p. 112). As discussed by Williamson (1991, p.
292-293), peculiarities in innovation-related governance require an extension of TCA to
capture the complications involved in this type of governance decision. Innovation-
related governance is distinct from other types of organizational governance in two
particular respects. First, the fear of appropriation (losing intellectual property to partner
firms) is heightened, as most firms believe innovation to be central to their core
competencies (Coombs, 1996) and future competitiveness (Howells et al., 2003).
Second, in certain environments, critical learning is done while performing innovation
activities (learning by doing). Since innovation is an iterative process (Garcia &

Calantone, 2002), an organization’s knowledge continues to build upon itself. When an

organization lacks knowledge in a particular domain, significant time and effort may be
needed to develop the needed capabilities internally.

According to Williamson (1991), the first of these peculiarities involves the role
knowledge appropriation (leakage) plays in this decision. F inns justifiably fear that by
sharing their knowledge with partner firms, they may lose intellectual property or create
competitors (Narula, 2001; Veugelers & Cassiman, 1999). The role of knowledge
appropriation has been investigated by several researchers (Cesaroni, 2004; Teece, 1986;
Veugelers & Cassiman, 1999) and the threat of knowledge leakage has been shown to
deter the contracting of innovation activities.

Williamson also notes that when making the innovation governance decision,
market governance (i.e., outsourcing) may be selected to facilitate timely development of
new products. There are two situations in which firms are likely to pursue outsourcing
for this reason: First, “when events are fast moving” and second, “if learning-by-doing is
essential” (p. 292). TCA-based research has considered technological uncertainty (e. g.,
Balakrishnan & Wemerfelt, 1986; Robertson & Gatignon, 1998; Walker & Weber,
1984), which is closely related to Williamson’s first timeliness condition. In a
technologically uncertain environment, specifications, desired outcomes, as well as
requirements for both expertise and equipment are constantly in flux (Howells, 1999).
However, the importance of learning has thus far been ignored in the TCA-based
innovation governance literature and is the second peculiarity of the innovation
governance decision that is of the interest here. The question remains unanswered: How
does the necessity of learning by doing impact the innovation governance decision?

Asset specificity has been shown to lead to internal governance (Nakamura & Odagiri,

2005; Ulset, 1996). Here, it is argued that this learning-timeliness effect moderates the
relationship between asset specificity and governance mode (see Chapter two for detailed
hypotheses). For instance, in an environment where learning is critical, a firm whose
innovation activities do not require specific assets will be extremely likely to externally
contract to speed innovations to market. In this case, the time and costs associated with
gaining necessary learning while developing specific assets will strongly deter an internal
governance decision.

This study will examine whether the expanded set of transaction cost arguments
suggested by Williamson (1991) —- including both the threat of knowledge leakage
(which, as mentioned previously, has been incorporated in other TCA-based innovation
governance studies) and the importance of learning by doing - better explains firms’
innovation governance decisions than does the traditional set of transaction cost
arguments. It is not sufficient, however, to show that these additional constructs have an
impact on firms’ governance patterns. TCA not only makes predictions for firms’
selection of governance, but also projects performance implications based on the
governance choice. Transaction cost logic argues that when the chosen governance
mechanism is best suited to the level of transaction costs, performance will be optimal
(Williamson, 1985, p. 22-23). As discussed by Geyskens et a1. (2006), Rindfleisch and
Heide (1997), and Sampson (2004), too often TCA-based studies have been restricted to
explaining governance form, without examining performance implications.

Once the antecedents’ impacts on innovation governance have been examined, it
is then possible to determine whether firms that have the appropriate mix of internal vs.

contracted. innovation activities (according to TCA predictions) outperform firms that

over-rely or under-rely on external innovation contracting. Previous innovation research
has shown that external governance has both tremendous possible benefits as well as
drawbacks. Though external innovation governance has been shown to aid
innovativeness (Nicholls-Nixon & Woo, 2003), it also tends to increase innovation costs,
not only through the direct spending on contracting, but also by increasing internal
innovation spending (Veugelers, 1997). In order to examine the impact of external
innovation governance on various performance dimensions, it is necessary to analyze
performance as a function of appropriate governance given the environmental
characteristics. TCA logic holds that there is no one appropriate governance mode:
conditions dictate appropriate governance (Sampson, 2004; Williamson, 1985). In this
study, in order to capture several dimensions of innovative performance, three distinct
outcome variables are examined: innovativeness, total innovation costs, and profitability.
Typically, innovation-related projects are judged by a variety of metrics, making a
diversity of performance measures imperative to deriving meaningful implications.
Firms that pursue innovation activities via the predicted mix of internalization and
contracting (based on the transaction cost arguments put forward herein) are referred to
as using “appropriate” governance]. Those firms which do not pursue innovation
activities via the TCA-predicted mix of internalization and contracting are referred to as
using “inappropriate” governance. Inappropriately governing firms can be divided into
two categories: those firms which over-rely on external contracting (i.e., overly market
oriented governance) and those firms which under-rely on external contracting (i.e.,
overly hierarchical governance). If the normative aspect of TCA holds, appropriately

governing firms will outperform both groups of inappropriately governing firms (all else

 

1 Throughout this study, the terms “contracting” and “outsourcing” are used interchangeably.

equal). Further, a comparison of the performance differences across the two different
groups of inappropriately governing firms (those which over-contract externally and
those which under-contract externally) will prove interesting to both TCA and innovation
scholars as well as managers of the innovation process. Previous researchers have
speculated that an over-reliance on external contracting can erode a firm’s technological
competence over time (Cavusgil et al., 2003; Piachaud, 2005). Equally interesting are the
potential repercussions of under-contracting. Researchers have speculated that this will
deny firms access to outside expertise and efficiencies (Howells, 1999), thereby
preventing firms from concentrating on their core competencies and increasing their
overall costs. Here, it is hypothesized that under-contracting is more harmful in the short
term than is over-contracting, since under-contracting firms face increased short term
costs and lack of flexibility. Conversely, it is hypothesized that over-contracting is more
harmful in the long term than is under-contracting since over-contracting can lead to a
lessening of a firm’s absorptive capacity and eventually the erosion of its core
competency.

Two recent research findings are central to the discussion of the relative impact of
the two varieties of inappropriate governance. In a TCA-based study of R&D alliances,
Sampson (2004) found that firms that rely on too hierarchical (i.e., too close to internal
governance) an alliance structure are dramatically less innovative than properly aligned
alliances. However, firms using a less hierarchical structure (i.e., too close to market
governance) than appropriate were only slightly less innovative than properly govemin g
firms. This suggests that the constraints placed on innovation by inappropriate

hierarchical governance are more damaging to performance than the costs imposed by

inappropriately selecting a market oriented governance mode. Leiblein et al.’s (2002)
findings also suggest performance asymmetry between the two categories of
inappropriately governing firms, but their findings suggest that in production outsourcing
decisions, firms selecting a less hierarchical than optimal governance structure more
severely under—perform than do firms selecting too hierarchical a governance structure.
In opposition to Sampson’s findings, this suggests that failing to safeguard assets by
inappropriately selecting a market oriented governance option more dramatically harms
performance than does selecting too hierarchical a governance option.

In general, the management of cross-boundary innovation is of tremendous
interest to practitioners and greater scholarly research is needed in this area (Hauser et al.,
2006 p. 694—695). More specifically, this research project contributes to the innovation
governance and (mis)govemance literatures by addressing three distinct and important
problems (summarized in Table 1) in addition to several other possible contributions
discussed in the next chapter. First, as discussed previously, the role of learning by doing
in innovation governance requires examination (Williamson 1991, p. 292). Does the
need to learn internally in innovation-related projects determine firm’s innovation
governance decisions?

Second, at present researchers do not understand the relative implications of using
governance structures that either impose excessive bureaucracy or do not provide
adequate safeguards from market hazards. Are firms better off over-relying on extemal
contracting or under-relying on external contracting? The possibly asymmetric

performance implications of the two forms of inappropriate governance compel further

investigation, and the use of multiple outcome dimensions in this examination will likely

add further insights (as called for by Leiblein et a1. 2002, p. 830-831).

Finally, TCA-based researchers have also called for longitudinal testing of the

relationship between (in)appropriate governance and performance (Macher & Richman,

2006, p. 63). In an iterative process such as innovation, inappropriate governance likely

has long term implications, making examination of longitudinal effects on performance

critical (N icholls-Nixon & Woo, 2003, p. 665). Further, by considering multiple

outcome dimensions over time, other insights may surface. For instance, it is plausible

that over- contracting may have severe long term innovativeness implications, but only

short term cost repercussions. Findings such as these could prove extremely valuable to

both scholars and managers.

Table 1: Summary of Contributions

 

 

 

 

 

relationship between (in)appropriate
governance and performance

 

 

Woo (2003, p. 665),
Macher and Richman,
(2006,p.63)

 

Contribution Called for by: Hypotheses
Investigating the management of Hauser, Tellis and All
cross-boundary innovation. Griffin (2006 p. 694-
695)
Investigating role of the importance Williamson (1991, p. H2
of learning by doing in innovation 292)
governance.
Examining implications of Leiblein et a1. (2002, H7, H8 and H9
(in)appropriate governance on p. 830-831)
multiple outcome dimensions.
Longitudinal testing of the Nicholls-Nixon and H7, H8 and H9

 

 

CHAPTER 2: LITERATURE REVIEW AND HYPOTHESES

Transaction Cost Analysis

 

As stated by Leiblein (2003 p. 940), “The basic idea [of TCA] is to match simple
exchanges with simple modes of governance and more complex exchanges with more
complex forms of organization.” TCA views firms and markets as alternative
governance structures. Firms will favor internal organization in environments where
transaction costs are prohibitive, and economic factors will favor the market (i.e.,
outsourcing) when transaction costs are low. Under certain conditions, the costs of
performing specific activities through the market may exceed the costs of organizing the
activities within the firm, in which case firms will internalize the function. On the other
hand, where the costs of performing activities through the market is minimal, firms will
favor the use of market governance (Williamson, 1975, 1985).

Transaction costs can be divided into three categories: adaptation, safeguarding
and measurement (Macher & Richman, 2006; Rindfleisch & Heide, 1997). Adaptation
costs represent efforts to adjust contracts to changing conditions and are a result of
environmental uncertainty. That is, when a firm may have to revise an agreement with a
partner company (thus facing substantial penalties) due to an unstable market
environment, the firm is likely to perform this function internally (Masten et al., 1991;
Ulset, 1996). Safeguarding costs characterize the costs of a partner acting
opportunistically after investments in the relationship have been made and are the result
of transaction specific investments (John & Weitz, 1988; Pisano, 1990). Measurement
costs represent the expenses associated with confirming that contracts have been

completed as specified and are a result of behavioral uncertainty. If it is very difficult to

determine the quality of a contracted good or service, the contracting firm may face
substantial costs to estimate quality. When these transaction costs are substantial,
internalization will be favored (Anderson, 1985).

When transaction costs are high, internalization is preferred due to three specific
characteristics of the firm (Rindfleisch & Heide, 1997; Williamson, 1985). These
characteristics of internal organization allow the firm to detect and control opportunism
as well as foster adaptation under conditions characterized by high transaction costs.
First, under internal organization, firms can measure both output as well as behavior,
where typically it is only possible to measure output from contracted firms. Second,
internal organization allows for longer term rewards (e.g., promotions) than contracting
typically does. Finally, the socialization process in organizations can help foster goal
congruence. These mechanisms help to control opportunism and encourage adaptation
within an organization.

TCA makes three assumptions (Macher & Richman, 2006; Rindfleisch & Heide,
1997). Bounded rationality presumes that decision makers’ cognitive abilities are
constrained (Williamson, 1996). Managers are constrained in decision making; they
cannot consider every possible course of action, nor can they analyze every possible
piece of data. Opportunism supposes that an exchange partner may act unscrupulously to
serve their own interests, and that it is difficult to establish the trustworthiness of
potential partners (John, 1984; Wathne & Heide, 2000). Risk neutrality is the assumption
that firms are neither risk averse nor risk seeking (Williamson, 1975, 1985). All three of

these assumptions hold within the innovation governance context.

10

Innovation Govemance

 

There exists a substantial literature focused on the innovation governance decision
(i.e., whether to outsource or internalize research and development (R&D) or new
product development (NPD) efforts). There also exist noteworthy research streams
pertaining to the structuring of innovation alliances (c.f. Robertson & Gatignon, 1998),
learning from customers or lead users (c.f. von Hippel, 1986, 2005) as well as using
venture capital and acquisition as alternatives or supplements to traditional internal
innovation efforts (c.f. Dushnitsky & Lenox, 2005). Despite the usefulness of this
research in understanding the full range of options available in organizing collaborative
innovation efforts, the discussion in this dissertation is largely restricted to research
pertaining to internally conducted (hierarchically governed) vs. externally contracted
(market governed) innovation. Research focusing exclusively on hybrid governance
mechanisms (i.e., alliances) is not the focus here.

As shown in Appendices A and B, there has been a good deal of research
examining innovation governance over the past 20 years. Searches of the top journals in
the Marketing, Management, Economics, and Innovation disciplines uncovered 19
empirical studies (Appendix A), as well as 11 case studies and managerially focused
articles (Appendix B). These articles are summarized in the respective appendices and
will now be synthesized and reviewed. First, the various antecedents of external

innovation governance will be reviewed, followed by discussion of the consequences of

external innovation governance.

ll

Antecedents of External Innovation Governance

Often, innovation governance researchers have used transaction cost arguments to
support one or several hypotheses, without providing a more complete test of TCA (e. g.,
Audretsch et al., 1996; Croisier, 1998). Consequently, some TCA arguments have
received much more attention than others in the innovation governance domain. First,
findings pertaining to each of these TCA tenants as drivers of innovation governance will
be discussed. After this, findings related to non-TCA-based drivers of innovation
governance will be reviewed. The research consensus (or lack thereof) for each
antecedent of innovation governance is presented in Table 2.

Asset Specificity refers to the transferability of assets which support a given
transaction (Williamson, 198 5). T ransaction-Specific Investments are investments that
are specialized to one or a few users or uses (Anderson & Gatignon, 1986). Succinctly
stated, where the level of transaction specific investments is high, safeguarding costs will
be prohibitive, causing firms to pursue hierarchical governance (internalization).
Williamson (1996) identifies six types of transaction specific investments: (1) site, (2)
physical asset, (3) human asset, (4) dedicated assets, (5) brand name capital, and (6)
temporal (for discussion of each of these categories and their treatment by researchers see
Lohtia et al., 1994). Within the totality of the TCA research stream, asset specificity’s
role in encouraging internal governance has been well established (David & Han, 2004;
Geyskens et al., 2006).

Examples of innovation-related transaction-specific investments include

laboratory equipment which is not easily transferable to other applications or

12

accumulated knowledge which is specific to one application (Lohtia et al., 1994). If
significant specific investments are made into a contracted research project, the
contracted firm may request additional funds before project completion since the market
no longer acts as an adequate safeguard. Firms select hierarchical governance when these
costs are expected to be prohibitive in order to avoid this type of bargaining situation
(Ulset, 1996).

Innovation governance research findings have been consistent and unambiguous
in supporting the relationship between asset specificity and internal innovation
governance. When specific assets are required, firms tend to internalize innovation
efforts due to prohibitive safeguarding costs. There have been a wide variety of
approaches used to measure asset specificity in studies related to innovation governance,
including sunk costs (Ulset, 1996), the use of proprietary technologies (Cesaroni, 2004),
skilled labor (Audretsch et al., 1996), as well as several measures of firm diversification
(Nakamura & Odagiri, 2005; Pisano, 1990). Presumably firms that are less diversified
have made more specific investments into the technology area. Despite the variety of
measures used within the five identified empirical studies examining the relationship
between asset specificity and innovation governance, all five studies made similar
conclusions: the presence of asset specificity is related to internal innovation efforts.
Accordingly, there appears to be consensus that asset specificity is related to internal
innovation governance.

Environmental uncertainty refers to “unanticipated changes in circumstances
surrounding an exchange” (Noordewier et al., 1990 p. 82). Consistent with numerous

researchers (e. g., Heide & John, 1990; Robertson & Gatignon, 1998) since Walker and

13

Weber (1984), environmental uncertainty is conceptualized here as having two distinct
dimensions: market uncertainty and technological uncertainty.

Market uncertainty is the unpredictability of demand in an industry (adapted
from Robertson & Gatignon, 1998). Transaction cost logic dictates that constant changes
in market demand will lead to difficulties in contracting, as these changes will necessitate
renegotiation, together with the accompanying costs of renegotiation. This relationship
has been well established in TCA research (Geyskens et al., 2006). With respect to
innovation projects, market uncertainty may cause fiequent changes to the development
portfolio, complicating and adding expense to external contracting. These changes may
necessitate renegotiation or cancellation of innovation contracts, which will likely carry
prohibitive penalties (a form of transaction cost). These transaction costs promote
internalization under high levels of market uncertainty.

Unfortunately, in the innovation governance literature, there have been relatively
few empirical findings related to market uncertainty as an antecedent to the governance
decision, with many researchers focusing solely on some measure of technological
uncertainty (as will be discussed next). Only two identified empirical studies have
examined market uncertainty’s impact on the innovation governance decision. These two
studies (Audretsch et al., 1996; Love & Roper, 2005) each examines the impact that sales
growth (one potential indicator of market uncertainty) has on the selection of internal or
external R&D governance. However, both studies show nonsignificant results for the
sales growth variable. Interestingly, Carson (2007) finds that market uncertainty has a
negative impact on two types of control that contracting firms use (but does not examine

market uncertainty’s impact on the governance decision). This indicates that contracting

14

firms believe that tumultuous environments require a greater degree of autonomy for
contracted firms. The lack of examination of the role of market uncertainty in innovation
governance decisions marks a firrther gap in the literature to be filled by the present
study.

Technological uncertainty refers to the firm’s inability to accurately forecast
technical requirements (Walker & Weber, 1984). As opposed to market uncertainty,
which favors internal governance to facilitate adaptation, the dominant view in modern
TCA research is that technological uncertainty is managed best through market
governance (Geyskens et al., 2006; Macher & Richman, 2006). In technologically
uncertain environments, selecting market governance allows firms the flexibility to end
relationships should technical requirements shift (Balakrishnan & Wemerfelt, 1986).
This is preferable since contracting with a new partner is generally a faster and less
involved process than developing a new technical competency internally. In the wider
TCA research stream, this relationship between market governance and technological
uncertainty has been well supported (Geyskens et al., 2006). However, this relationship
has not been clearly established in the smaller innovation governance literature: various
studies have shown that technological uncertainty is related to both hierarchical and
market governance. As will be discussed, three identified empirical studies (Calantone &
Stanko, 2007; Swan & Allred, 2003; Ulset, 1996) find that constructs associated with
technological uncertainty are related to external innovation governance, while two other
empirical studies (Croisier, 1998; Nakamura & Odagiri, 2005) imply that technological

uncertainty is related to internal innovation governance. It appears that in the innovation

15

governance context, there may be two conflicting effects imposed by technological
uncertainty.

First, technological uncertainty may promote external governance through the
flexibility afforded by market contracting. Several research findings indicate that
technological uncertainty leads to externally governed innovation. Both firm level (Swan
& Allred, 2003; Ulset, 1996) and industry level studies (Calantone & Stanko, 2007) have
shown that technological uncertainty is positively related to external innovation
governance. These findings are in line with the dominant view in TCA research, that
technological uncertainty favors external governance (Geyskens et al., 2006).

Conversely, technological uncertainty may also promote internal governance.
The process of contracting is complicated by technological uncertainty. That is, where
technical requirements cannot be foreseen, it is difficult to describe project requirements
contractually, which raises the transaction costs associated with external contracting and
leads to internal governance. There exist both case and empirically based studies which
support this relationship between technological uncertainty and internalization. For
instance, Narula (2001) finds through interviews with innovation managers that one of
the most common explanations for not relying more heavily on contracted innovation
activities is the difficulty in negotiating and enforcing contracts caused by the
technological uncertainty inherent to innovation. In terms of empirical support,
Nakamura and Odagiri (2005) find that when the pace of innovation is fast, firms will
tend not to externally contract R&D. Further, Croisier’s (1998) study includes several
distinct findings suggesting that low levels of technological uncertainty encourage

external (market) governance. First, Croisier finds that an R&D project being at an

16

advanced stage (thus having lower technological uncertainty) is positively related to
external R&D sourcing. Further, a low number of technologies used and a short project
duration (both indicators of low technological uncertainty) are also associated with
contracted R&D.

To sum, there have been a diversity of findings with respect to the innovation
governance implications of technological uncertainty. In this application, it is unclear
whether technological uncertainty raises the transaction costs associated with external
contracting by complicating the specification of desired technical details or lowers
transaction costs by allowing contracting firms the flexibility to change vendors (or both).
At present, the lack of consensus here calls for further examination of the relationship
between technological uncertainty and innovation governance.

Behavioral uncertainty is defined as “the degree of difficulty in verifying
whether compliance with established agreements has occurred” (Geyskens et al., 2006 p.
525). Transaction cost logic dictates that when behavioral uncertainty imposes
substantial performance evaluation costs, internal governance will be favored. This
relationship has been supported in the TCA literature as a whole (Geyskens et al., 2006).
In the innovation governance domain, behavioral uncertainty would impose significant
transaction costs when, for instance, the costs of ascertaining the quality of a
development project performed by a contracted supplier are prohibitive. In this case,
internalization will be preferred since the costs of controlling and monitoring quality
internally will be less than the costs of measuring the finished output from the supplier.

In the innovation governance literature, there have been relatively few findings

related to behavioral uncertainty. Apparently, researchers in this area have chosen to use

17

either a measure of technological uncertainty, behavioral uncertainty or appropriability
(which will be discussed next), but few studies have used more than one of these, despite
these being distinct constructs which have shown differing effects on governance in other
TCA research (Geyskens et al., 2006).

There are two identified empirical studies relevant to behavioral uncertainty’s
impact on the innovation governance decision. Steensma and Corley (2001) find that the
perceived threat of partner opportunism, which is closely related to behavioral
uncertainty, makes an internal governance decision more likely. The other study (Pisano,
1990), presents a nonsignificant finding for a variable related to behavioral uncertainty.
Additional support for the relationship between behavioral uncertainty and internal
governance can be found in the innovation alliance literature, where Robertson and
Gatignon (1998) find that firms are more willing to enter into R&D alliances when
innovative performance is easily measured. In summary, there have been relatively few
examinations of behavioral uncertainty’s impact on R&D governance, but these studies
suggest that, in line with traditional TCA logic, firms will be more likely to internalize
innovation functions when potential evaluation costs are prohibitive.

Appropriability is the extent to which results from innovative activities can be
protected and are not easily diffused within an industry (adapted from Veugelers &
Cassiman, 1999). The threat of losing intellectual property by involving partner
companies in innovation efforts raises the perceived transaction costs of contracting these
efforts and motivates firms to internalize innovation efforts under these circumstances.

Weak appropriability (high threat of knowledge leakage) has consistently been

empirically shown to have a negative relationship with external contracting of innovation

18

activities. In fact, all three identified empirical studies support this relationship between
weak appropriability and internal innovation governance (Nakamura & Odagiri, 2005;
Steensma & Corley, 2000; Veugelers & Cassiman, 1999). Several case studies provide
further support for the negative relationship between weak appropriability and external
innovation contracting, with appropriability concerns being listed among managers’
foremost concerns when making innovation governance decisions (Narula, 2001; Tapon,
1989). There appears to be consensus in both empirical and case based research that

weak appropriability negatively influences external technology contracting.

Non Transaction Cost Determinants of Innovation Governance

In addition to the transaction cost dimensions previously discussed, other
potential drivers have been considered by researchers, including those suggested
anecdotally by managers as well as those called for by other theory bases. Findings
related to these antecedents are discussed so as to be inclusive of other potential
explanations for this governance decision, though the current study draws upon only the
transaction cost dimensions previously discussedz. Findings related to these other
antecedents of innovation governance will now be briefly examined.

Low cost goals and profit margin. Despite the widely held notion that firms
outsource all activities with cost savings as a primary goal (Piachaud, 2005; Quinn,
2000), empirical evidence consistently suggests that a low cost goal is strongly negatively
associated with external innovation contracting (Swan & Allred, 2003; Veugelers &

Cassiman, 1999). In general, higher profit margins are associated with conducting more

 

2 Though transaction cost logic may be used to support arguments for these drivers, they are presented as
non transaction cost determinants since they are not among the formal set of transaction cost dimensions.

19

innovation activities externally (Calantone & Stanko, 2007), though this relationship has
been shown to differ across low and high tech industries (Audretsch et al., 1996). That is,
high technology industries, high margins are associated with external technology
contracting, while in non high tech industries, high margins are associated with internal
development. Collectively, this evidence suggests that access to expertise and other
appealing characteristics of potential partner firms (i.e., “pull” factors) are more
important to decision makers than is cost savings when making innovation governance
decisions (Audretsch et al., 1996; Swan & Allred, 2003; Veugelers & Cassiman, 1999).

Firm size has been shown to have an effect on innovation governance habits by a
number of researchers. Small firms tend to either exclusively develop technology in
house or contract development externally, whereas large firms tend to use a portfolio
approach (i.e., both internal and external governance) (Veugelers & Cassiman, 1999). In
general, large firms are more likely to pursue contracted innovation activities than are
small firms, since larger firms will have a greater number of R&D projects and will tend
to contract at least some these (Nakamura & Odagiri, 2005). In addition to firm size, the
increasing scale of innovation projects has been noted to promote external contracting for
some firms (Howells et al., 2003; Love & Roper, 2002; Narula, 2001).

Other Antecedents. Aside from the antecedents of external innovation
governance already discussed, Swan and Allred (2003) examined a number of possible
antecedents of innovation contracting which otherwise have not garnered attention from
researchers. These researchers found that a goal of product differentiation is negatively
associated with external contracting, indicating that managers believe that by contracting

innovation activities they will tend to develop commoditized products since competitors

20

likely have access to comparable technology via contracting. Also, these same authors
found that the distance between marketing and R&D operations (in miles) contributed to

the likelihood of external contracting.

21

Table 2: Antecedents of Innovation Governance: Research Consensus

 

TCA-Based
Determinants of
Irmovation Governance

Consensus from
Innovation Governance
Literature

Relevant Empirical Findings

 

Asset Specificity

Positively related to
internal governance

Nakamura & Odagiri (2005),
Cesaroni (2004), Veugelers
(1997), Ulset (1996),

 

 

Audretsch et al.(1996)
Market Uncertainty No consensus due to lack Love and Roper (2005),
of examination Audretsch et al.(1996)
Technological No consensus due to Calantone & Stanko (2007),
Uncertainty conflicting findings Nakamura & Odagiri (2005),
Swan & Allred (2003),

Croisier (1998), Ulset (1996)

 

Behavioral Uncertainty

Positively related to
internal governance

Steensma & Corley (2001),
Pisano (1990)

 

Appropriability

 

Weak appropriability is
positively related to
internal governance

 

Nakamura & Odagiri (2005),
Steensma & Corley (2000),
Veugelers & Cassiman
(1999)

 

Non TCA-Based
Determinants of
Innovation Governance

 

 

Low cost goal Positively related to Calantone & Stanko (2007),
internal governance Swan & Allred (2003),
Veugelers & Cassiman
(1999), Audretsch et a1.
(1996)
Firm size Negatively related to Nakamura & Odagiri (2005),

 

 

internal governance

 

Veugelers & Cassiman
(1999)

 

22

 

Consequences of External Innovation Governance

Researchers began to place more emphasis on the outcomes of external
innovation governance around 1999-2000, once a base of research addressing the
antecedents of external contracting had been established. Consequently, at present, the
literature pertaining to the outcomes of innovation governance is relatively
underdeveloped compared to the body of research discussing the antecedents of these
governance modes. There are only six identified studies which consider performance
implications of innovation governance. Some of the outcomes of external innovation
governance that have been considered include innovativeness and various performance
measures (e. g., project performance, sourcing performance), costs, as well as others. Of
the six identified studies focusing on the outcomes of external contracting, half used a
single performance measure (for exceptions see Nicholls-Nixon & Woo, 2003;
Rothaermel et al., 2006; Steensma & Corley, 2000), meaning that there have not been a
great many studies examining any one outcome variable. The impact of external
innovation governance on each of these outcomes will now be examined:

Innovativeness/ Sourcing Performance. Research findings surrounding the
innovativeness and performance implications of external innovation governance have
been mixed. The use of a variety of different types of innovation relationships (e. g.,
contracting, alliances, etc.) has been shown to be positively related to the number of new
products developed (Nicholls-Nixon & Woo, 2003) since different types of linkages can
be used to gain different types of knowledge. For instance, partnerships with universities
can be used in learning about new advances in scientific knowledge, whereas contracts

with design firms lead to knowledge which is then directly applied to commercialization.

23

Collectively, these types of knowledge contribute to a firm’s ability to develop new
products. Similarly, the use of contracting for activities which are also performed
internally has been shown to foster growth of the product portfolio as well as new
product success, since this allows a firm to access knowledge developed outside the firm,
while maintaining its competencies (Rothaermel et al., 2006). Other researchers have
found that external contracting, and particularly early stage idea contracting, is negatively
related to competitive success since contracted ideas will typically be easily imitable
(Kessler et al., 2000).

It is difficult to derive consensus from a small group of studies, but it does appear
likely that the relationship between innovation governance and performance depends on
other factors such as appropriateness of external governance (as is the focus of this
study), partner selection (as discussed by Piachaud, 2005), or the type of control
mechanisms used (Carson, 2007).

Costs. As discussed previously, though cost cutting is often thought of as the
primary rationale for externalizing innovation activities, empirical work has shown that a
low cost goal is negatively associated with external innovation contracting. A related
question is, what impact does external innovation contracting have on innovation costs?
Unfortunately, there are very few empirical findings related to this question, with only
two identified empirical studies examining the relationship between innovation
governance and costs. In one exception, Veugelers (1997) found that where firms
possess an. internal R&D capacity, spending on externally contracted R&D tends to
stimulate internal expenditures. This seems to indicate that spending on external

contracting may place demands on the internal organization in interpreting results from

24

external sources, and synthesizing them with other findings (both internal and external).
Kessler, Bierly and Gopalakrishnan (2000) find no relation between either early or late
stage external innovation contracting and new product development costs.

Time to market. Despite the importance of speeding new products to market
(Bayus, 1997), there has been little examination of the role of innovation governance on
new product development (NPD) speed. In one rare exception, Kessler, Bierly and
Gopalakrishnan (2000) find that external contracting (and particularly late stage
contracting) is related to slower NPD. This may reflect the extra time required for an
external firm’s employees to learn about a particular product or application. It is also
plausible that when internal development teams struggle to solve late stage technical
issues, they may then resort to additional external development to help solve these issues,
causing further delays. Case based research, such as Quinn (2000) and Chatterji (1996),

indicates that external contracting can be used to help speed products to market.

Appropriate Governance and Performance

The basic prescriptive (or normative) premise of TCA is that when governance is
appropriate, or well suited to the characteristics of the transaction, performance will be
Optimal. However, where governance is inappropriate, or not well suited to the
characteristics of the transaction, performance will suffer (Williamson, 1985, p. 22-23).
As discussed earlier, it is expected that more hierarchical governance modes will be
favored when high levels of transaction costs are associated with external contracting.
The use of governance appropriate to the transaction’s characteristics allows for superior

performance through optimal efficiency when compared to other governance alternatives

25

(Brouthers, 2002). On the other hand, the use of inappropriate levels of governance leads
to inefficiencies imposed through either unnecessary bureaucracy or evaluation costs and
is associated with inferior performance and organizational failure (i.e., insolvency).
Stated differently, “To use a simple mode of governance to manage a complex
transaction would be to risk contractual breakdown, whereas to use a complex mode of
governance to manage a simple transaction would be to incur costs without gain”
(Macher & Richman, 2006 p. 5).

Typically, extant innovation governance research related to performance has
examined the direct relationship between external contracting and performance variables
(e. g., Kessler et al., 2000; Steensma & Corley, 2000), rather than the relationship between
the use of appropriate governance (given the transaction’s characteristics) and
performance, as is dictated. by TCA (Sampson, 2004; Williamson, 1985). Researchers’
failure to consider the appropriateness of governance given the characteristics of the
transaction may be the underlying reason behind the numerous conflicting results
pertaining to performance (as discussed in the previous section).

Appropriate governance has been referred to using various labels: alignment
(Leiblein et al., 2002; Nickerson & Silverman, 2003; Sampson, 2004; Silverman et al.,
1997), fit (Brouthers, 2002; Brouthers et al., 2003; Leiblein etal., 2002), conformity
(Anderson, 1988), transaction cost-enhanced governance (Brouthers et al., 2003), and
predicted governance (Shaver, 1998). Several articles refer to it by numerous labels.
Throughout the present study the terms “appropriate governance” or “appropriate

governance levels” are used (Nickerson & Silverman, 2003).

26

In order to provide a thorough review of research related to (in)appropriate
govemance’s role in determining performance, searches of top journals in marketing,
management and economics were conducted. As well, several published and
unpublished reviews of the TCA literature (David & Han, 2004; Geyskens et al., 2006;
Leiblein, 2003; Macher & Richman, 2006; Mahoney, 1992; Rindfleisch & Heide, 1997;
Zhao et al., 2004) were consulted. In this section, the scope of the review expands to
focus on the entire literature surrounding the performance implications of (in)appropriate
governance for two reasons. First, though there exist numerous TCA reviews, the small
but growing literature surrounding the outcomes of (in)appropriate governance has not
yet been well synthesized. Second, the innovation specific portion of this literature is not
yet developed enough to warrant review on its own; there are only two identified studies
considering (in)appropriate govemance’s impact on outcomes related to innovativeness
(Leiblein et al., 2002; Sampson, 2004). In order to include as many studies as possible,
research focusing exclusively on alliances is now included.

Within the larger TCA literature, research progressed similarly to the innovation
governance literature previously discussed. Many TCA researchers have examined the
antecedents of governance choice, though studies examining the performance
implications of governance choice were rare until recently. Seemingly, the literatures
pertaining to the antecedents of governance choice and this decision’s role in determining
performance evolved quite distinctly from one another (for more discussion see Leiblein,
2003). In fact, at present there exist only a handful of empirical studies (Appendix C)
focused on the costs and benefits of (in)appropriate governance based on the

characteristics of the transaction (Macher & Richman, 2006).

27

As can be seen in Appendix C, researchers have examined the impact of
(in)appropriate governance in a few distinctive settings since Anderson (1988) first
examined the role of “observance” with the governance mechanism (reps or direct sales)
suggested by characteristics of the film’s sales environment. Though this particular study
shows no significant relationship between appropriate governance and sales efficiency,
all seven other identified studies examining the performance implications of
(in)appropriate governance have shown that the use of appropriate governance is
positively related to performance or negatively related to firm failure3 (Argyres &
Bigelow, In press; Brouthers et al., 2003; Leiblein et al., 2002; Masten et al., 1991;
Nickerson & Silverman, 2003; Sampson, 2004; Silverman et al., 1997).

Clearly, the most significant contribution from this emerging literature stream has
been to show that use of appropriate levels of governance leads to superior performance.
F inns which inappropriately select too hierarchical a governance mode suffer the costs of
excessive bureaucracy; the competitive market no longer serves to combat inefficiencies.
On the other hand, firms which inappropriately select market governance suffer the costs
of unchecked opportunism, which can be managed better within firm boundaries where
stronger social norms and the ability to measure behaviors (not only outcomes) contend
with behavioral uncertainty (Nickerson & Silverman, 2003; Williamson, 1985).

The benefits of selecting appropriate governance have been shown in a variety of
contexts, including the governance of automotive manufacturing (Argyres & Bi gelow, In
press), shipbuilding component production (Masten et al., 1991), semiconductor device

production (Leiblein et al., 2002), international entry mode (Brouthers et al., 2003),

 

3 Note that while Silverman et a1. (1997) do not report a significant relationship between misalignment and
firm failure, they do state that raising the minimum frrm size for inclusion to $1.2 million in annual sales
leads to misalignment reaching statistical significance in its relationship with firm survival.

28

trucking fleets (Nickerson & Silverman, 2003), and telecom R&D alliances (Sampson,
2004). As can be expected from such a variety of contexts, there has been some diversity
in the outcome measures examined. While several scholars have used firm failure as an
outcome measure (Argyres & Bi gelow, In press; Silverman et al., 1997), one used a
perceptual performance measure (Brouthers et al., 2003), and two have used self reported
financial measures such as selling efficiency (Anderson, 1988) and organization costs
(Masten et al., 1991). The largest group of researchers has used secondary financial and
performance measures such as return on assets (Nickerson & Silverman, 2003), transistor
density (Leiblein et al., 2002) and patenting (Sampson, 2004).

As research in this area has advanced, another interesting research question has
emerged: are the performance implications of inappropriately selecting a market oriented
governance method equivalent to the performance implications of inappropriately
selecting a hierarchical governance method? Two recent research projects have
addressed this question, though they have each reached contrasting conclusions.
Sampson’s (2004) study of telecom R&D alliances finds that firms relying on too
hierarchical an alliance structure are dramatically (231-307%) less innovative than
properly aligned alliances. However, firms using a less hierarchical structure than called
for are only slightly (4-12%) less innovative than properly aligned firms. This suggests
that the effect of too much bureaucracy placed on innovation efforts through over-
govemance is more damaging to innovativeness than the unchecked opportunism
associated with selecting an inappropriately market oriented governance mode.
Conversely, Leiblein et al.’s (2002) study of the production of semiconductor devices

finds that firms selecting a less hierarchical than optimal governance structure more

29

severely under-perform than do firms selecting too hierarchical a governance structure.
This suggests that failing to safeguard from market vulnerabilities by inappropriately
selecting a market governance option more severely negatively impacts performance than
does the excessive bureaucracy imposed by inappropriately selecting a hierarchical
governance option. Unfortunately, other studies examining inappropriate governance
have not reported comparisons between the two groups of misgoverned firms, instead
choosing to compare only appropriate vs. inappropriate governance.

To sum, this fledgling research stream has consistently shown that inappropriate
governance is negatively related to performance. However, at present researchers do not
yet appreciate the relative performance implications of the use of governance structures
which impose unnecessary bureaucracy or do not sufficiently safeguard from market

hazards.

Hypotheses

Asset Specificity refers to the transferability of assets which support a given
transaction (Williamson, 1985). Transaction cost logic holds that where investments are
not easily transferable outside of a particular relationship, market forces will no longer
serve as adequate safeguards (Williamson, 1975, 1985). In this case, partner fimrs may
use these investments to their advantage in negotiations. For instance, there have been
several examinations of TCA with respect to the decision to use a direct (internal) sales
force or extemalize by using a sales agency (e. g., Anderson, 1988). In the case where the
costs of training sales staff represent a specific investment (i.e., the investment made in

training staff to sell a particular product is not easily transferable to selling other related

30

products), this will dictate internalization of the sales function (all other factors held
equal). If the firm were to rely on an external agency in this scenario, once this
substantial investment in training has been made, the producing firm would then be
susceptible to being held hostage in contract negotiations. The sales agency could take
advantage of the knowledge that if the producing firm switches agencies, they will need
to endure the training costs once again. As previously mentioned, the market no longer
acts as an adequate safeguard once these specific investments have been made. In the
case of the innovation governance decision, firms that pursue innovation projects
requiring investments which are more specific to the use or user will be more likely to
pursue internal innovation (Robertson & Gatignon, 1998). More formally stated,

H I : Asset specificity is positively related to degree of internal innovation

governance.
Importance of Learning by Doing refers to the extent that significant experience through
internal pursuit of innovation activities is necessary to achieving innovative output. As
proposed by Williamson (1991), one potential explanation for why firms externally
contract innovation concerns learning by doing. In environments where learning by
doing is critical, firms are not able to achieve returns fiom internal innovation efforts
until experience has been gained.

As stated in H1, firms that require substantial specific innovation investments will
tend to internalize innovation efforts. However, should these firms compete in industries
in which learning by doing is necessary in order to achieve returns from internally
governed innovation, these firms will be even more likely to internalize innovation

activities in order to maintain their current level of knowledge, to take full advantage of

31

existing investments and to protect their knowledge by limiting interaction with outside
firms (Cantwell & Santangelo, 1999). In environments where learning by doing is very
important, knowledge is likely to be viewed as key to future competitive success, and
intellectual property may be protected more fervently. One method of protecting
intellectual property is limiting co-operation with external firms. In this case, the
prohibitive transaction costs associated with knowledge that is critical to the firm’s
competitive success dictate that internal governance will be preferred. Contracting will
be complicated by the transaction costs imposed by the combination of specific
investments and the necessity of learning by doing.

Again, as discussed in H1, when specific investments are not required, firms will
tend to use market governance. In environments in which learning by doing is necessary
in order to achieve returns from innovation efforts, those firms whose innovation
activities do not require specific investments will be even more likely to use market
governance, as the time and costs involved in climbing the experience curve will be
prohibitive, particularly where the generality of the investments (i.e., non-specificness)
dictates that these activities can be contracted to other firms. More formally stated,

H2: Importance of learning positively moderates the relationship
between asset specificity and degree of internal innovation governance.

The positive relationship between asset specificity and degree of internal

innovation governance will be stronger (weaker) where the importance
of learning is high (low).

32

Environmental Uncertainty. As have previous researchers (e.g., Barney et al.,
1992; Robertson & Gatignon, 1998; Walker & Weber, 1984), two dimensions of
environmental uncertainty are conceptualized: market uncertainty and technological
uncertainty.

Market uncertainty is the unpredictability of demand in an industry (adapted
from Robertson & Gatignon, 1998). When market demand is uncertain, contracting is
typically hampered by the need for constant adaptation through ongoing renegotiations
(Rindfleisch & Heide, 1997), and the firm’s internal decision making processes become
more appropriate as authority structures are needed in order to more efficiently make
decisions in this environment (John & Weitz, 198 8). Environments characterized by
market uncertainty necessitate renegotiation of contracts, which often involves significant
penalties (a form of transaction cost).

In reference to innovation governance, in environments where market uncertainty
is high, managers will tend to constantly be re-evaluating new product development
goals, meaning that projects may be constantly modified, cancelled and created. This
state of flux imposes substantial transaction costs on market contracting, thus favoring
internal governance. Conversely, in environments where there is little market
uncertainty, contracting is dramatically less complicated since new product development
goals will tend to be more stable. More formally stated,

H3: Market uncertainty is positively related to degree of internal
innovation governance.
Technological uncertainty is the firm’s inability to forecast technical

requirements (adapted from Walker & Weber, 1984) and may result from changes in

33

specifications of the desired component or end product. In this case, it is argued that high
levels of technological uncertainty favor market governance — the opposite effect as
market uncertainty has on governance (Balakrishnan & Wemerfelt, 1986; Geyskens et
al., 2006; Ulset, 1996). In technologically uncertain firms, selecting market governance
provides the option of terminating the relationship with the current partner and selecting a
new one should the firm undergo a dramatic shift in technical requirements (Balakrishnan
& Wemerfelt, 1986). Another possibility, as argued by Swan and Allred (2003), is that
in situations where technology is constantly in flux, due to constraints on internal
expertise managers may have little choice but to rely on the market to support their
innovation efforts.

This relationship between technological uncertainty and governance has not been
clearly established in the innovation context. In the innovation governance context, it
may be that the contracting difficulties imposed by technological uncertainty overwhelm
any potential flexibility benefits gained through the ability to switch partners.
Contracting under technological uncertainty is complicated by difficulties specifying
technical details which are constantly evolving. Qualitative research (such as Narula,
2001) reports that the difficulty in negotiating and enforcing contracts is a primary reason
that firms do not rely more heavily on external innovation contracting. Despite the
existence of arguments supporting both possible relationships here, in concert with the
bulk of contemporary TCA research, it is hypothesized that high levels of technological
uncertainty favor market governance. More formally stated,

H4: Technological uncertainty is negatively related to degree of internal
innovation governance.

34

Appropriability is the extent to which results from innovative activities can be protected.
and are not easily diffused within an industry (adapted fiom Veugelers & Cassiman, 1999
p. 65)’. Strong appropriability indicates that fums are able to protect intellectual property
from innovation efforts and to profit from successfirl innovations. Traditionally, the
pharmaceutical industry in developed countries has been thought of as an industry with
strong appropriability (Penner-Hahn & Shaver, 2005). In this type of environment,
intellectual property protection laws have been established and are actively enforced. On
the other hand, weak appropriability indicates an environment in which firms are less
able to protect intellectual property stemming from innovation efforts and are less likely
to profit from successful innovations as these gains may be diverted to other firms
capitalizing on the intellectual property.

Appropriability is the first of the added problems related to innovation
governance noted by Williamson (1991, p. 292). In environments where firms cannot
adequately protect their intellectual property, firms will select internal governance so as
to most effectively shield knowledge-related investments. The potential consequences of
contracting innovation activities where appropriability is weak include creating future
competitors from current suppliers and having intellectual capital redirected to current or
future competitors. Both these possibilities raise the transaction costs of market
governance in industries where appropriability is weak. When the threat of knowledge
leakage is high (weaker appropriability) firms will avoid external innovation contracting,
relying instead on their stronger ability to protect intellectual property internally. More

formally stated,

 

‘1 Consistent with this definition, appropriability ranges from completely appropriable by the innovating
firm (strong appropriability) to public (weak appropriability).

35

H5: Strong appropriability is negatively related to degree of internal
innovation governance.

Contracting Experience reflects an organization’s previous experience in contracting
innovation activities to external organizations and is closely related to a firm’s behavioral
uncertainty. That is, firms with substantial contracting experience face reduced
behavioral uncertainty since they are more accurately able to assess the contracted firm’s
performance and manage the threat of opportunistic behavior (Gatignon & Anderson,
1988; Leiblein & Miller, 2003; Mody, 1993). When it is very difficult to ascertain the
quality of a contracted service or product (i.e., high levels of behavioral uncertainty),
substantial measurement costs may be necessary. When these costs are substantial,
internal governance will be favored since under internal governance both behaviors and
outcomes can be measured.

Firms with greater experience in this type of contracting will face reduced costs
involved in market contracting, such as search and co-ordination costs (Pisano, 1990;
Powell et al., 1996). Conversely, consider the costs and risks associated with externally
contracting innovation activities for a firm with no experience in doing this previously.
This firm faces high search costs since this is a new activity, and is prone to opportunism
since its lack of experience both prevents it from detecting opportunism as well as
hinders it in contracting in ways which best protect itself. More formally stated,

H6: Contracting experience is negatively related to degree of internal
innovation governance.

36

Performance Implications. The normative aspect of transaction cost logic states
that when the level of governance is appropriate given the level of transaction costs,
performance will be optimal (Williamson, 1985 , p. 22-23). Appropriate governance is
defined here as governance which is suitable to the characteristics of the transaction as
dictated by the contributors to transaction costs discussed herein. Too often scholars
have only examined the choice of governance form without investigating its impact on
performance (Heide & Stump, 1995; Rindfleisch & Heide, 1997; Sampson, 2004).
Before considering potential performance asymmetries between the two varieties of
inappropriate governance, first the question of whether firms selecting an appropriate
level of governance outperform firms selecting an inappropriate level of governance is
examined using multiple outcome dimensions (innovativeness, costs and profits) over
time. Firms that choose appropriate governance modes will not have to endure the
excessive bureaucracy or unchecked opportunism brought about by inappropriate
governance, and are hypothesized to outperform inappropriately governing firms on each
outcome dimension.

H 7A: Innovation governance inappropriateness is negatively related to
innovativeness.

H 73: Innovation governance inappropriateness is positively related to
total innovation costs.

H 7C: Innovation governance inappropriateness is negatively related to
profit.

37

Performance asymmetry between firms inappropriately over and under-externally
contracting. There have been very few empirical findings pertaining to the relative
consequences of the two forms of inappropriate governance. However, there exist
possible explanations for each category of inappropriate governance being more
detrimental to firm performance than the other, along with arguments suggesting a
temporal aspect to possible performance asymmetries (see Table 3 for an overview of
these hypotheses).

Firms which under-externally contract face several threats, all of which are
thought to impact short term performances. First, the failure to contract externally may
equate to lost short term flexibility, since it is generally a faster process to switch
suppliers than to develop a new technology in-house (Balakrishnan & Wemerfelt, 1986).
Further, excessive internal innovation governance may lead to cost inefficiencies as
market competition does not serve to keep costs in check. This is particularly important
as innovation efforts have become increasingly capital intensive (N arula, 2001). Firms
which over-rely on external contracting may be able to avoid these substantial costs.
These arguments indicate that firms which under-contract may perform worse in the short

term than those firms which over-contract. More formally stated,

 

5 Short term refers to performance implications up to one year post governance decision.

38

H 8A: Inappropriately under-externally contracting is more negatively
related to short term innovativeness than is inappropriately over-
externally contracting.

H88: Inappropriately under-externally contracting is more positively
related to short term innovation costs than is inappropriately over-
externally contracting.

H8 C: Inappropriately under-externally contracting is more negatively
related to short term profit than is inappropriately over-externally
contracting.

Conversely, researchers have uncovered numerous drawbacks of over-relying on
external contracting (Leiblein et al., 2002). Upon examination, many of these effects are
longer term in nature. The over-reliance on external contracting can result in eroding the
technological core of an organization (Howells, 1999) leading to the eventual inability to
manage innovation efforts as the firm’s absorptive capacity dwindles (Lane & Lubatkin,
1998). External innovation contracting has also been shown to be positively related to
time taken in developing new products and negatively related to competitive success
(Kessler et al., 2000). Firms also risk exposing their knowledge to other firms and
possibly creating future competitors (Piachaud, 2005). These potential drawbacks to
over-extemally contracting are more gradual processes, meaning that the consequences of
over-extemally contracting may not be severe in the short term, but given time the
erosion of a firm’s innovation competency will have severe performance implications.

However, in terms of innovation costs, under-contracting is hypothesized to
increase costs more than over-contracting across time periods. Firms which under-
contract are prone to cost inefficiencies, since the market no longer acts to keep costs

competitive. Over time, these cost inefficiencies may become even more pronounced, as

the cost implications of excess bureaucracy compound. Over-contracting also has

39

negative cost implications, as renegotiation costs must often be endured when contracting
is used inappropriately. However, these costs will be less than the cost of inefficiency
imposed by under-contracting. Firms which over-contract will likely learn to select
partners well and structure contracts to avoid renegotiation penalties. However, firms
which under-contract will be relatively less able to control inefficient spending.

For these reasons, firms which rely too heavily on innovation contracting will
more severely under-perforrn in terms of longer term innovativeness and profit than will
those firms which do not rely enough on innovation contracting. However, firms which
under-rely on contracting will face higher innovation costs in both the short term (H8B)
and longer term. More formally stated,

H 9A: Inappropriately over-externally contracting is more negatively
related to longer term innovativeness than is inappropriately under-
externally contracting.

H 9B: Inappropriately over-externally contracting is less positively
related to longer term innovation costs than is inappropriately under-
externally contracting.

H 9C : Inappropriately over-externally contracting is more negatively

related to longer term profit than is inappropriately under-externally
contracting.

40

Table 3: Hypothesized Temporal Outcomes

 

 

Time Periods

 

Short Term

Longer Term

 

Innovativeness

Under-contractin g
more negatively
related

Over-contracting
more negatively
related

 

Total Innovation

Under-contracting

Under-contracting

 

 

Costs more positively more positively
related related
Profit Under-contracting Over—contracting

 

more negatively
related

 

more negatively
related

 

41

 

CHAPTER 3: MEASURES AND ANALYSIS OVERVIEW

Data Sources

 

Several data sets are necessary to test the hypothesized relationships discussed in
Chapter 2. The Survey of Industrial Research and Development (SIRD) is the most
central to testing these hypotheses. The SIRD is conducted jointly by both the National
Science Foundation (NSF) and the US. Census Bureau and contains data for over 2000
firms pertaining to R&D costs, R&D employment and R&D outsourcing/collaboration.
The SIRD is merged with several other data sets to assemble the necessary set of
variables. The Standard Statistical Establishment List (SSEL) is prepared by the US.
Census Bureau and contains firm information such as sales, industry classification,
location, etc. Standard & Poor’s Compustat database contains financial data (i.e., balance
sheet, income statement, and statement of cash flow items) for publicly traded companies
in the US. Finally, the National Bureau of Economic Research’s (NBER) US. Patent
Citations Data File contains data pertaining to patenting activity and citations. See Table
4 for a summary of constructs and Operationalizations.

These datasets are accessed through the Census Research Data Center (CRDC)
where the data can be accessed exclusively by approved researchers. Once analysis is
performed, all output to be removed from the CRDC must be approved by Census Bureau

staff to ensure disclosure requirements are met.

42

Construct Operationalizations

 

Asset Specificity refers to the transferability of assets which support a given
transaction (Williamson, 1985). To obtain an indication of the degree to which a firm’s
innovation assets are closely tied to the individual firm, the citations made by a firm’s
patent applications are examined (Argyres & Silverman, 2004; Rosenkopf & Nerkar,
2001. ). Patent applications must cite patents which are used or improved upon (both
those patents held by the firm as well as patents held by other firms or individuals).
Since innovation is, by definition, “an iterative process” (Garcia & Calantone, 2002 p.
112), the specificness of an organization’s innovation assets (including people, pr0perty
and knowledge) is indicated by the source of the knowledge that the organization builds
upon. Here, the average percentage of self citations made (i.e., self citations/all citations)
in the firm’s patent applications during the past four years is used as an indicator of asset
specificity. Companies that build upon (and thus are knowledgeable with regard to) the
work of the wider community of firms are said to have more general innovation assets,
since the skills and knowledge resident in the innovation organization is less particular to
the firm. On the other hand, companies that tend to build on their own knowledge are
said to have more specific innovation assets, since the skills and knowledge resident in
the innovation organization is more particular to the firm.

Self citation is negatively correlated with the “breadth of impact” the innovation
has, meaning that firms which tend to build upon their own work tend to produce
innovations that are later built upon only in a narrow range of technology areas (Argyres
& Silverman, 2004). Further, Tratjenberg et al. (1992) shows that the percentage of self

citation is negatively correlated to total future citations (from both other companies and

43

the firm itself), but highly positively correlated with the number of future self citations,
meaning that (all else being equal) heavily self-citing patents will tend to be less
frequently built upon by external firms.

Importance of Learning is the extent that significant experience through internal
pursuit of innovation activities is necessary to achieving innovative output. This is
measured here as the industry ratio of full time equivalent (F TE) scientists and engineers
to sales (in thousands of dollars), which provides an indication of the importance of
internal learning by doing in the industry as a whole. The average ratio of FTE scientists
and engineers to sales in an industry provides an indicator of the quantity of internal
learning required in R&D. An industry level measure is appropriate here since this
construct’s domain concerns the necessity of internal learning prior to achieving results
from innovation efforts within a particular environment (i.e., industry). F irms are
responsive to both micro factors (factors pertaining to their organization and individual
contracts) as well as macro factors (such as the importance of learning in their
environment) when making governance decisions. However, a firm level ratio of FTE
scientists and engineers to sales is not appropriate here since that would provide an
indication of how much learning the firm has done, and not the importance of learning in
that innovation environment.

Market uncertainty is the unpredictability of demand in an industry (adapted
from Robertson & Gatignon, 1998). This measure gives an indication of the likelihood
and extent of unanticipated shifts in industry demand. Similar to Leiblein and Miller

(2003), Levy (1985) and others, autoregressive time series analyses of each industry’s

44

sales over the past five years are performed, with the sum of squared errors divided by
industry sales (SSE/Sales) serving as the measure of market uncertainty.

Technological uncertainty is the frrm’s inability to forecast technical
requirements (adapted from Walker & Weber, 1984). The NSF segments R&D into three
categories: basic research, applied research and development. Basic research represents
original investigation for the advancement of scientific knowledge that does not have
specific and immediate commercial objectives. Applied research represents investigation
in the discovery of new scientific knowledge that has specific commercial objectives.
Finally, development represents technical activity concerned with translating research
into products or processes (Wolfe, 2005). Here, a firm level indicator variable is used for
technological uncertainty, since firms within the same industry may have substantial
differences in their levels of technological uncertainty (and since the domain of this
construct concerns “the firm’s inability to forecast technical requirements”). Potential
projects have varying degrees of technological uncertainty (Raz et al., 2002), meaning
that firms have some control over their degree of technological uncertainty through their
selection of projects, making it appropriate to measure this variable at the firm level.
This variable is coded as 1 if the firm conducts any basic or applied research and coded 0
if only development activities are conducted. Since basic and applied research have a
higher level of uncertainty than does development, whether or not a company pursues
activities other than development indicates the firm’s level of uncertainty in its
innovation organization. Howells (1999) discusses the degree of uncertainty in various
types of innovation projects, ranging from true uncertainty, which is associated with

fundamental research and invention (what is deemed basic research here) to very little

45

uncertainty, which is associated with activities such as developing new models and
making minor technical improvements (which would be classified as development
activities here). As research progresses from basic research through development, there
become fewer unanswered questions, decreasing the firrn’s degree of technological
uncertainty and making the array of possibilities easier to describe contractually
(Croisier, 1998).

Appropriability is the extent to which results from innovative activities can be
protected and are not easily diffused within an industry (adapted fiom Veugelers &
Cassiman, 1999 p. 65). Here, this is operationalized as the number of patents issued in an
industry divided by the total R&D expenditures in that industry (millions of dollars). An
industry’s propensity to receive patents reflects the perceived strength of intellectual
property protection in a given industry (Kim, 2004). For instance, in industries in which
patenting offers little protection of intellectual property, relatively few firms will endure
the time and costs required to file patent applications. Conversely, in industries which
offer a reasonable level of intellectual property protection, firms will tend to be diligent
in applying for patents on any development which may potentially be valuable (Penner-
Hahn & Shaver, 2005).

Firm ’5 prior contracting experience is an organization’s previous experience in
contracting innovation activities to external organizations. This is measured here using
the historical ratio of externally contracted R&D to total R&D (internally pursued R&D
and externally contracted R&D). This provides an indicator of the firrn’s experience in
contracting innovation activities, which is closely related to its ability to contract

successfully and detect and control partner opportunism. Previous researchers have used

46

a count of the number of previous relationships in preceding years (Powell et al., 1996;
Robertson & Gatignon, 1998). Pisano (1990) used the historical ratio of intemally
conducted R&D projects to the total number of R&D projects at time t-4. Here, the mean
historical ratio from 3 and 4 years prior to the governance decision is used. The t-3 value
is normed to have the same mean value as does the t-4 value. These two values are then
averaged for those firms sampled in both years, with the single year response used for
firms sampled only once. By using the mean of t-3 and t-4, this allows for a larger
sample size, as firms need only be included in the SIRD in one of those years to have a
valid measure. This use of this historical measure also helps to avoid simultaneity with
the governance variable. That is, fewer of the contracted R&D projects at t-3 and t-4 will
remain in progress at t than projects at (for instance) t-1. This minimizes the impact of
projects which continue into the focal year.

Control variables. Firm size is measured using the log of the number of domestic
employees. The industry classifications adopted for this study are those which the
National Science Foundation uses in their publicly available data. Groupings of two digit
SIC codes are used to classify firms into fifteen industries.

Degree of internal innovation governance is operationalized as the percentage of
total R&D spending (both internal and contracted) on internally pursued R&D. As
discussed by Rindfleisch and Heide (1997 p. 41), a continuous degree of internal
integration (i.e., 0% to 100%) has been used by numerous governance researchers,
including the percentage of components manufactured internally (Walker & Weber,

1984), and the percentage of trucking miles driven by company drivers vs. the total of

47

company drivers and owner/operators (Nickerson & Silverman, 2003; Silverman et al.,
1997).

Innovation governance inappropriateness is operationalized as the absolute
value of the residual from the regression analysis predicting the degree of internal
innovation governance (H1-,H6), which will be referred to as the stage one regression
analysis (the two stage approach taken here will be discussed in detail in the next
section). Residuals are interpreted here as the degree to which a given firm deviates from
TCA predictions in its organization of innovation governance. As stated by Nickerson
and Silverman (2003, p. 33): “If transaction cost economics is correct in its prescriptions,
and if the theory has any consequence, then we expect larger deviations to be associated
with diminished performance.” This measure is equivalent to those measures used by
Silverman et al. (2003) and Nickerson and Silverman (2003), who used a continuous
governance measure (as does this study).

The first stage dependent variable (degree of internal innovation governance) is
constrained in range from O to 1, necessitating the use of a first stage estimation
technique which accommodates censored dependent variables, such as Tobit analysis
(estimation techniques will be discussed in depth in the following chapter). As discussed
by Breen (1996), when performing Tobit analyses, the estimated parameters can be
interpreted with respect to four types of expected values. Here, the expected,
unconditional value of the realized variable (internal innovation governance) is used as
the expected value in calculating the residual (actual internal innoVation governance-
predicted). The predicted value is constrained to the range of realizable values (ranging

from 0 to 1). For calculation details, please see Breen (1996, p. 27). This differs from

48

the typical method of calculating predicted values from regression analyses, simply
calculated as E(yi|x,)=XiB, which could lead to predicted values outside the range of
realizable values.

This approach is also conceptually consistent with the work of other researchers
who have used dichotomous governance modes (typically make or buy) and examined
the residuals from probit or logit analyses as measures of inappropriate governance
(Anderson, 1988; Brouthers et al., 2003; Leiblein et al., 2002; Masten et al., 1991;
Sampson, 2004).

Innovativeness is operationalized as the log of citation weighted patents received
for patents applied for in the focal year and the subsequent 3 year period, similar to
Sampson (2004) and Dutta et al. (2005). The four year period is used due to the time
required for innovators to recognize the potential value of an innovation and prepare
patent applications. Previous research has shown that patents are highly related to
inventions as well as new products developed (Grilliches, 1990; Sampson, 2004).
However, patents relate to a wide variety of innovations, ranging from those with
tremendous value to those with very little, meaning that patenting is an imprecise
indicator of innovativeness. Future patent applications must cite patents which are used
or improved upon, meaning that patent citations provide an indication of the influence of
an innovation. Prior research has shown that weighted patent citations are highly related
to the value of an innovation (Hall et al., 2001; Trajtenberg, 1990), providing a more
meaningful measure of innovativeness than does patenting alone.

It is important to distinguish between forward patent citations (citations received

in the future) and backward citations (past work cited). In this study, forward patent

49

citations are drawn upon for the innovativeness measure (to assess the future evaluation
of a company’s innovations) whereas backward citations are used for the asset specificity
measure.

Weighting is done as follows. First, the average number of citations received by
all the patents in the entire sample is calculated for each year. Next, the number of
citations assigned to each patent is then divided by this average number of citations,
providing the weight for each particular patent. The summed value of citation weights
for all a firm’s patents provides the innovativeness indicator (Dutta et al., 2005).

Total innovation costs. To assess both the internal costs (e.g., payroll, training,
supplies, equipment, etc.) as well as the cost of contracting innovation activities
externally, the reported costs (in thousands of dollars) for internal R&D as well as
contracted R&D are summed, with the log of this total serving as the measure of total
innovation costs.

Profit is the focal year’s before tax profit (in millions of dollars).

50

Table 4: Constructs and Operationalizations

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Construct Measure Year(s) Dataset
Asset specificity Average percentage of patent self-citations 1992-1995 NBER
made
Market Sum of squared errors from autoregressive 1991-1995 Industry Level
uncertainty analysis of industry sales from previous 5 SIRD
years divided by industry sales
Technological Indicator variable — 0 if firm’s R&D efforts 1995 SIRD
uncertainty are entirely development, 1 otherwise
F irm’s prior Historical ratio of external/ total R&D 1991, 1992 SIRD
contracting
experience
Importance of Industry average R&D headcount/sales 1995 Industry Level
learning SIRD
Appropriability Industry level ratio of number of patents 1995 Industry Level
divided by total R&D expenditures in that NBER
industry. Patent intensive industries are said
to have higher levels of intellectual property
protection.
Degree of internal Firm level ratio of internal R&D/ total R&D 1995 SIRD
innovation spending
governance
Innovation Residual from stage one regression. -- --
governance
inappropriateness
Outcomes
Innovativeness Log of citation weighted patents applied for in 1995-1998, NBER
4 year period 1996-1999,
1997-2000
Total innovation Log of Internal + External spending on R&D 1995, 1996, SIRD
costs (both 1997
internal and
external)
Profit Profit 1995, 1996, Compustat
1997
Control variables
Firm size Log of domestic number of employees 1995 SSEL
Industry 2 digit NAICS groupings 1995 SSEL J!

 

 

 

 

SIRD: Survey of Industrial Research and Development
SSEL: Standard Statistical Establishment List
NBER: National Bureau of Economic Research’s US. Patent Citations Data File

51

Analysis Overview

 

There are two stages of analysis, consistent with previous studies examining
inappropriate governance’s impact on performance (Anderson, 1988; Brouthers et al.,
2003; Nickerson & Silverman, 2003). For the first stage, the dependent variable is the
degree of internal innovation governance. For the second stage analyses, the dependent
variables are performance variables. This overview is intended to briefly summarize the
analyses; significantly more detail pertaining to the analyses is disclosed in the following

chapter.

Stage One: Determinants of Governance

The initial analysis examines which variables significantly impact firms’
governance decisions and provides the residuals, which indicate each firm’s deviation
from predicted governance. Since the stage one independent variables are at both the
firm and industry levels, regression with nested effects (also known as multilevel
modeling) is employed (Luke, 2004).

Multilevel modeling has become commonplace in the social sciences, including
marketing (e. g., Olson et al., 2005). In this case, the analysis predicts values of the
dependent variable (degree of internal innovation governance) based on predictor
variables at both the firm (level 1) and industry (level 2) levels. Additionally, since the
variable “firm’s prior contracting experience” is an inverse lagged form of the stage one
dependent variable, maximum likelihood (ML) estimation is used for the stage one
analysis. ML remains consistent under this scenario, unlike some other estimation

techniques.

52

Dependent variable:
Degree of internal innovation governance (level 1)
Independent variables:

Asset specificity (level 1)
Market uncertainty (level 2)
Technological uncertainty (level 1)
Appropriability (level 2)
Firm’s prior contracting experience (level 1)
Importance of learning (level 2)

Asset specifity*lmportance of learning

Control variables

Firm size (level 1)
Industry (level 2)

Stage Two: Inappropriate Governance ’s Performance Implications

Next, the impact of the magnitude and valence of the first stage residuals on the
outcome variables is examined. For both analyses described below, three outcome
variables are examined over three distinct time periods (see Table 5). Examining the
temporal effects of inappropriate governance provide an indication of the effect
persistence. As mentioned previously, four year timeframes are examined for the
innovativeness measure. Contemporaneous (year zero), year one and year two analyses
are conducted for both total innovation costs and profit. For the purposes of'H8 and H9
(the time based hypotheses) the contemporaneous and year 1 analyses will be considered

“short term”, while the year 2 analysis will give an indication of longer term effects.

Table 5: Outcome Measures Over Time

 

 

 

 

Innovativeness Total Innovation Costs Profit

0 Focal year and subsequent 3 0 Focal year (year 0) 0 Focal year (year 0)
years, as done by Sampson 0 Year 1 0 Year 1
(2004) 0 Year 2 0 Year 2

0 Years 1-4

0 Years 2-5

 

 

 

53

 

Two distinct analyses are performed to assess the implications of (in)appropriate
governance:

First, absolute residuals from the stage one regression are used as an independent
variable (with control variables) to examine if appropriately governing firms outperform
inappropriately governing firms across the three outcome variables. Analysis using only
absolute residuals assumes performance symmetry between firms under-relying on
external contracting and those over-relying on external contracting, since the absolute
residuals do not reveal valence of inappropriate governance, only the magnitude of
inappropriateness. This is necessary to determine whether appropriately governing firms
outperform inappropriately governing firms (H7). Next, this symmetry assumption is

relaxed.

Dependent variables:

Innovativeness (level 1)
Total innovation costs (level 1)
Profit (level 1)

Independent variable:

Innovation governance inappropriateness (level 1)

Control variables:

Degree of internal innovation governance (level 1)

Firm size (level 1)

Industry (level 2)

The valence of the residual (negative or positive) is employed to examine
performance differences between the two varieties of inappropriateness via an indicator
variable. The coefficient for this variable can be interpreted as the effect of under-

contracting. This variable is used alongside the variables from the previous set of

analyses to predict the three outcome variables of interest.

54

Dependent variables:

Innovativeness

Total innovation costs

Profit

Independent variables:
Under-contracting

Innovation governance inappropriateness
Control variables:

Degree of internal innovation governance
Firm size

Industry

(level 1)
(level 1)
(level 1)

(level 1)
(level 1)

(level 1)

(level 1)
(level 2)

55

CHAPTER 4: DATA PREPARATION, ANALYSIS AND RESULTS

Preparing Stage One Data

 

The data used to test the first stage model is assembled as follows. First, the 1995
Survey of Industrial Research and Development (SIRD) is linked to earlier year’s
editions of the SIRD (1991 and 1992). The 1995 SIRD provides the measures for several
of the constructs in the first stage analysis, while the 1991 and 1992 editions of the SIRD
are needed for the contracting experience variable. The year 1995 is selected as the focal
year for the stage one analysis since it is the most recent year that allows for the
availability of subsequent longitudinal performance measures. Specifically, the most
current version of the NBER patent database provides coverage through 2002 (Hall et al.,
2006). This time lapse allows for the three distinct four year patenting windows and for
citation of these patents afterwards, both of which are necessary to measure
innovativeness.

The SIRD is conducted at the parent company level (“establishments under
common ownership or control”) (Wolfe, 1999, p. 10) making it imperative that other data
sets used here are consistent. The National Bureau of Economic Research’s Patent
Citation Data File (NBER) is reported at business entity level (Hall et al., 2001)
necessitating that results for all business entities making up a parent company be
combined prior to merging the SIRD with the NBER. To arnalgamate business entity
data to the parent firm level, the NBER is first linked to the Standard Statistical
Establishment List (SSEL) using a naming algorithm similar to that described in Kerr and
Fu (2006). The SSEL is a comprehensive listing of all US. business establishments. The

algorithm used removes punctuation, spaces and trailing company information such as

56

“Company”, “Inc.” and “Ltd” so that (for instance) a listing of “J. K. R. Biochemicals
Incorporated” would be successfirlly linked with a listing of “JKR Biochemicals Inc”.
Once the SSEL and the NBER have been linked by this naming algorithm, patent data are
summed for parent companies using the ownership information resident in the SSEL, and
parent company patent data are merged with the NBER.

The dataset is then filtered so that it contains only US. based companies. This is
necessary as many of the variables in the SIRD pertain solely to R&D operations within
the US. This filter removes a potential confound from non US. based companies who
may perform a portion of their R&D activities in the US. (which may be captured in the
SIRD), but whose innovative output is largely a function of operations outside of the US.
(not captured by the SIRD).

Industry level measures are assembled from the publicly available industry-level
version of the SIRD (Wolfe, 1999), as well as by summing the NBER data by industry
classification. Prior to analysis, the data are examined to ensure all values are within
reasonable boundaries. This resulted in 359 usable observations (i.e., observations with
no missing data) for the stage one analysis (see Table 6). Correlations and descriptive

statistics for first stage variables are shown in Table 7.

57

Table 6: Industry Classifications

 

Industry SIC codes n (Stage
One)*

Food, kindred, and tobacco products 20,21 15
Textiles and apparel 22,23 <10
Lumber, wood products, and furniture 24,25 <10
Paper and allied products 26 <10
Chemicals and allied products 28 37
Rubber products 30 18
Stone, clay, and glass products 32 <10
Primary metals 33 17
Fabricated metal products 34 23
Machinery 35 50
Electrical equipment 36 47
Transportation equipment 37 13
Professional and scientific instruments 38 33
Other manufacturing industries 27,31,39 14
Nonmanufacturing O7-lO,12-l7, 40-42,44-49, 50—55 64
Total 359

*Note that ranges are reported instead of n for small industries due to Census Bureau
disclosure policy.

Preparing Stage Two Data

 

To perform the second stage analysis, performance measures must be assembled
and merged with the first stage data set, which contains the residuals fi'om the first stage
analysis. The second stage analysis requires more recent editions of the SIRD for the
innovation costs variable, the NBER patent citation data for innovativeness data and
Compustat for profitability. Note that the NBER is linked using the same procedure as in
preparing the stage one analysis (i.e., the naming algorithm). Correlations and
descriptive statistics for second stage variables are shown in Table 8.

As will be discussed in detail with respect to each individual analysis, sample size

varies for each stage two analysis since data are not consistently available for every

58

second stage dependent variable. Most firms in the stage one analysis are active in the
patent database (used for the innovativeness measure) in all three time periods in
question. As expected, sample size decreases over the three time periods in question.
When using the SIRD data as a performance measure (innovation costs), there is a more
significant loss of data in ensuing years, as not all firms are re-sarnpled in the SIRD
annually (Wolfe, 1999). Finally, when using the Compustat data (profitability measure),
only publicly traded firms are listed in this dataset, resulting in a markedly smaller

sample for these final analyses.

59

 

HO.VQ*** mo.VQ** _.VQ*
ammuz
0o; into—d- .ISONd ***m~. mood ***wN.o- N—od ***mN.o- GMWA owcfi ONE SEE w
2: .13.? was- :33 Sod .........&.o 1.2. $3 82 £55588... 5
8.. Red- .........2.o Seo- weod- 1&2- 2 3 :23 Saunas @8880 e
2: .1 S- Sod- 1&5. mad Es ~85 basses 32.289 m
2: 88¢ Sod ........:.o- Sam 83 3588:: 352 a
2: .83 23. Ed ammo weeszo aoaaeoaea m
2: 1.25 $86 985 causes aa< N
co; wfd 936 855300 =o5m>0=5 BEBE _
w e e m a m N _ an .em :82

 

 

moumafim o>uatomoQ use mecca—otou owfim HE E 033.

60

RES: coax—oboe 2: 2:5 9 now: £ cmoa
5m mag vs... .880 cosgocfi do.“ 32 .mmo:o>=m>o:E com womeoo: 880830 8:: 2: e8 EoEoSmmoE 30:30 23 85 202

1|.
6

5:... Eva... av?

0mm”: ammue ammuz sen: 0mm“: .Nmns

 

oo._ aim-Nd- ..:.:..cm.o 3.146 ..:.:..mod _.:_:..mm.o ommg owes 3% E5 0
ammnz ammne 2.“: awmnz 5m":
co; 12.26- Ed- _.:.:..v_.o- Nwood wad mmmd oo:mEo>ow cougoS—ZmEBE m-
an M“: Rn: ommne 3m“: wmoaofitmemmafi
co.“ de 356 anod- avwod Domed oonmfio>ow :2355: v
EH: En: mun:
cog art-Yo Isa-We cawve 3“de ESQ m
ammuz gm":
co; ..:.:..~K.o Ammo; minw 38o "5:965: N
“mm“:
co.“ Noam meN mmo=o>3m>o=£ _
o w v m N fl .>oD .9m :82

 

 

mogcflm oZEtomoQ ecu Hose—8.50 owfim wcooom ”w 2an

Stage One Analyses

 

A multilevel model is employed (using SAS Proc NLMixed) since both firm and
industry level variables are used. This analysis accounts for the right censorship of the
stage one dependent variable (innovation govemance), since many firms in the sample
performed 100% of their innovation activities internally. The estimation is analogous to
a Tobit analysis with random industry effects. This type of mixed model for censored or
truncated data has been used previously in biomedical (Thiébaut & J acqmin-Gadda,
2004) as well as ecological (Br0phy et al., 2007) research. The SAS code used. for
estimating the full model is in Appendix D.

To separately observe the effects of the firm level variables, industry level
variables and random industry effects, the model is first estimated with random industry
effects only, followed by the random industry effects with industry level variables, then
random industry effects with firm level variables and finally the full model (Table 9).
The full model will not converge, with the gradient indicating that the random industry
effects are preventing convergence of the model. Given that the full model with random
industry effects did not converge, and these random effects are nonsignificant (p > .05)
when included alongside either the firm level or the industry level variables, it is logical
to conclude that industry-level variation is largely explained through the industry level
variables included in the model; there do not appear to be any significant unexplained
industry idiosyncrasies. Thus, the model can be estimated using a standard Tobit model

without random industry level effects.

62

Table 9: Stage One - Results of Mixed Models

 

Dependent Variable: Degree of internal innovation governance

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Only Random Random Industry Random Industry
Industry Effects Effects and Industry Effects and Firm
Variables Variables
Alpha l.0734*** 1.0327*** 1.2059***
(0.02929) (0.4410) (0.05330)
Asset specificity 0.1990*
(0.1115)
Importance of -0.06780
learning (0.1242)
Market uncertainty -0.006 1 9
(0.006985)
Technical -0.03632
uncertainty (0.02091)
Contracting -1.2455***
experience (0.08079)
Appropriability 02005“
(0.1043)
Firm size -.01435**
(0.006571)
Random variance 0.006979* 0.003611 0.000801
parameter (0.003690) (0.002616) (0.000802)
Random industry 1 Industry *** 1 Industry * 15 Nonsignificant
effects 1 Industry ** 14 Nonsignificant
l3 Nonsignificant
N 359 359 359
-2 Log likelihood 227.4 223.7 31.3
AIC 233.4 235.7 45.3

 

Ustandardized parameter estimates
(standard errors in parentheses)

*p<.l

**p<.05

*** p < 01

63

 

Stage One Results

 

The first stage Tobit results (Table 10) indicate good model fit, with a Dhrymes
R2 of 0.554 (Veall & Zimmermann, 1996). Multicollinearity does not appear to be a
serious concern, as there are no correlations among these independent variables above
L30} with the highest variance inflation factor at 1.24, well below the standard cutoff of
10 (N eter et al., 1996). Residual plots do not reveal patterns indicative of
heteroscedasticity, and estimation of the model using the Tobit with heteroscedasticity
option in PROC QLIM did not materially affect the results, indicating that
heteroscedasticity is not a concern.

The asset specificity coefficient is positive but does not reach a standard level of
statistical significance (p=.102), failing to support H1. While no direct relationship
between the importance of learning and innovation governance was hypothesized, a
significant negative effect is observed (p < .05). As hypothesized (H2), a positive
interaction between asset specificity and importance of learning is observed (p < .05).
The effect of market uncertainty is nonsignificant, failing to support H3. Conversely, H4
is supported, as technical uncertainty has a negative relationship (p = .0506) with internal
innovation governance. Appropriability was hypothesized to have a negative effect on
internal innovation governance, but a significant (p < .01) positive impact is found here,
contradicting H5. Contracting experience was hypothesized to have a negative impact on
internal innovation governance (H6), which is confirmed by this analysis (p < .01).

Finally, the control variable firm size had a significant (p < .01) negative effect.

64

Table 10: Stage One - Tobit Results

 

Dependent Variable: Degree of internal innovation governance

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Construct Result Hypothesized Result
Relationship
Intercept l.l87***
(0.0552)
Asset specificity 0.182 + (H1) Not supported
(0.1 12)
Importance of learning -0. 131 ** NA
(0.0526)
Asset specificity x 1.369** + (H2) Supported
importance of learning (0.662)
interaction
Market uncertainty -0.0010 + (H3) Not supported
(0.0028)
Technological uncertainty -0.0404* - (H4) Supported
(0.0207)
Appropriability 0.117*** - (H5) Contradicted
(0.0444)
Contracting experience -1.244*** - (H6) Supported
(0.0781)
Firm size -0.0162** NA
(0.0066)
N 359
Log likelihood -7.82
Null model log likelihood -127.44
Dhrymes RZ 0.554
Tobit coefficients
(standard errors in parentheses)
*p<.1 **p<.05 ***p<.01

65

Stage Two Analyses

 

In the second stage analyses, the first stage residuals are used alongside control
variables to predict three performance variables — innovativeness, innovation costs, and V
profit. Similar to Brouthers et al. (2003) and Leiblein et al. (2002) (both of which use
discrete first stage dependent variables), OLS regression is used to estimate the second
stage analyses, and the first stage dependent variable (internal innovation governance) is
also used as a control in the second stage. This ensures that any performance effects
attributed to governance inappropriateness are not simply a byproduct of the level of
internal innovation governance. For instance, consider a firm which is severely over-
contracting (having a low level of internal innovation governance): by examining both
the effects of a high level of contracting (via the level of internal innovation governance)
as well as the level of inappropriateness, the effects of the inappropriateness of the
governance can be discerned from any potential effect from having a high level of
contracting.

For all second stage analyses, multicollinearity is not a concern (as shown by
variance inflation factors below 5 for all analyses). Further, for each second stage
analysis, the White test was performed to test for heteroscedasticity. No significant
heteroscedasticity was detected in any of the stage two analyses. Stage two analyses are
performed both with and without an “undercontracting” indicator variable as can be seen
in Tables 11, 12 and 13. This variable is used to examine whether there are performance
asymmetries between the two forms of inappropriate governance, and is discussed further

below as H8 and H9 are examined.

66

As previously mentioned, for purposes of these analyses, short term will refer to
contemporaneous (t) and the subsequent (t+1) analysis, while the latest outcomes (t+2)

provide an indication of longer term effects.

Innovativeness

 

The first dependent variable of interest in the second stage is innovativeness,
measured using the log of weighted patent citations for patents applied for during a four
year timeframe. Three distinct four year windows are examined, 1995-1998, 1996-1999
and 1997-2000. The large majority of firms in the stage one sample are actively
patenting during each time period. However, firms do not appear in the patent database
with greater frequency in the latter time periods (1995 n=321, 1996 n=294, 1997 n=273).
To ensure that there is a sufficient sample for each industry, firms are reassigned to
“other manufacturing industries” in the rare case when sample size in a particular
industry drops below 5 (sample size for “nonmanufacturing” is consistently adequate).

The results of the initial second stage analyses (Table 11) show a persistent
negative relationship between innovation governance inappropriateness and
innovativeness. Innovation governance inappropriateness has a highly significant
negative relationship with innovativeness in the 1995-1998 timeframe (p < .01), with this
negative relationship persisting into the 1996-1999 (p < .05) as well as 1997-2000
(p < .05) tirnefrarnes. With significant negative effects in each of the three timefrarnes
examined, H7A is supported by these results.

Under-contracting (i.e., too much internalization) was hypothesized to be more

negatively related to innovativeness than is over-contracting in the short term (H8A), but

67

 

less negatively related to innovativeness in the longer term. When the under-contractin g

dummy is used in this analysis, no significant effects are detected in any of the three time

periods of interest. This indicates that there are no significant asymmetrical

innovativeness repercussions of under or over-contracting, failing to support both H8A

and H9A.

Table 11: Stage Two - Innovativeness Results

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Innovativeness Innovativeness Innovativeness
(1995-1998) (1996-1999) (1997-2000)
Contemporaneous (t) t+2
Intercept -4.623*** -4.725*** -4.034*** -4.240*** -4.077*** —4.185***
(1.202) (1.225) (1.339) (1.353) (1.438) (1.451)
Innovation -5.312*** ~5.244*** -4.838** -4.648** -5.241** -5.l42**
governance (1.860) (1.868) (2.237) (2.244) (2.439) (2.447)
inappropriateness ~
Under-contracting -0.1 19 -O.284 -0.1 71
(0.262) (0.269) (0.280)
Internal innovation -0.1 11 0.106 -0.182 0.325 -0.0559 0.234
governance (0.993) (1.103) (1.102) (1.202) (1.176) (1.271)
Firm Size O.894"‘M 0895*" 0836*“, 0837*" 0822*” 0822*“
(0.0663) (0.0664) (0.0688) (.0688) (0.0727) (0.0728)
Industry Controls Yes*** Yes*** Yes*** Yes"‘** Yes*** Yes***
N 321 321 294 294 273 273
R2 0.447 0.447 0.420 0.423 0.405 0.406
Adjusted R2 0.417 0.416 0.387 0.387 0.370 0.369
F 1532*" 14.40"“ 1254*" 1188*" 1167*” 1094*"
Unstandardized coefficients
(standard errors in parentheses)
*p<.1 **p<.05 ***p<.01

68

 

Innovation Costs

 

The next performance measure of interest is innovation costs, measured by the log
of total R&D costs (both internal and external) for each of three years: 1995, 1996 and
1997. Since 1995 data are also used for the stage one analysis, the full sample (n=359) is
available for this period. However, the sample sizes are considerably smaller for 1996
(n=229) and 1997 (n=181), as only firms which completed the Survey of Industrial R&D
again in these years can be included (see Wolfe (1999) for a description of the SIRD’s
sampling procedures). Again, as discussed with respect to the innovativeness analysis, in
the case where the sample size in any particular industry drops below 5 due to data
unavailability, these firms are reassigned to “other manufacturing industries” (sample
size for “nonmanufacturing” is consistently adequate).

Innovation governance inappropriateness was hypothesized to have a positive
relationship with total innovation costs (H7B), meaning that appropriately governing
firms are hypothesized to realize a cost savings. However, in all three time periods
examined (1995, 1996, 1997) there is a strong negative relationship between innovation
governance inappropriateness and innovation costs (p < .01 in 1995, p < .05 in 1996 and
1997). These results (Table 12) contradict H7B, and show a strong, persistent negative
effect of innovation governance inappropriateness on total innovation costs. Firms that
govern “appropriately” actually face higher innovation costs than do firms governing
inappropriately.

Under-contracting was hypothesized to be more positively related to innovation
costs over all time periods of interest (H8B, H9B) than is over-contracting. However, the

results show a strong negative relationship between under-contracting and total

69

innovation costs, persisting over all three time periods of interest (p < .01 in 1996, p < .05
in 1995 and 1997). Tests of R2 change between models with and without the under-
contracting indicator variable show a significant increase in variance explained by
including the additional explanatory variable in all three periods (p < .05 in 1995, p < .01
in 1996 and 1997). The analyses including the under-contracting indicator variable
indicate that under-contracting firms face lower total innovation costs (contradicting both
H8B and H9B). Conversely, those firms which over—contract face higher total innovation

COSIS.

Table 12: Stage Two - Innovation Costs Results

 

 

 

 

 

 

 

 

 

 

 

 

Innovation Costs Innovation Costs Innovation Costs
(1995) (1996) (1997)
Contemporaneous (t) t+1 t+2
Intercept 3.472*** 3193*" 2.140** 1.870M 1.469 1.272
(0.859) (0.887) (0.953) (0.945) (1.021) (1.007)
Innovation -3.886*** -3.827*** -3.l25** -3.l33** -3.115** -3.031**
governance (1.347) (1.338) (1.445) (1.424) (1.498) (1.475)
inappropriateness
Under-contracting -0.487** -0.580*** -0.599**
(0.198) (0.216) (0.235)
Internal innovation - l .433* -0.695 -0.465 0.322 -0.924 -0. 107
governance (0.732) (0.786) (0.732) (0.779) (0.789) (0.840)
Firm Size 0859*" 0865*" 0.960*** 0.972*** 1016*" 1017*“
(0.0500) (0.0497) (0.0610) (0.0603) (0.0688) (0.0677)
Industry Controls Yes*** Yes*** Yes*** Yes*** Yes*** Yes***
N 359 359 229 229 I81 181
R7 0.572 0.579 0.647 0.659 0.673 0.686
Adjusted R 0.550 0.557 0.622 0.633 0.646 0.657
F 2678*” 2600*" 2605*“ 2558*" 24.42“" 2398*“

 

 

 

 

 

 

 

 

Unstandardized coefficients
(standard errors in parentheses)

*p<.l **p<.05 ***p<.01

70

 

m

The final performance variable of interest is profit. Since the profit data used here
comes from the Compustat database, data are only available for publicly traded
companies that can be successfully matched to firms in the first stage dataset. This
results in a markedly smaller dataset (n=77 (1995), n =79 (1996), n = 83 (1997)), which
cannot accommodate indicator variables for all 15 industries. Accordingly, a variable
indicating either manufacturing or non-manufacturing industry is used. The limitations
when drawing conclusions from such a small sample are discussed in the final chapter.

Innovation governance inappropriateness was hypothesized to have a negative
impact on profitability (H7C). Though the analyses for all three time periods show a
negative coefficient for inappropriateness (see Table 13), standard statistical levels of
significance are not reached (1995 p=0.134, 1996 p=0.218, 1997 p=0.277) , failing to
support H7C.

Under-contracting was hypothesized to more negatively affect profit in the short
term than does over-contracting (H8C), but less negatively affect profit in the longer term
(H9C). Analysis of the 1995 data reveals a nonsignificant coefficient for the under-
contracting variable, though there are significant (p < .05) negative effects in both the
1996 and 1997 analyses. Further, tests of R2 change between the models with and
without the under-contracting variable indicate that models including the under—
contracting variable explain significantly more variance in both 1996 and 1997 (p <.05 in
both years). With a significant negative effect observed in one of the short term time

periods (1996), H8C is partially supported. In the final time period examined, under-

71

contracting is more negatively related to profit than is over-contracting, contradicting

H9C. See Table 14 for an overview of the second stage findings.

Table 13: Stage Two — Profit Results

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Profit Profit Profit
(1995) (1996) (1997)
Contemporaneous (t) t+2
Intercept ~3l6.61 -165.60 -885.32 -612.75 -743.32 -472.87
(1224.84) (1223.41) (1456.10) (1426.30) (1684.92) (1644.05)
Innovation -2859.05 -2907.71 -2786.11 -2854.21 —2853.36 -2832.80
governance (1884.34) (1874.45) (2243.26) (2189.06) (2603.46) (2533.89)
inappropriateness
Under-contracting -425.71 -805.70** -956.63**
(318.33) (370.63) (413.86)
Internal innovation -1699.21* -1245.89 -1577.77 -707.15 -1520.61 -448.61
governance (950.21) (1004.01) (1131.11) (1174.09) (1315.57) (1361.81)
Firm Size 279.02*** 264.71*** 346.93*** 319.83*** 320.61*** 290.35***
(66.69) (67.18) (79.03) (78.11) (90.44) (88.99)
Manufacturing 198.95 59.55 188.45 -69.38 216.81 -96.59
Control (268.72) (286.87) (318.97) (333.07) (367.22) (382.25)
N 77 77 79 79 83 83
R7 0.238 0.257 0.237 0.284 0.1606 0.215
Adjusted R2 0.200 0.204 0.196 0.234 0.118 0.164
F 5.62*** 4.90*** 5.75*** 5.78*** 3.73*** 422*“
Unstandardized coefficients
(standard errors in parentheses)
*p<.1 **P<-05 ***p<.01

72

Table 14: Summary of Stage Two Findings

 

A. Governance inappropriateness and performance:

 

 

 

 

 

Hypothesized Relationship Result
Innovativeness Inappropriate Governance negatively Supported

related (H7A)

Total Inappropriate Governance positively Confiadicted

Innovation mated (H73)

Costs

Profit Inappropriate Govemance negatively Not supported
related (H7C)

 

 

 

B: The two varieties of inappropriate govemance’s performance implications over time:

 

 

 

 

 

 

 

 

 

Time Period
Short Term Longer Term
Contemporaneous (t) t+2
and t+l
Hypothesized Result Hypothesized Result
Relationship Relationship
In novativeness Under-contracting Not supported Over-contracting Not supported

more negatively more negatively
related (H8A) related (H9A)

Total Under-contracting Contradicted Over-contracting Contradicted

Innovation more positively less positively

Costs related (H8B) related (H9B)

Profit Under-contracting Partially Over-contracting Contradicted
more negatively supported more negatively
related (H8C) related (H9C)

 

 

 

73

 

 

CHAPTER 5: DISCUSSION, MANAGERIAL IMPLICATIONS, AND FURTHER
RESEARCH DIRECTIONS

 

Discussion
Determinants of Governance

 

The examination of the determinants of the innovation governance decision
strongly confirm Williamson’s (1991) suggestion that additional apparatus is required to
understand the outsourcing decision in the innovation context. Both the importance of
learning by doing and appropriability are shown to substantially impact firms’
governance decisions. Further, this study demonstrates that the governance decision is
impacted by both firm level. factors as well as environmental (i.e., industry-level) factors
as will be discussed below.

These results indicate that specific assets only promote internalization in an
environment where learning is important. That is, firms requiring specific innovation
assets operating in an environment where learning is important are likely to internalize.
The specificity of the company’s activities, coupled with the necessary time and expenses
involved in learning dramatically increase the costs of extemalization. In this
environment, once a contracted firm possesses specialized assets and has conducted
substantial learning, the contracting firm would be at a tremendous disadvantage in a
renegotiation situation. Contracted firms will demand prohibitive compensation to
develop specific assets and draw upon their significant learning (Ulset, 1996). Further, in
this scenario firms desire to develop and protect their specific investments, which act as a
deterrent to potential competitors in an environment where learning is very important.
Conversely, in environments where learning is less important, the presence of asset

specificity is shown here to not significantly promote internalization. In this situation,

74

specific assets unaccompanied by the need for learning do not act as a deterrent to
outsourcing. In environments where learning is less important, specific innovation assets
can be developed relatively quickly, without dramatically slowing down the process of
getting new products to market, meaning that contracting firms will not be at the same
disadvantage in renegotiation. This finding improves the understanding of asset
specificity’s role in the innovation governance decision; this role has previously been
examined without the moderating role of the importance of learning (Audretsch et al.,
1996; Ulset, 1996).

The significant direct effect of the importance of learning indicates that (all else
equal) in environments in which significant learning is required before companies can see
benefits from innovation efforts, they are more likely to outsource in order to bypass this
learning curve entirely (Quinn, 2000; Ulset, 1996). A high importance of learning
appears to be a likely indicator of industries likely to have an established network of
innovation services contractors, which have developed to support companies’ need to
speed products to market in this environment (Chiesa et al., 2004).

The finding that market uncertainty did not have a significant impact on the
governance decision is not completely surprising given the lack of consensus in this area
(see Table 2: Antecedents of Innovation Governance: Research Consensus) and previous
findings of nonsignificance (Audretsch et al., 1996; Love & Roper, 2005). It is possible
that market uncertainty may have conflicting effects, making its true outcome ambiguous.
As per traditional TCA logic, environments characterized by market uncertainty may
complicate the process of contracting because fluctuating product requirements make

documenting project requirements problematic (Geyskens et al., 2006). On the other

75

hand, highly uncertain environments may also promote outsourcing as firms wish to
manage the risks of this uncertainty by keeping headcount low (Howells et al., 2003;
Steensma & Corley, 2000). The findings of nonsignificance in this study (and others
studying the governance of innovation) suggest that the effect of market uncertainty may
be another area in which the innovation governance decision deviates from more typical
governance decisions. This will be discussed further below as firture research directions
are considered.

Technological uncertainty is found here to have a negative relationship with
internal innovation governance. Under technological uncertainty, firms favor contracting
because it is quicker and less expensive to change contracting partners then to develop
new internal competencies (Balakrishnan & Wemerfelt, 1986). This supports previous
empirical findings in the innovation governance area (Calantone & Stanko, 2007; Swan
& Allred, 2003; Ulset, 1996), and is in agreement with the consensus from the larger
TCA literature, suggesting that the innovation governance decision does not differ from
other sourcing decisions in this respect.

The finding that contracting experience isnegatively related to internal innovation
governance indicates that contractually inexperienced firms are less apt to outsource.
These firms face greater search costs due to inexperience as well as an inability to
properly structure agreements and protect themselves from opportunism, raising the
transaction costs associated with contracting and making it logical to perform more of
these activities internally (Pisano, 1990; Steensma & Corley, 2000). Contractually
inexperienced firms face greater behavioral uncertainty and, accordingly, tend to refrain

from outsourcing innovation activities.

76

It was posited here that in environments characterized by weak appr0priability,
firms will be less willing to work with other organizations (i.e., contract providers), since
their intellectual property cannot adequately be protected in these relationships
(N akamura & Odagiri, 2005; Steensma & Corley, 2000; Veugelers & Cassiman, 1999).
However, the opposite effect is observed here. This indicates that in industries in which
intellectual property protection is strong, firms choose to pursue innovation activities
internally, so as to have full control over the valuable knowledge accrued. Conversely,
where intellectual property protection is weak, firms tend to outsource innovation
activities since technological competence is not viewed as a viable path to greater
profitability (Narula, 2001). Where there is no opportunity to develop sustainable
advantage (whether due to weak appropriability or other factors), firms will favor
outsourcing over internalization (Steensma & Corley, 2001). In fact, the desire to imitate
competitors has been discussed as one of the explanations for the growth of innovation
outsourcing (Chatterji, 1996). In line with Swan and Allred’s (2003) finding that a
differentiation goal is negatively associated with outsourcing, and Bardhan’s (2006)
assertion that outsourcing is appropriate when innovations become commoditized, these
results indicate that in environments in which firms’ competencies can easily be
appropriated by competitors (making differentiation through intellectual property

unachievable), firms opt to outsource.

77

Performance Implications

 

Prior to discussing the impact of governance inappropriateness on various
outcome measures, it is important to call attention to the consistently nonsignificant
(p > .05) relationship between internal innovation governance and all three dimensions of
performance considered here. These consistent nonsignificant effects across the three
sets of performance analyses support transaction cost logic in that there is no one correct
form of governance. That is, neither outsourcing or internalization inherently result in
superior performance (Leiblein et al., 2002). Rather, performance is a function of
selecting the form of governance most appropriate to the firm’s environment (Sampson,
2004; Williamson, 1985). It is noteworthy that this holds across the three outcome
variables of interest, despite innovation governance inappropriateness having differing
relationships depending on the outcome variable. This suggests that the notion of their
being no correct form of governance may be universal across performance dimensions.

As hypothesized, appropriately governing firms are shown to be more innovative
than inappropriately governing firms. This adds to the findings of Sampson (2004) who
found that properly aligned alliances are more innovative, as well as several other studies
finding that appropriately governing firms outperform inappropriately governing firms
using other single performance outcomes (see Appendix C). With respect to
innovativeness, Williamson’s ( 1985) contention that appropriate governance is related to
optimal performance is supported here. Appropriately governing firms do not suffer
from the effects of governance mechanisms that either insufficiently safeguard or impose
unnecessary bureaucracy. The effect of this inappropriateness is shown to persist through

all three time periods of interest here. This indicates lasting innovativeness repercussions

78

stemming from inappropriate governance, meaning that inappropriate outsourcing
decisions will likely harm an organization’s innovativeness for years to come.

In terms of the asymmetrical innovativeness implications of under and over-
contracting, the results from this study do not reveal asymmetric effects from under or
over-contractin g. When viewed in light of the previous finding that inappropriateness is
negatively related to innovativeness, these findings can be interpreted to suggest that
governance inappropriateness is negatively related to innovativeness, regardless of the
variety of inappropriate governance employed (under or over-contracting). This finding
is in contrast with both Sampson (2004) and Leiblein et al. (2002) who find asymmetrical
innovativeness implications from the two forms of inappropriate governance (though as
discussed in Chapter 2, these two studies make opposing conclusions in this regard).

In terms of total innovation costs, innovation inappropriateness is shown here to
have a negative effect which persists over all three timeframes of interest. This indicates
that appropriately governing firms face higher total costs than do firms governing
inappropriately. This finding represents a significant departure from the existing
literature, as no previous studies (see Appendix C) have found any negative consequence
relating to appropriate governance.

There appear to be additional costs associated with governing appropriately, and
although (as discussed previously) appropriately governing firms are more innovative,
there is a cost associated with this innovativeness benefit. To reveal the relative effects
on innovativeness and innovation costs, supplemental comparisons of the respective
coefficients for both the innovativeness and innovation costs analyses are conducted.

Coefficients from the various analyses forecast that a firm governing 1% from predicted

79

governance would face a penalty equivalent to 11.51%, 10.54% and 11.37% of total
weighted patent citations (in the three periods respectively), while lowering dollar costs
by 8.56%, 6.94% and 6.92% (again in the three periods respectively), as compared to a
firm governing as predicted (all other variables held constant)6. While there is a cost to
governing appropriately, the benefit to innovativeness is disproportionately greater. To
illustrate one potential explanation for this finding, consider firms that inappropriately
utilize an “internal only” approach to innovation by choosing not to outsource any
innovation activities. While avoiding the additional costs associated with a “hybrid”
innovation organization (those firms which conduct both internal and contracted
innovation activities), these firms would also likely be less innovative. This finding is in
line with both Veugelers (1997) who found that in firms conducting internal R&D,
spending on outsourcing actually acts to stimulate internal costs, as well as Narula (2001)
who found that an internal competency must be maintained to successfully contract
innovation activities.

There are also asymmetries in the total costs that under-contracting and over-
contracting firms face. Under-contracting firms face lower total costs than do over-
contracting firms, with this effect persisting over time. As discussed by Swan and Allred
(2003) and Veugelers and Cassiman (1999), low cost goals have been found to be
negatively associated with innovation outsourcing. In situations where firms are

primarily concerned with cost savings, it is difficult to gain approval for external

 

6 Innovativeness calculations are performed as follows (cost calculations performed in the same manner):
Coefficients from second stage analyses are multiplied by mean values for independent variables with the
exception of the inappropriateness variable, which is first fixed to 0.00 (representing predicted governance)
then 0.01 (representing governance 1% from the prediction). A percentage is then calculated using the
ratio of the difference in the number of expected weighted patent citations to the number of expected
weighted patent citations predicted at appropriate governance (note that this is the antilog of the
innovativeness measure).

80

contracting. Similarly, in the present study, firms that over-contract may be less
concerned with cost savings (due to labor shortages, sales growth, or other factors) thus
subjecting themselves to higher total costs.

As mentioned previously, there are decidedly smaller sample sizes for the profit
analyses. These results need to be held to greater scrutiny than other results found here,
and replication of these profit results with a larger sample (including control variables for
industry) would likely prove fruitful in further research. The current study did not find a
significant relationship between inappropriate governance and profitability. It may be
that profitability is too “noisy” a performance indicator; there may simply be too many
contributors to firrns’ profitability to isolate this effect here. While appropriately
governing firms are more innovative, they face higher total innovation costs, with no
significant profit effects detected.

Over-contracting firms are shown to be more profitable (p < .05) in the two latter
years analyzed, with no significant difference between under-contracting and over-
contracting firms in the contemporaneous analysis. The results from the latter years
indicate that under-contracting firms are expected to be on average (all other variables
held equal) $806 million less profitable in year t+1 and $957 million less profitable in
year t+2, when compared to firms that are over-contracting7. Though it is difficult to
make conclusions based on results from such a small sample, it does appear that the
negative effects of under-contracting are more damaging to profitability than are those of
over-contracting. Under-contracting is thought to harm performance in two ways, each

of which negatively impacts profit. Under-contracting firms do not expose their

 

7 Note that the coefficients for the profit analyses can be interpreted directly; the profit variable is not
logged as are the other two performance variables.

81

company to outside expertise, thus missing valuable learning opportunities (Howells,
1999; Howells et al., 2003; Kessler et al., 2000). Further, the loss of flexibility suffered
by under-contracting firms is detrimental to performance (Balakrishnan & Wemerfelt,
1986; Chesbrough & Teece, 1996).

The emergence of this effect in the latter two time periods indicates that there may
not be an immediate negative consequence to under-contracting. There are two possible
explanations surrounding the time elapsed before the profit repercussion is significant.
First, products take substantial time in the development process; projects may not impact
a company’s revenues for years after the corresponding spending on R&D occurs (Bayus,
1997). Second, it is also feasible that failure to expose the internal innovation
organization to outside expertise may not have a negative effect in the short term, but
continued lack of exposure over time (whilst technologies continue to change) has a

significant negative profit impact.

Managerial Implications

 

Knowledge of the drivers of innovation governance can be useful to managers in
two respects. First, by understanding these factors, managers may better forecast
demands on both the internal innovation organization as well as the need to manage
contractors and accompanying spending. Secondly, these factors will also be of interest
to managers in contract innovation services businesses. Through understanding the
factors that make companies more likely to outsource innovation services, firms seeking
to attract this business can better target potential customer organizations as well as tailor

their service offerings. Specifically, the findings here suggest that both firm and

82

environmental characteristics define the situations in which firms are most likely to
outsource innovation efforts:
1. Firms not possessing specific assets operating in environments characterized
by a high importance of learning.
2. Firms facing high technological uncertainty
3. Firms possessing high levels of contracting experience
4. Environments characterized by a high importance of learning

5. Environments characterized by weak appropriability

Through understanding the performance findings here, managers can better
understand the short and longer-term implications of their decisions to internalize or
outsource innovation activities. First, there is no one correct form of governance.
Neither outsourcing nor internalization inherently lead to performance (Leiblein et al.,
2002; Sampson, 2004). As is confirmed here, performance is a fimction of selecting
governance appropriate to the conditions within which the firm operates. Appropriately
governing firms are generally more innovative than those firms governing
inappropriately. However, appropriately governing firms also face an increase in costs
which is proportionately smaller than the gain to innovativeness.

Understanding the effects of over and under—contracting should also prove
valuable to managers of innovation organizations (see Table 15). In terms of
innovativeness, the nature of the inapprOpriateness (under or over-contracting) does not
have a significant effect, though as discussed previously, appropriately governing firms
are more innovative than those inappropriately governing. However, under-contractin g

firms face lower total innovation costs than do over-contracting firms. In cost-conscious

83

organizations it may be easier to gain approval for internally conducted projects (Swan &
Allred, 2003). F irms which are over-contracting may be less concerned or simply have
less control over costs due to unexpected growth, tight labor markets or other factors. In
terms of profitability, over-contracting firms are more profitable, though this effect does
not become significant until the year after the governance decision. This indicates that
under-contracting firms tend to be less profitable due to lack of exposure to external

competencies.

Table 15: Performance Implications

 

 

 

 

 

 

 

 

 

 

Innovativeness Innovation Costs Profit
Governance Negative effect Negative effect No effect
Inappropriateness
Under-contracting No asymmetric Negative effect Negative effect
effect
Over-contracting No asymmetric Positive effect Positive effect
effect
Conclusions

 

This study examines the factors contributing to the outsourcing of innovation
activities using a sample of 359 US. firms. Further, the implications of the
appropriateness of the chosen level of governance are considered with respect to
innovativeness, innovation costs, and profit. The management of cross-boundary
innovation is of interest to both scholars and managers, and these findings advance our
understanding in several specific ways (see Table 1: Summary of Contributions).

First, the role of the importance of learning by doing in the innovation governance
has been established (Williamson, 1991). The understanding of both the direct impact of

the importance of learning as well as its interaction with asset specificity is shown here to

84

 

be relevant to this governance decision, and indicates a distinction between the
innovation governance decision and more typical governance choices.

The use of multiple outcome dimensions in this study allows a much richer
understanding of the implications of appropriate (and inappropriate) innovation
governance (as called for by Leiblein et al., 2002). Here it is shown that governance
inappropriateness is negatively related to innovativeness, yet positively related to costs.
Though the effect of inappropriate governance on innovativeness is proportionately
greater than its effect on costs, there is a tradeoff associated with the additional
innovativeness realized by appropriately goveming firms. Appropriately governing firms
do not outperform inappropriately governing firms across all possible outcomes. Further,
under-contracting companies are shown to have lower total innovation costs, but also
lower profitability when compared to over-contracting firms. When viewed together,
these findings begin to provide a more complete understanding of the implications of
inappropriate governance, and provide a foundation for future researchers to develop
further. It is hoped that the results from this study provoke greater theoretical and
empirical questioning of the relationships between governance appropriateness and
various dimensions of performance, so that transaction cost analysis can grow as a theory
base to make strong predictions with respect to the relationships between the various
breeds of governance inappropriateness and different dimensions of performance.

Finally, this study examines the relationship between (in)appropriate governance
and performance over time (Nicholls-Nixon & Woo, 2003). Here, the persistence of
these effects has been shown, and in the case of profitability, an effect which does not

take effect until time elapses is observed.

85

Limitations

There are several limitations of this research project. First, this project relies
exclusively on the use of secondary data. Though primary data collection at one point in
time would not allow for the analysis of performance over time as has been done here,
primary data collection could amplify or corroborate the first stage results obtained here
(particularly the findings with respect to the importance of learning and appropriabilty).
For instance, it is possible that the level of appropriability may change over the course of
the product development cycle across industries. That is, there may be differences in the
propensity for new product ideas or partially developed technologies to leak outside of
the firm in various industries, though the measure for appropriability employed here only
considers those innovations which have progressed to the point of patent application.

Though the examination of multiple outcome measures over time here marks
progress from previous research, examination of three sequential time periods is likely
insufficient to discern long term effects such as those stemming from the loss of
absorptive capacity or intellectual property (which will be discussed under future
research directions).

The profit analyses conducted here were particularly limited in terms of the
sample size (and thus, the control variables which could be used). Several industry
indicator variables are significant in the innovativeness and innovation costs analyses,
meaning that potential industry profitability idiosyncrasies have not been accounted for
by the use of the manufacturing indicator variable in the profit analyses. In short, it is not

possible to rule out the confounding effect of industry in this set of analyses.

86

The current study drew a sample of US companies only, and only examined these
companies’ domestic outsourcing. Examining the outsourcing habits of companies
around the world, as well as both their domestic and international innovation contracting
is necessary to gain a more complete understanding of the performance implications of

these innovation contracting decisions (see Bardhan, 2006).

Future Research Directions

 

The finding that appropriately governing firms face higher innovation costs
suggests an agenda for future research. First, managers will prioritize various aspects of
performance differently across organizations and environmental conditions: Under what
conditions would the (proportionately smaller) cost savings observed by inappropriately
governing firms be more valued than the (proportionately larger) penalty to
innovativeness? Presumably this cost savings would be more valued in organizations
with a shorter term perspective, such as firms with liquidity issues or a focus on short
term cost savings. The firm’s temporal perspective has been shown to influence the
relative importance of various new product outcomes (Hultink & Robben, 1995) as has
firm strategy (Griffin & Page, 1996) and stage of the development process (Chiesa &
Frattini, 2007). The potential role of outcome prioritization in determining governance
appropriateness necessitates further research. For instance, incentives based upon short
term metrics may motivate inappropriate governance decisions, which as shown here may
have a severe impact on innovativeness.

Further, which (if any) other performance dimensions are banned by appropriate

governance? Here, costs are shown to be positively related to governance

87

apprOpriateness. Other outcomes need investigation in the future — such as speed to
market, flexibility, and product quality. If there are other performance outcomes that are
negatively related to appropriate governance, what are the commonalities between these
performance dimensions? For instance, it is conceivable that dimensions more closely
related to the resulting innovation (e. g., innovativeness, product quality) will be improved
by appropriate governance, while it may be possible to improve upon dimensions related
to the process of developing the innovation (e. g., development costs, development time)
through inappropriate governance. Only through future initiatives observing a wide
variety of outcomes can this and similar research questions be addressed.

This study used three distinct timeframes for each performance variable of
interest. Though this was sufficient to discern the persistence of effects, longer term
analyses may reveal findings not uncovered in the current study. Some of the long term
effects that were thought to result from over-contracting include a loss of absorptive
capacity, as well as the loss of intellectual property to partner firms. It is probable that
these sort of long term effects do not reveal themselves without a lengthier performance
view. Argyres and Bigelow (In press) found that the effects of inappropriate governance
selection may take up to 20 years to observe, depending on the maturity of the industry.
Extending the present findings over a significantly longer time span is another area for
fruitful future research.

Finally, there remain unresolved issues with respect to the determinants of
innovation governance. Particularly, the effects of appropriability and market uncertainty
require further inquiry in the innovation governance context. Environments characterized

by strong appropriability were hypothesized to encourage outsourcing, as firms can be

88

more assured when working with partner firms in environments where intellectual
property is better protected (Nakamura & Odagiri, 2005; Steensma & Corley, 2000).
However, it was found here that in environments where appropriability is strong, firms
will favor internalization, as this represents an opportunity to develop an inimitable
competence (Bardhan, 2006). Under which conditions does strong appropriability
encourage firms to work with partner organizations, and under which conditions does it
discourage this behavior? For instance, it is possible that the nature of the outsourcing
relationship and the types of control used by the contracting company (Carson, 2007)
could make outsourcing arrangements more effective under strong appropriability. It is
also possible that appropriability impacts the supply-side of the market. That is, where
intellectual property is well protected, contract innovation services organizations may be
less prevalent; necessary knowledge may only exist within traditional organizations as
this knowledge has not widely diffused to other organizations such as contract innovation
services firms.

The possibility of two opposing effects stemming from market uncertainty in the
innovation context requires further examination. Market uncertainty may complicate
outsourcing by requiring frequent renegotiation with contracted firms, but also encourage
outsourcing; firms in uncertain markets may be forced to outsource to keep up with
quickly changing market needs. Thus far, no known studies have found a significant
effect of market uncertainty in the innovation governance context, with the current study
and two previous studies (Audretsch et al., 1996; Love & Roper, 2005) finding the effect
of market uncertainty to be nonsignficant in the innovation context. Interestingly, the

TCA literature as a whole finds this to be a significant effect (Geyskens et al., 2006),

89

suggesting yet another distinction to the innovation governance decision. Future
researchers may find it helpful to consider market uncertainty as two separate
dimensions, as suggested by the previous speculation of conflicting effects stemming
from market uncertainty. Through examining the differential effects of both the extent to
which contracting is made difficult due to market unpredictability and the extent to which
market unpredictability favors risk management through outsourcing, market

uncertainty’s impact on the innovation governance decision may be better understood.

90

 

80380 8mm...—

08 8 80088888 0. :8 80.280
0888080800 0:. 8 «00.8.8 03:08
8080 0 08 888050 00—00 00888”:
000380 Emma. 05 8 >80 8: Q8888
88080 8 .0888 20888 may—00:88.0

 

 

 

 

 

0: 80888 80808: 000880 :8: 8 38800888 8088802
88088 38080 8 @8800 80.0802: 0: -88 80 8:88:88 8 088:: 8:088:
8808 2.85 000380 :8: 8 w80800 w88008=08 AA000~¢0_n:80\x.00__w..0.V 888050 888.:
88080 8 @8800 303:8: 0_ :0000000 :86 008088 8:88:80 00:00 008880 .8808 8.88 .8 88.82.
88828: 808808 000 000380 .m.D 3. 080380 Dim £0888 09080.08 €088.80: Ahoomv 088m
@038 SQA _ 88 0888880800 53: _0>0_ 8088: 082 8808800 :80008 88888”: :80 0880—00
8. v 3 828
0:30 80:0...0088
088080 :80
.8880 080 x0 8 80088 80888 308808
08:02: 0 08 8808 8:08:28 ”8880 808 8.0230
8 088 :8: 8 08088 0.880: 80:0 .8888 Ammo—0808
0>8 :000 588008 800—88 80 088 303808 080 05 .8 8888880 538000
0380000 .8880 80:0 9888808 8 080.088 .8888 0.08880 - 88808:
308880: 802: S 008808 0:08
008088.808 “0:830 no 8088: 08888 080.088 038000 8:880 .8888 8:80:82
0 08 0:80 80:0 80 08088.0: 08088008 08088809 08:08:88 .8 8808
“0:820 80 8088: 03:00: 0 08 080808.80 8:980 .8 08.0808 00:850. .302: 5880008. :88 5880008
:08 08 8 888 08080 08088008 w8080080 00 808880 x0 8 080 8v 800880 8080028 0:880 8:00:02
088850 8 8088: 08:02: 80:: 0 08 9:2 w8808a0 08880888 >03 800: 8880 8:000:08 0:30 80:0 .8 .883
28800 080 08 8: :88 8.800 .009 x: 080:0 C: m “80 N 0380852 8 09C. .808 08 .8 088 028000 Creed 80000
0.0.280
8 .8: $888.88» .883.
08:08.: .80 80800 0880082 88:0:8— 0038...; 80:80:06:— \ 085.8

 

 

 

 

 

< XHQmem/x

 

88003: 0808860 888088: 88088: 1% 88093

91

 

EW 0.. 8.8.82 .888 .888
:08800 .008 08 0: :8000 8.8808
00:00.8 000880 0:0 .8800 .3

0:08: 0:0
00:08.: 582
8 08: 000880

88:08:02
0.00 0.800

 

 

 

 

 

 

 

 

 

805 m888w800 .088800 80890—080 8080—2080 0:88.: 0:0
8:008 w8:000 088.: com 005800— 8808 .8800— <2 <2 <2 8:802 008:0
80:8 0>00w0:
00: 8:088 .8 00000 >80 00:?
dd: 00:800800 :0 0880:0980
8 0808: E08800: 000 :00880080
0:0 :0000800080 88.0888 A903: :0:0>0:8
80:00.. 08888 Q00: ”850: 08800800 8 000:0 .30: 8:08:88
088008800 :0 08:08:08: :80 EN 8 88: 58:08:88.0: ”80:08
0:0 8: 8
.0888 0>00w0: 0>0: GA :83 D88 8880 8000:
000088 .8 00000 0:0 88:00 80:0: 088008800 .30: :08 8880800080 308800.: B02
00:3 .90: :0 M80880 .8 0008—8:— :0 088880 8000888 80:00., 0:0 :0:0>0::_
:0 0808: 8,800: 0>0: 8088880 08:: A8: 8 0A: .080 $8888 Q03: 88...: 8 00:80:00:
0:0 800080880 8000888 80:00., w8008:08 Q00: .80 :0 ”000: 0:0 08:80? £83 88000
.085 “A8: 8 08:08:00 Q85 :80 _ 008800 08.: 0:0 Z 08880:: 8.6: 80:08 0:0 8: 80m 0:0 8:80:02
0:088:80
0:0 80808 08808
808800 .9888: 8::
258 .3583 .8 .0:
.0880: 58:88 0808800
.0w0 :8”— H00_:0_00> 88:00
A8888 8000888
0080080 0008 .8 8:80 8 00000—0.— 8088 000080 _00:._0> .8 8.6— 680888
080:0: 8:008:80 0: 8 80:0 .8008 0008 0:0 8:008 0088000: 0:0 0808800 8:80..
8: 0: 80:: w800:0080 08885 0000000 800 0088000 .9008 0:0 00:8: .8808: :8: 008888 00:00 8088802
8:008 30: 0:0 0:00:80 8:008 08 08080 088.: w8080080 .8 088.: 8:: 88.88: w8080080 088080 A80: 08885
08 .8 080 0:: 0: 080—00 30.0800: 08::8000088 :8: :83 08888 8:008 .3 0088.80: 008880 .8 85 32:8 0:2. 0:0
0: 8 850:0 0: 8880080 080,—. mm: 800888 _0080> 0.80: 8 085 :08: w88:0080 088005 82 88:00:83:
08.280
.8 :8: $88800.» 8:80..
08:08..— 000 80800 088.800: 8:00:88: 08:88.» 80.80000:— \ 08:80:

 

 

92

 

5:50 82:58:

:5: :8: 88:5 5850: 50—0582
:0 8:5: 8qu: 5:50:58: 2580:
:0: t0:::: :58: 5:50 85:52

:5: 8:28 808:? :0 890 588::5 :0
835:: 85 8838: 5:50:58: 85:80:

 

 

 

 

 

 

 

 

 

:0: t0:::: :25): 5:50 5:05:02 3:: :0:2::o: 5550:.
:8 8558: Q5: :0 5:85 5838: -53 58:5 2050:: 0:3 :08 :0 808:? :0 585852):
5:50:38: 02 .:::::0 18:58: 80:88:55: 50203 855:: 80:53 30—0582 05255
:5: Q5: 58:05 8525 5838: .m.D .288: :0 8:»: 5:85 3050—0502 $83 00 3 :5
5:50:28: 2:58: a :0: t0:::: :25): 85 5 55.: :N 8:02 30—05825: day: 15:25 5 58585: :05 2-50:0: Z
58:58
05 0: >8:
m: 8050::
:::0:500
a 8553
.:::0:500 ”59:5:
55 :0 :0:0::0:: 85 0:5 5883: 5? ::08m
58:58 a 85 :00585— 2: 88885 8:2. .::2m 15:25
5058: :30—0582 :0 5:35:26 :5: 8:05: :3 8:088:
25 85 85:: 59:5: :58: 2F ..m.D 5 :88: :0 :880:
50:50:50 5:288: 8:555: :28: $5825
5505025 0:95: E5528 58:8: Gum: .550502 52250::
0: :0: ::2 8350502 52250:: :8 :a 50:: 35:: 3:8: 858 m5:: 5:: 85 3 8:30 2:5:
8:28:58 Q5: 5 :8>5 3:28: 85 .8528 530% 5 30:05:02 :0 88588: .85: 58500
2:5: 88085 585850 30:05:02 :8 5 :558: 85 38:5 .85.: 8::::: A5882:
:0 5:535: 85 .5255: 550502 5:850:50 88: 85 225:3 $20582 8:0: 50:88:
:0 855:: 85:: a 0:: 085 :25, 880:: 2:02 5:35.: :5: 85:20: :0 835:2 $8.8 50880
.:0:m>0:5
:0 88:0: 8528 m5m800m :0: 800
:5: 8:: :0 8: 38852 88:85
..m.D 05 5 88 8558:: 858:500
flax—:5
:0 :5: A20358> .550:
85:5,: :5: 52:00 80:88:02 55889 8355:» 55:28::— \ 805:<

 

 

93

 

0030— 03 0005 :00? 00095 0:02.030

.:0E0_:000

0:808:80 00 30:5

.0:0000_:0w00 .0w50_0§0000

 

 

 

 

 

 

 

000.: 0 000 0330500 0305803 00—2000 :0 $0050: #:0838000 00 #:0503508 60:80.30: 35:00
00 3:300:00 00H. .swE 0:0 0000:0000 >w£0~£000 000:0 505 00009005 A:0_:0_:~00: #01 E0: 6935260 305090502
£030 :00? 0:0 .32 000 0w:_0_0§080 000::0000 0a: 0000 00:80.30: 00 300505 0505803 000 50:50:00 .00 >80000<

E08035“:— :0:B 0006003 0033 03:09:80 #0: 0:0 900—000—0000 0005 .05—5 3000:0800 .00 COOS x0100
:fim 0:00—98 803300050 .00 000:5. .m.D 33:: m— #:0803502 0055260 000:0 aE0_::t0%~0 .00 000:: 0:0 0:00:02m
00000.00 02:00: 00000 020: Bongo—:80
0:30am $325 and 50m
600000.00 0>00w0: 805_-:0: .005 00—3
000.000 03:00: 00: 00:0 :50 A800,“?
0>00w0: 000:: :0: 000 000:: 3:23
0000:: 02:00: 0 000 3053380
900m .900M 00 0000 0:: .00 0::0. :—
000bo :000>0:5 0:60
03:00: a 00: 0:20 000—008 0:0 000000 :0: Guam E0650 8:508: .5300w
0>0mw0: a 00: 0080 E2050 530:1: 00:00.5 .«0 00000 0:03 .0500 000.09: 00050:m
400000 03:00: a 00: 008 5000:0050 0000 50:00:00 0:3: .008 8000:0050 .00 00::0:00m
A2630: 0:0 00:00: 000:0 0:0 0000:00 >0E00E .093 :000:0000 .0w5:0000: 05 00 005:2.
swan; “00.0.00 0>00w0: a 00: 030 0:0—0 :0 3:0 #:0500350 000:000w 3:000:00“:—
dfim 0:05: 335005 00 b_0:0:00: 00:00:03 Q03— .b_0:00:_ Dam 00000000 @003
000:0:c5 b03000: 0:0—050350 900% 00:20 MD mm: 0:02 w5x0t00:3 .0:0:S0~9:0 ES: £30 900% :0:0& 0:0 0>0q
50:25 030000800 :wE 50$ 33300:
0:: #:0052000 00:00.: 505 33200: 4000008005
.0_00w 5000500000 .005 E0383: :0 50:
000200000 0_ :0205—000 000000: 3:00.05 ..m.D >009: w:_00:00
000 02:00:80 00030:: E03835 0320:0000
.003. 0:: $000.12: :003000 8:800 0:000: :_ 005888 0003 000000:
00000005 50$ b03000: ._00w 0000 .Aw:_§0£::0:0 003005, 0:00:00005 00:00.08 50:85 0:0E0w0:02
30. 50.5 300,003: #:0500050 00:00:: 30905 00 0:000:3500 aw:_0u:00 03.000800 €058.30 :000>0:£
~00: b03000: £00m 80000000000 00:00:05 000000 :00000005 30—05—00. 00:00:: .Qdm 0:0 gag—SE 00:00::
505 303003: 00300000 0_ 00:: 000503 030000: :0 swan: 00:00:: :002509 3:306 ._00w 00 3:50.. AMOONV
5903000 30—22000 00:00:: 00:0me .m.D 53 0:800:30 0:02 0:0me 0000-32 ._00w 5000500005 00:: 0:0 :0Bm
0_0>_::0
00 3:: 3033...; 02:50..
$505,: 0:: 0:00:00 0000000002 0:00:03: 00300:; 0:00:00005 \ 0.85:0.~

 

 

 

 

94

 

.322500“ 032:: 8 3233::
«305:: “06:: Bunny—atom mien—cm :mE

:30— 0000»

Ym “5800:0008
00:05:38.:
30:30:30

003053 600500:
v =0 :0 2880003:
8 “0309 b23398

 

 

 

8 £000— 38 8308—004. 30:80:30 0 8.: @0553 “:02.“in again—05:
:wE 0025 3:0 :5 .00§£._0a 0003 8598900: macaw .8059»: 00=0§ot0a «50“ 48201600
9:058 :wE 5:? 6030800 3 0:020:58 98 55800050 w.:0=50 £58005 3:002:50.“
m: w~:0=00= 60:80:: How 005258009 8 “5:50—05: 30:05:80 0:023:00 3:058 awe—23000 mo 09G.
53 5:588 05 b5§8§ as: .0528: 5:593 .50 032503 1.58.
5235.5: $005535. 30300: 3:05.058 63:03:5— .m0>:00.30 5:033 2080?:02
9:80:20: 05 00.30.09: $5950 05E 8:3 0080:: no 8055800 0309830 .8358»: .«o >E0v00<
no 8&0: 05 .05 «00%.; 00.50008 55:? £00.85 :000 wimgo: J=0Em20>0c 58800050 .b=ESE: 88S >350
002058509 So.“ :0 :0 0:230uw0m mafia—80 won 98 :2:m_:_00< bzfiwmao $005235. 303223009 . 98 «80:00am
.uaaomawmm
0:0 :00; 0
380>000f0w5§>w0
0303308
8.: 3:50:0on
.05 Mom—m 030
00>0000*Em:=35amo
no 000::
ofl< 00:80.35
Mar—35:0,:
1:82:58 ‘3 “00::
Ho.“ 5088 Emu—mafia
«£30500 $00. :0
0:90:53 003 :00m 090.5260
0:02.:ch £0: 03 team: $5.50: 0333550
28 320.: soEm_=c0< £229.25 0.0:.— uow bfisroaqo @039: .3902 mafia—00:
02E: 0 “En 0.00—0 0389508 .3 205. :50 05—5 .09an ”8203600 £030
07%..“ :0
.3 :5. Amv0_:a_._a> :2... :3.
09:25— 1.; 33:00 00320—002 E06550: 0030.05» 2.015906:— \ 0.85.30

 

 

 

 

 

 

 

95

 

3235 :00 Ea

A..008>:00 500803800

 

 

 

 

>088— 808w8~ 00 0000 00 080800s€ab
00 000303 AQQM 38008 108080008 5.0082000 0000
dag 108008 808 8000 00 0000 8088 08:80:88 .00 0808030 .0068 0338
008090088 388—08000 80080008 00008805: 0800000: 00—0 00: 00 008 0083000
0880003800 3800 :0: 80¢ 8000 30608 6880008200 500803800 .«0 00380:“ 3800p.
00 0000 00083008 08080 3080—08000 8 08:80 @003 >863 30:08.9 8080803800 80809882
8080808800 0880880800 0003 10080—08000 3208000 5.0083000 0000 0&0080m
>083 880 8000 00 0000 0080308 w880 w88=¢00 .00 08080:“ 3068.» 08880 $33 =0800=2
08.80 3080—08000 w800098000m 0000.008 3‘ 0:02 .00 0008002 3080—08000 w800008000m 0:0 08.80? Z
.0088 3108080
30302008 00 300.: 002 000 056 88: 0_
bzfiatmoaam 00083 .w80800 38080
0303008 00 0088— b0>zaw00 0_ 08002000
0000 .00 0088008H .w80800 38080
0203008 2020 00—08 000 $000 8000
00800000 3:88 0805 48000.00 3.58 0880 30m 088003000 08 60—08000
0 000 00 3002 0008 08 880080.08 00090008 00 0x0: 000 00000 .08039088
38008 w820> 088m .w80800 38080 $8088 $8080 00 00088000
000 38008 53 00: 00 0000 08.6 038 3808000 80308088 00809800
0000053 538080 >w208000 00800 00 088m 2093 .00 00830098 803080.08 zozom 8000000m
00:0: 8 3808000 00—00000 80302008 wctaoflscflz 00 38080 38008 00 0083.595 833 0080000
00580 00 300:— 0008 0.8 08.6 =08m 8880mm w: 0002 .38008 .bzsataoanm .08m 080 000—0w=0>
3338000
00 0002 .00 w80800 .00 000—803: 00000
:83 0800—00 0: 08 88088 _080§m
A0002 .«0 w80800 30080803 0000000 03800800
0000000 3.8000800 :83 880003800 .8000 80800—030 80808802
020090 00: w80800 38025 0050808 000000 00000 0039008 Dag 8°ch
.380800 0wS0 000— 30080808 00800800 0088008 0003 .03208000 .«0 8800000 00802030000
0008008 30: 00—050 00 0003 08: .m.D 090— 2 8 00800—0800 38080 .0002 .00 w80800 080
00 0000—00 3038000 8 88088 .0800xm 0000.008 QAZ 2. <2 >80 I <2 38080 $80800 :08me 3005 82003—1
0.00:8.“
a0 08: A0v038._0> 1880—.
0w80=E 0:: 80800 0000000002 80000009 003880> 80.82—08.— \ 0.85.30

 

 

 

 

 

 

 

96

 

000 900,00 38080 000 38008 0080008
0000 080 8 0000 000ww00 88080

.2000 30: Q03“ 38008 00 0000—00
30.000000 :5 3000 30: 90.30 300080
00 0000—00 >_0>00w00 00 0000— 00:05
.003 0005 000 0895 05 38.
30c Q0000 38008 00 0000—00 >_0>0_000

 

 

 

 

 

 

 

 

 

000 A500 0000 30: @030 38080 00080000”:—
00 000030 >_0>00w00 00 8808 0000 808000080 0000 30— 00 0800 3000000000
0800 6008000080 0000 30— 8 >80 08: .0008 00:80 .5308 000—008 000 30008008
Q0000 38008 000 38080 500 00 0000—00 A00 00 00>v .8808 0000 0000 “5000008 00 3800—.
>_0>8000 00 3000008 38000 .3000 E 0800 Quwm 0000 E 000 @030 38000— 38000 .00000000 0008000 @005 0008:.
000 30— 0008 Q05— 38008 000 38080 war—000000008 30— 0003000 03000 A00 00 00>v :80 .0000 00000000000 000 808002
00 0000—00 >_0>0_000 00 m800000 9.3— 0089 002 000 0000000005 Q0030 3808m .080 800 $800000 Q02 . 000000005~
00000008 dag
00 w800000 a: 0000 0083 8 $0003 38008 w8000008
003. 38008 00 00 00000 000000 flag 8 8000000080 000
38008 00 00008 00 000 00080w0w00 >8>000 0>000000-00
Odd 083000000 >850 003— w00080 0000 003 00000008 >w0_08000 38080 .00
00 00 000.0 00003 >000 :5 H300 38008 8050000 00000000 0000000000 8000800000 90¢
00 00008 028000 0 000 w80000080 38008 00 00000000 0>000000=00 .900 0000000000
D0010 8000880 000 00000000 00 w800000 800000
0008080 8000880 000 00000800 00.800800 .>_0>0_000 0000008 000880>0w 008800
800008 =< .93 38008 w880_0x0 w88000000 Q0010 8300008088 0008000080 80008 000000000030 >800 000000000
5 030:9 .8008 088 20 0 000 00820 000 05 000 000 .0825 588000035 .000 5.8 900: 020035
80000080 003 000 0080008 80000000 800008 .0000000
000000 0000000005 3000000000 Q03: 00 88>: .00000 30800808 0000000
000080090 000—008 0005 0000000000 .>=000_w 0800 4000008 000 00 000000 .00 000800
000 0000000 30— 000 00000 0800808 0002000 0000.800 ”0080800 4000000 08 8 0082000000 000880002
30— .000000 .00 000800 30— 0080—08000 33000000 000080>0w 00 00088 600800350 D030
.00 000800 32 .0w000 Q0030 00000>0< 38080 .23 0002 Odd .0> 0000a 0800 Odd 500: 00000000
.500 000 39
0003000 300000
000>300
00 :00 3038000» 30.0000.
88080 000 800000 0000000002 000000009 00.08000» 00000000000 \ 00000.20

 

 

 

97

 

.0008 0000000 000008 005800

0>08 000 000000000 380.0 0008 8800
300800 008808 0 8800000 0080000 238
00 000000 0800 30000008080 00000.00
8800000 8080 00000000 8 00000000000

00 $3 30300800000 >8 0000 >80 000
008000 Q30 00000800000000 38008
00 0800080 00 00 0000000 >380:

00 008 .8000 8800000 000000000 0 >8 0000

 

 

 

 

 

 

 

 

 

000 00 8800000 38080 00.0 80800000000 000>300
.>=0000>0. 5000000 88308 .0080 30000000 00 00000 80880002
0 00 0000 00 000 880080 .0000> 00.0 00 300008008 883 00000000 8008000 .0000000 .00808 .00 80.008000.
0000 03.8 000 Q03— 38008 0008080800 000 .m.D 3 >380 000 .00000 00800808 .80080 80000000
00 0000000 80308000 38080 00.300080 Q03: 000 080000 00000 $0008.00 .8000. 308 300: 3000—2
000 00 000.00 00 0038 000000800 0002 +3: .00 >0>00m <2 >0>000 I <2 .080: 0000000 00:00 00 0000mm 000 0.000080
0000000 Q03: .00 003> 03000 00 0000—00
>3>0800 .>0_0>00 30008000 00 000030
>3>0800 000 008800 3000000 0.80:0
.00000 8000 3000000 00
0000—00 >_0>00000 00 00008005 30000>
00000 8000 3000000 00 0000—00 >_0>0_000 00008008
000 008800 300000 03000000800010 .000>0_080 030 30000>
0.80:0 080 >8 .00 00000000
00008008 30000> 803 ow 000 .00000008 0.80:0 .0080
0080000030 0>0800 0 0>08 008800 :000000 80000000 00000 >0000000 000080800
0300000000000 0.80:0 000 00000000 000 38080 0.80:0 000 000>080m
00000 0080 0000002000008 3000000000 >8 000 0000000 ov .0080 0000000 003> 00800080
00008008 30000> 8003 0080000030 .0000._.000 Q0020 030000008800 03000 00000080 .00000 8000 .00 3800—.
030000 0 008 00000 8000 3000000 00008 0.— 8 0002 0.80:0 3000000 .>:0>00 30.88000. 3003 0005
0000000800 000 >080
000000008 8000 30: 8 008 500008000800
0_0>_000
.00 00.0 30.800000» 300000.
8808.0 000 800000 0000000002 00000000: 00.800000» 80000000:— 3008004

 

 

98

 

.809»: 0:
:0: 3350:: Q03— 05005 8 33:03::
32385 33:: 0: :0: Enid

00:02:00 3:23: 30:00 «00:80:50

085 Om «00:00

0030 353::

9 038333 3050:: m0
0380808 0:00.“ Aiowouao
5 $00.85 0803800 m0
598:5 0053098 A30» «man
5 E08335: 5 $0202:
Odd _3>:0EQ03>00

E 90300:: Data :30

no 8.355 boamE wflogom
A033: .«0 598:5 E32:
30—38 «0200:: “30:00:99“

 

 

300$:me 003 03 0:05. .:0:0N_3:.§E 8003:: flag 0:30 05 E 08030.5 Qfim 385:0

9 0033: E0350: 03 0:08 8:: :0 320530005 flag 5:: 055 32053803 00:03m

3&0: 830% 60:0:098 Gag S753 30:33:53: 33:23:00 30: :0 “095:5 0:030:00 0>nub£§<

D03— AEBEQE 303 33:50:00 < .3 0:02 :0 00:0: 5 @5393 59:5: =08m 83: 0:85
0.33::

.«0 3:: 3333...; 3:30..

$5.53 6:3 38:00 0.53.5602 2.0.0595 00.335» 30—52—06:— \ 205.3.

 

 

 

 

 

 

 

99

 

00000 0 5 0.0030800 w50005 0500000 0 0000000 0000000 000. 000—0 000000w0005 0000000 m 00 00000005 5000 $055500

030—3000— 000 500000200 m50_00_ 000.00 00 :03 00 “0000000 000 3030000 0003000 0000—00 000 003 030—000 0005 030000005 000 00000009
02wa00 000 A8000 0000000 300 w500000 .00 _00w E00500 000 00 300 w500000v 00000000 03000000 00000 000 0000000000000 00 00000 .05—30000
2 000000 4000000000 00005000 00 02 30000500 0003 w500000000 00000 00500000 0000 00000000 000500 00.0. 005—005w 000 0000030000 .00w0000m

 

30300000000 000000000 00.0 00000000630 .m 00000900 03—50005 .0 0000000 0000 03000
@0000 .m 030—3000— bflfifiofl 000 000095 .N 0000000030 30— .00 00 30000000 .0 ”m500000000 00.0 00000000

 

00000000 .w A000 0000000008 000030000 .0. 00000 .0 0000000000 00530 .m 0300030 .0 00000me 000000w0002

2m $205000. .N 500000500 .0 ”0000000, w500000000 00.0 000000000 000 0.000000 _000>0m .00000> 00000000 30.000000.

00 550005 00 0000000 00003000 05 00.0 000500—030 300005 m5w00000 .00 00003000 0.0 0000 £032 w500000 300000000 00.0 00000000

0000 .m500000000 03 0000 000500000 0005 000M052 $500000 $000000 5 00w00=000 .000w00000 00000005 005—0000w 00008000000000 $88 00000000
000050090
0000 005 0030.0 Q03— 5 3050000000003 .000000000000025 003000 0.:0< .0> 000-0005 .m
00000000000

00 w500000000 00000000000 0000000 00000>005 00003 .0000005 000 .300000 3000050090

0030 00095 00000000 013000000 20 00 :03 00 0000— 000 0000—0002 000—002 00900 .0> 000—002 0000— .0

000000000

00 000000000005 05: 00 b00000000 :30 00003 5000000000 =0_m>0000000000 30a— .0> 003: .m

.00000500w00 0000>005

0000000000 00.0 00:00 00000 000 60500003 0050w00 0003 00005 00000305 0000003 .00000>005

0005000000 000 00:00 00000 000 00000500w00 000000000000 .0000>000_ 000500000< .0> 000000>m .N

0200000000005 0000000000 00 000 000000005 _05w0000 000000?

0000000000: 00 000—0 0000 00 00000>005 000000 0. 0000—00 0003 0000000000 35902 .0> 00000Q .—
0505 00 0000—00 000000000000w 000 000000000 0000000000000 00 $5500 02.0 000_0>0D

AS00105 Ego—w w5300w 50030 000 00000 .0500

 

 

 

 

I 00.50000 M53050 5 00000000w 353—00000. 00 30005 0000000 :00 000 0000000000 00000005 00000300 0003000 000E0m0002
30—05—000 .9.va 00 0000000 0030: 0000000 0000 005 00000000 50¢ .00 000000002» 00 M500500000 000m .00 950000.000 05000>m 00005

0000000 000009—000 00000005 $00,000 0005000 >00>0 000 0w0000>00 00000000— 0000 005.0 000002 000 w500000000 0520,.— 383 000000m
.000000

005050— 0000000 \ 0.00—00¢.

 

 

00000030 000000260 0000>000~ 000000 00 _000wm002 000 000% 0000 ”m 5000090V

m 099meme

100

 

 

88:80:33: :8: 8:8 :8 380 88888: 8 8:85 .m

 

 

8:58.88 83on 5:5 :83888 888— 0: 8:88:88 3082888.: .N 88: awe—08:8. 88:83— 8888832
03mm .8 #8828 3:88 8380 ._ .8 88838 Guam 283:5
“888: 388m :8 Dag 3888 :o: .8 8: 2: :83 88.: 3:38—08:88 .8 £308 2: 88:85 888 mm 8:: m32>88~ a. 38:38:
8888 Quad 3828 :8 32088:.
“Quad 3:888 w8:::ot:w 882 838 :0 88088 03 8888 n33: 5:: 8,8828 88.: 88:8: 883:8» :8 8088830 Soomv 2:3 Z
832828 8:: 88::6 38:88 5:: “80:38? 8: 88:8
85:3 2: .8830: 888580 8388882: :0 :88 80:58:82: 8838 on 9 88:8 85:3
2: 35 :88: 8 3:8 : .380 8:383: 8:888 883 8:3 88:3 828 «8:: 35 88883 <OH .888.“
2: 8.: 858 $285.88 8883 8:: 8:: 83 b3 8: 35 88883 >md 2: :8 8x58 38? 88> 88 >383— 8: I 8::
:8: :8 80:30.88 >mm 3838 80:80:08 <OH :8. 88:3 2: .88: 2: .8 8:8: 308880 :3 :H 8 8:38:38 38:88 8::88m
88:8 8888 A33: .8 3383: 3 8: 8: 880m 883—. 8 88888 :oEooD
A: 888:. 88.88 3:23: :8 8388 :82. .8808: 8:888:88 8:88:83 828:8: “888880 30:88:88: 838 :8 388832
88:38 £88.: 08:38.. 8:: 2 2: a: x8888. Dav: ‘8 8:: 8:8; 8: 2:88: 805:3 BE. 3 .8 8388 .8 83885 $83 88:0
.8888 32888 8 80:88:88 88 8 88888 888.:8 88$: :8 88:83
388m 8083 83388 8888 8:23:8— 88 :88 :8 82 82 83:88.8: 83 805:3 2F
8:88: :o::3 83> 32 68888898 88 83. 8 m8.“ .8 883:5 .m
83888 8888250 .v
38:88 8388 8:3: 5:: 83838: :8 88:8— 8.3:m .m
8:53:30 8 8088:38m .N
8:898 .8 88o>:88< .—
828: =:m
8032288: 38:8:3 5:: .MD 8:
88 o. 88:88 8888: 23: 8: o: 8:: 832 88:08 883:: :8 88:88 8 83888 8 flown—38 5:5 £52888
$5.5 8:88 8: 830388: ‘8 898:: 2: 5:: 8838888 888 98:88 83 58:05 .N :o :88: 38:22:05—
:83888 3:8 :8 8:8 :83 Data 8 3:81:88 8888:— ._ :888: :8 .85 8838 88889832
828 58 83.35. 83288. :8 A88
A888o>ow 3:88 8:: >8: 8 8888 832 :8

>333 88808:: 888 £888: 2: .8 8:3: 88:83 2: :88: 888.: 38>8 8808: 88:8 8.:

3888 :88: 8083.:

883—. .8330:

 

8:888 8:23:8— 8: :33 .v

 

 

888883 83858:: @883 .m 8088832
8:38: 283:3 8:382 .N 8:38:38: 888 8380
8:888 2: 8388m< .— 2: 8 m832 888838: 388m @0ch
_a::::n

8:68

 

288:0

 

\ 228.2

 

 

101

 

.8088 :0:0>0:8 5:05: :5 98m 8888 :0 :0_mm:8_: a 5:: 80:30:00 :055 0::
Age—83:0: 503:5 0:880 w88005 .8285 :0 0:08 0:: w805::0v 8:58:00 8:530: 80:5:800
:w8=::: 80:5,: :5 CE: .800 5803803 8.: 05 0:880 Dag w:Em:: 8:25: :0 5888:: a m: 802

>w208—00: 8 80858
53: 83038.8 ON 5
:03 5 0:80:05 5::88

 

 

 

 

.:08:8_: 8: 88:80:88 0:8 8:58:00 :0: :0 85:08: 0E~ Abfimtoog 0:5 803 88080888: :050 :5 :08: 8580: 858::—
50880: :088 0: 3:88:88. 85 8580: 580888.: 80:: me5: 858:8 (:0 8090 805:, 8 5:05: :0 :83 8:8: :5 88:5
588083 8:58 8580: 85:88 0:: :0 0:8 0:: :5 Dav: 88800 :0 538w 0:: 8888:: :055 0E. Odd 88:8 88:85 8:03 8:030:
88:80:30: 0558 80: :5 «:05: 88:5 0: 30:
8888:: 8:: :055 0:; 8:080:08 :08 5:: 358:8 0:08 :00:— 35083 05: 35:38:00 38:65
..w.0v 00588 :0: 8.88:8 003:8 :: .8058 808838: :5 350 .25: 80:: :0::>0:: 05 8035:”.—
828 .n
:88 882 .v
88:: 0552 .m
8805:. 85825 .N
«:85 08:83: .—
w:05>0::: 008880 .33
:80 8058538 8.05: 5: 8038008 302 .v
82:55": 58885 .m
8:828 :0 x88 592w 850:0 .N
:580: 5903 $8380 ._
80588 w::0::0m::0 :0: Em: m: w88: :33 .8050: v 3030M
.05: 538w 8:: .8058 Odd :5 838:8 8080352
:: 80:. 0: 83:55:00 882 :0 w885858 580:8 :0 b0: :0: 2:00 >05 I 580: 5 :08:8:: 8: 08:0 80:5 885: w80::0m::0 52m
.808: 0:08 8:50:00: 0:23 $3-8 08:: :5 800 858:8 :032 50 888880 5:: 805880 :0E:< 858:8 :0 8:88:85 883 880
.8888: 05 3508880 .m> 005:7.~ .m> n55: 8:0: :3 800:: 58:00: 0058080
8:80: 0:3: 808880 08:8 0: 5:08:88 8:88 83:58 8:8 88:5
8005:? :0 55 8:080:80 :0 0:0: 53 w8::08 0:08 808: 00:: 5:: 8508: 832808:
55:80:83 .m
dem :0 0::5: 25H .N
3588:: ._
05:80: 0:08 :0: Did :0 8:088:50 8: >55
_::::0:
88:8: «x0800 \ 805.3:

 

 

 

 

102

 

00000080 .000 505 w=0_0 £00080 88088 05 00 B008 0wsm w :0 080020 805:0 08:0

00:08 20000 $00080 w80800
.8 00m000 =0 :0 8008080 00:00:: ”08. 80>: 000080 .8 0008800 00:00 00 08:00 ”000080 80808 080.080
0 00 00w0:08 8: 8 86800 “000800 008 0005 00 0:0 8 00800 50803 080 00800—080 3:03:08 000.;

03208000 000800 8508008 380000003 00 :03 00
808 8808800 0:0 0008802 850805 6080080 300008008 0088— 0080580008 800—00: 60080008
w80080:c £088.80 080.05 000.08: “00808 00.880088 0005 8 0080580 08080008 0803‘

00000080 080
008050 3800 0. 8008800 8080 8888 880808080 000m 00 803 w 00 80055000 0:0 00w080>00 0:

8888300 8080 80.00 0000—005 .m 0880 0:850 0:808:70 .N 5.000% 88308 8000508— 2080085 .—
020 08.0

80500800 w80088 .0 880nm 0.02 .m 0880008
308.0000 00 8004 .N .8000 0030— 80800—050 000000 000 000: 0:0 88800800 080% 00000008 .0

 

 

 

 

000w 0580a 00800800 800808 8088802

. 0888: 8800030 880008 030—08000.

.0000 300:0 0:0 080800 008 00088 00800000 080 800000—08 _000>0m 0000800 cm .8 0:08 0 505 8080000”—

.380800: 35080 98000000 8 0000008 M88808 805 8 >55 80:00:00 05 00380 00 808000 85:0 08H. 8808080 80¢ $8080 3003 E08080
.008 8800888
008000 05 :0 00000 0000088080 00003008 0:0 ~08?» 8088 8008000 88.5000 0 0000900 805:0 0800
8000808000 0880 0: :00008080 00008008 0:8
008—8800 0 08:00: 0:0 8000808 808080—0800 505 800000 8 080> 0:80:0w 8000808 080008
000008 00 8800800008 0080—0800. 00000 0. 0.0008000 8 000008080

088$ 05 .0000 85 8 0083888 0050 880 80080000 0000080 09 :00 8000888 0:080:08< 3030M

0008030

.m88:08:0 .0080 000880 05 08080:

00 28 05 888.8000 8 30880 8 008000 8000808 00 00b 05 80830: 88080880 88:80 “00.0 8000 0208003 503 w:0_0 6003 0000 H

8 8000888 00 000800 8 0855 5:580: 800008 8000000000 08m :00 88080080 005 0:80 805:0 0.; 808808 000 008—0080 0:0 8w:080080

00800800
.0:u:00
00808.”— 80800 \ 805:0.

 

 

 

 

103

 

$833830 38 :o :83
AEmEztoamo .333 3.360: 56:62? :83 SEE: 56¢?on BE 3:0:5w3 45,—. $me .852“ 2F

60:05: 98va 38032.5 :5 3:359:00 mm :03
3 £5505 £233: :8: Emzogua a: 3: 3:25:00 05 :5 $8.35 59.3% 3 mmo:w:==3::
0:205 $5505 .3333? mm: :ozgoS: Eucsooaganm Ho» .Dfim :0 Eu 93.265

 

 

:05: >5 55 22: 65% 25: Fousooascanav was 93:: E 42:36.“: «o 833 gum—om .m :ouaflgwRu
5:969: E £888: 2me Mo 85:25: wEwE .N .mEE 18:55 :5 8330m—

353: mo mmo:o>uoo.cm ._ v5 Eozsoomgnm uwufl 2:5:8m mo

”mugging Ea 3:053: 023:: Guam .3 :85: E Comm 12:25 .3 338 3:52. Aawma

95385 9.23 8a 892: 3:5 35 3st Ba nasal—mam :80 8:03:35 633 88 a mom: 85.5 2: wfibhocg mo 56385 53% $88".
_a:::cn

mafia—Fm 28:80 \ 985.3.

 

 

 

104

 

 

50555::0 :0:—05:00:: :0 0:000 0:: :000x0 >05 >00:0:00::n
:0 0:000 0:: :0:: :005 :05 3::0:0:::: 0:: :0:: 003—30::
55:0 05:. 0:5,: 3:535: :0>0 :000 0: 000:0>:0>0:5 5
:0:0::0: $~Tv 0 >50 60:00:00 0: 0:05 8:050 30:33:05

055:0 50:: 30::0:
.30: :0 05: $00080 3:55:

 

 

 

 

000: 0 0:05? 0000 0:: 5 Q0530: .055: :0:w=00_5 .0550: 50:5: 3:80:05: 0:00»
:0:: 05355:: 0:05 £08m-— mm 0:0 05:: 3:53 b:0:0:: .00:0::0:x0 00:0:3 .005: 50: w5:m:0 300:0:3 5:055:90
.5: :0:00 0: :0:: :0:00:00 0: 0:05 00:0:3 30:33:05 :0:: .00::B0:00 3050:0530: :0>0 3:0:050 b5c0 :0 ::0 00:50:00.:
0:05 0 0:053 0000 0:: 5 0:05:50 50:0: :0:—E03 :0 0:0: 55:0: :0 33:05: H33: 50:0: :3 :0:—$03 0:005:00 5500: 33 :0
:030: $5 5 05:00: 00:0::0>0w :0:w=0m::: .=0:0>O .0505; 00:35:00 :0 5:005 £53: 0503:: :050: 005050 3:52. :83
50:55:05: 0::>0:: 553:0 590:: 3:5: 50:: 0:30: :0 595:: 6:00: 00:52 80003355: D03: 50003:. :00:::0m
555:5 05:05 0 :0 5000350 0 w55: :0::500:
0: m0: :0::050 3353505 :5: .5055 :0 :0 00w0:0 3:00 05
:55: 50:5: :0 23: 0: 5&0: 53:50:80 :000 :0:000:0::
0:0:0: w0_ :00: ON 0 00 50: 00 0:3: :05 :5 :0:: 03500::
505:0 0:5. 0:5: :0:w=0 :0:: 00:0: :0:00:w :0 :0: 0:5: :0:—:50 00w:
:0:m_30:5 5003:: 55:5 0:: w55Q .055: :0:w:_0 m0 :0 33585 ::0: 00:0:0m
0:0: 355.: 0500 0:: 0:3: 05:: :0:w:00:: 5083:: 55:5 35:85:05 5055502
:0 0: :0:: .500050 0:05 0:: ::0 0:05 050505: :0 :00 0:0 :0:: 0:05 005:5: :0::3: mm:— 300::
5:30:05 :0:0::0:: 503:0: 00:20:50: :5500: 00:0::0>0m :5: 05 :0: :0 :0::05: .050 05:0: -22 55:5 :3 303:5
0:: 50:: 35500: 0:: 00 :053:0:0:0:0 0: 5055352 5:: .0w0 5:: a30:55:: 05w:m 35305030 2305050 mi ::0 00:35:
30:05
0055.553
00505050 2:05:53 035:5» w:::0m 3:50:
35.50.59: 0: :0:0_0- 0:555: :50 :0:—:02 :0 05055559 050050 50:00.03: 3:055:

 

 

 

 

 

 

503000: 000:0:0::0:::< 00505050 30555 ”O 5::0::<

U VASE/5:?

105

 

 

 

 

 

 

 

 

 

535050 353505: 3:52.
0:50:00: 0:35.. 0:: 0850 5:00:85 :00000b50t005 ::050: A550: :0 0050050: 5055502
:05? 0:5: 000:: 30:5: 3505550 :0: 000: 5: 55030: $505050 :0000 55850:: 5:50:05: 00050: 0:w0:0::m
:35: 0 0035 3:00:85 :05? 05:: 000:: 800:5 30539: ::050: .0:0::::0 :0 :0:5:: 05050::0: :0:0:::00::00 Amoom:
90:05 :::0:: 3:5 50:: :05:::0:0: 00: ..:::05 0505050: 050 ::0 .050 5:: .05:0: :5 : 3050—0500:; :0 :0:0::0:: .3 :0 ::0—::0:
:50: 0: 05: :0>0 5055305:
0: 53503.: ...0::00: 050:0 :0: 000:: :0::0w0:
00:53:, ::0: w5::0:w :0 30:03:00 50:» :000 5555008 :0::
0:05 0: 505500 0 5: .:0::0:0: :0: 0: 50:55:05: :0 05:0:
03: :0 50:02 .45: :53 5:03: 0>:0m0: ::0 5005:50 0
00: 5:303: .30_:0::0> 5:500 0:050 3:260 ::0 .050 63:0 333:0
:00—:0:: :0:: 000: .0w0:0>0_: 00353.» 550 3:260 ::03 :00: 00::0w030 :5wo: .0030w030 353:0:0 00:0:0m
0: 5:005: 33.5 :0:. :0 55553:: 0:: 50:: 003:6: 550:: 50:05:05: .050: :0550 .0> 503:: 0>:0::0:55:<
00:0: 55:: 0:: :0 00:0E0>0w 0:03:00 0:: :0::3 0: 0050: :0 :0::::: 3:0: .0550? ::0: 005::50 :0 Amoom:
0:: 00 :0:0::.:0:5 0:0 :053 33:58: 0:58: 0: :00: 0:0 .0553: ::0: .w50::0>:0 .354: 0300.0. :0::0:0::: 05:: 35.53%
50:55:: 0 0: 0000526» 0050: 3:0:: 5:8. 50:: 03:53:: :00—:0:: :0:: 000: :0 0:0:m :0 :53: 5503:. .m.D ::0 50:00:52
3:00:35
:0550 3:0:?
0:5 55:5 .0:0:0_::00: :0 0550> :55: 3:50:
3M2 .00:0::0:x0 .050 0:5: 355 5055502
5:: $505050 :00000b50::005 350005 ::0 35:00 0:w0:0::m
...0::.:.: :::::: :0:: 3:0: 55::0: 00:05 550 0:50:80 .b50::005: 3:::00:0: .:0::Q :0 :5030: Amoom: 5503
50:53: :055: E: :0 2:050:50 :000 5:000:05. 95:00:00 35550: 5505005 05:00:50: 0:05 550 ::0 0:05:03:
055: .0550 0:05 550 350:: 0: :00: 00:05:00 :_:0:: 0:50:80 $505050 :000< 00:05:0::0: 3:535:05: .0:0:::0:m
30:05
00:0 ::0>0w:
8:05—0:50 ::05 :w=0 0.:0_:0> “530m 3550—.
0:0_::0.:::< 0: :0:0_0~— 55:5,: ::0 .5502 :0 05:55:03: 050050 550001 \0.:0:::<

 

 

106

 

00:3 :8 @0000:

0:5 £55000 >00— .«0 00:38:8—
.0wv0365_ 35.30090
$5823.00: _0>0: 55350::
3:085:30 .56 05

 

 

 

 

 

 

 

 

::0ocawu 0:: 0.20530 :05: :0 mo 0:6 .3080 2.200% 8:08:05 $355083:
ago—MESA.” 8: mm 0053:0030 (8 x005 0E. $05650 w:_=0m 5235062: 028 6:: “0300:: A380 8 00:8 00:0_0m
00:00:: 8 8:85:38 30:38: ::0 005883—00: .02“: mo 30:00:3an,“ £550: :8 w5=0m\mo_mmv 00:00 @3508 “:0:—0&5:
.006 ban—:80 :23 :00: mm $5 .5800qu 03032 a: £300: :00: .0500: 33:02.38 4:253 @5850 82me 32:3 AmwaS
Set 6003800 :009 00: :0053030 no 08:5: 05 00:0 028:0: :8 :00: 5233:0002 w5=0m :_ 000:8 001% :Om:0w:<
dong—Emmy 05 8 03:10:00 0:0 000.: 0050 :00E0EES 05 8
60:53:“ 09 2:05 00:00am:w_m 8303.8: 0:: .05 ::0—0 ::0
£5 :o ::098 go: 0: >05. 003:: 2 man ::0: wfisozam :05:
$2.2 b5 :5 £3 3 81:: Bow wfinegm :0:? £60m... @3005:
000085 3000 .05 30% >2: 0:30:50 00:09:26» 26250 no 3000 $5520 m::0>0_ 50:0ngva
05 00:00—00 :0: EEG .2 3.003058: :00: 02E 2:03 3000 :02 56:25 3:98:30— C0=QQ=m ::0 mogogom
0005 :2: :32? :05 mm = .880 5:00:ng .3 32060:: €8.00 wfiu00:mw:0 .bmxflmfion. S 31052:: 304 mo
00 :00: 0:0 ::0: £5: 00:35 05 02: 0030:? 0:30 0:: :0:.H 032.28% 00000 58:60:; .0000 5:03ng 60:09:00 10:52. 23:
:2060—0 bi :o 00.5: 05 :8 000:: 03 0:238:00 .395 52:3: 6:60:00 .«0 00:83::z go 880 00 550355 :0 .0 5.32
A3383: 3:52.
A551: N; we Sou 0::0>0: 0 :0 me. A : wfisomfi 003029, 5:26 .m> $03.6 ::0E0w0502
.anwscwfi 08E 0050003 33.60% «ESE—3:: damn 75:00 Mo 02:00 $5090.60 003—:50 we 0&0Ebm
SE EEEEE 05 00:: >05 mm .05 0.0: 085:0 05 swag: .mocmt208fi0 30: .2050 550983 085 Ago: .10
00:05:88: 9 «00.20: >_0:00£:w_0 B: n: ::0sawzmfiz .0w0E3 ABA 00 590:8: 05:0,: FEE wfixozfi. .m.D :0 :0:—335
A0203:
00:05:03
00:: E0200 «:0:. :w:: 03:25, w:_:0w _: FEE.
Swinging: 3 003—0% muEuEh ::0 60502 a: 35553.09 05:8:0 £03000“ 3.85.3.

 

 

107

APPENDIX D

Appendix D: SAS Proc NLMixed Code For Stage One Estimation

/*SAS code for censored ML estimation*/

PROC NLMIXED DATA=TEMP1 OPTCHECK;
/* USE RESULTS OF PROC MIXED FOR STARTING PARAMETERS*/

FARMS BETA1=? BETA2=? BETA3=? BETA4=? BETA5=? BETA6=? BETA7=?
BETA 8=?;

PI=2*ARSIN(1);

MU: ALPHA + A_I + BETA1*ASSETSPECIFICITY +
BETA2*IMPORTANCEOFLEARNING + BETA3*INTERACTION +
BETA4*MARKETUNCERTAINTY + BETA5*TECHNICALUNCERTAINTY +
BETA6*APPROPRIABILITY + BETA7*FIRMSIZE;

/*OBSERVED (OBS=l) RIGHT CENSORED (OBS=O)*/
IF OBS=l THEN LL=(l/(SQRT(2*PI*SIGSQE)))*EXP(-(IG-
MU)**2/(2*SIGSQE));

IF OBS=O THEN LL=PROBNORM((MU-IG)/SQRT(SIGSQE));
L=LOG<LL);

MODEL IG ~ GENERAL(L);

RANDOM A_I ~ NORMAL([O,SZA_I] SUBJECT=LGINCIDID OUT=RANDOM1;
PREDICT MU OUT=OUT;

RUN;

108

REFERENCES

 

Anderson, E. (1985). The salesperson as outside agent or employee: A transaction cost
analysis. Marketing Science, 4(3), 234-254.

Anderson, E. (1988). Strategic implications of Darwinian economics for selling
efficiency and choice of integrated or independent sales forces. Management
Science, 34(5), 599-618.

Anderson, E., & Gatignon,.H. (1986). Modes of foreign entry: A transaction cost analysis
and propositions. Journal of International Business Studies, 1 7(3), 1-26.

Argyres, N., & Bigelow, L. S. (In press). Do transaction costs matter for firm survival at
all stages of the industry lifecycle? Management Science.

Argyres, N., & Silverman, B. S. (2004). R&D, organization structure, and the
development of corporate technological knowledge. Strategic Management
Journal, 25(8-9), 929-958.

Audretsch, D. B., Menkveld, A. J ., & Thurik, A. R. (1996). The decision between internal
and external R&D. Journal of Institutional and Theoretical Economics, 152(3),
5 19-530.

Balakrishnan, S., & Wemerfelt, B. (1986). Technical change, competition and vertical
integration. Strategic Management Journal, 7(4), 347—359.

Bardhan, A. D. (2006). Managing globalization of R&D: Organizing for offshoring
innovation. Human Systems Management, 25(2), 103-114.

Barney, J. R, Edwards, F. L., & Ringleb, A. H. (1992). Organizational responses to legal
liability: Employee exposure to hazardous materials, vertical integration, and
small firm production. The Academy of Management Journal, 35(2), 328-349.

Barragan, S., Cappellino, C., Dempsey, N., & Rothenberg, S. (2003). A framework for
sourcing product development services. Supply Chain Management, 8(3/4), 271—
280.

Bayus, B. L. (1997). Speed-to-market and new product performance trade-offs. The
Journal of Product Innovation Management, 14(6), 485-497.

Bienstock, C. C., & Mentzer, J. R. (1999). An experimental investigation of the
outsourcing decision for motor carrier transportation. Transportation Journal,
39(1), 42.

Breen, R. (1996). Regression models: Censored, sample selected, or truncated data.
Thousand Oaks, CA: Sage Publications.

109

Brophy, C., Gibson, D. J ., Wayne, P. M., & Connolly, J. (2007). A modelling framework
for analysing the reproductive output of individual plants grown in monoculture.
Ecological Modelling, 207(2-4), 99-108.

Brouthers, K. D. (2002). Institutional, cultural and transaction cost influences on entry
mode choice and performance. Journal of International Business Studies, 33(2),
203-222.

Brouthers, K. D., Brouthers, L. E., & Werner, S. (2003). Transaction cost-enhanced entry
mode choices and firm performance. Strategic Management Journal, 24(12),
1239-1248.

Calantone, R. J ., & Stanko, M. A. (2007). Drivers of outsourced innovation: An
exploratory study. Journal of Product Innovation Management, 24(3), 230-241.

Cantwell, J ., & Santangelo, G. D. (1999). The frontier of international technology
networks: Sourcing abroad the most highly tacit capabilities. Information
Economics and Policy, 11(1), 101-123.

Carson, S. J. (2007). When to give up control of outsourced new product development.
Journal of Marketing, 71(1), 49-66.

Cavusgil, S. T., Calantone, R. J ., & Zhao, Y. (2003). Tacit knowledge transfer and firm
innovation capability. Journal of Business & Industrial Marketing, 18(1), 6-21.

Cesaroni, F. (2004). Technological outsourcing and product diversification: Do markets
for technology affect firms' strategies? Research Policy, 33(10), 1547-1564.

Chandy, R. K., Hopstaken, B., Narasimhan, O., & Prabhu, J. (2006). From invention to
innovation: Conversion ability in product development. Journal of Marketing
Research, 43(3), 494-508.

Chandy, R. K., & Tellis, G. J. (1998). Organizing for radical product innovation: The
overlooked role of willingness to cannibalize. Journal of Marketing Research,
35(4), 474-487.

Chatteiji, D. (1996). Accessing external sources of technology. Research Technology
Management, 39(2), 48-56.

Chatterji, D., & Manuel, T. A. (1993). Benefiting from external sources of technology.
Research Technology Management, 36(6), 21-26.

Chesbrough, H. W., & Teece, D. J. (1996). When is virtual virtuous? Organizing for
innovation. Harvard Business Review, 74(1), 65-71.

Chiesa, V., & Frattini, F. (2007). Exploring the differences in performance measurement
between research and development: Evidence from a multiple case study. R&D
Management, 37(4), 283-301.

110

Chiesa, V., Manzini, R., & Pizzumo, E. (2004). The extemalisation of R&D activities
and the growing market of product development services. R&D Management,
34(1), 65-75.

Coombs, R. (1996). Core competencies and the strategic management of R&D. R&D
Management, 26(4), 345-355.

Croisier, B. (1998). The governance of external research: Empirical test of some
transaction-cost related factors. R&D Management, 28(4), 289-298.

David, R. J ., & Han, S. K. (2004). A systematic assessment of the empirical support for
transaction cost economics. Strategic Management Journal, 25(1), 39-58.

Dushnitsky, G., & Lenox, M. J. (2005). When do firms undertake R&D by investing in
new ventures. Strategic Management Journal, 26(10), 947-965.

Dutta, S., Narasimhan, O., & Rajiv, S. (2005). Conceptualizing and measuring
capabilities: Methodology and empirical application. Strategic Management
Journal, 26(3), 277-285.

Engardio, P., & Einhom, B. (2005). Outsourcing Innovation; First came manufacturing.
Now companies are farming out R&D to cut costs and get new products to market
faster. Are they going too far? Business Week (March 21), 84-94.

Garcia, R., & Calantone, R. (2002). A critical look at technological innovation typology
and innovativeness terminology: a literature review. Journal of Product
Innovation Management, 19(2), 1 10-132.

Gatignon, H., & Anderson, E. (1988). The multinational corporation's degree of control
over foreign subsidiaries: An empirical test of a transaction cost explanation.
Journal of Law, Economics, & Organization, 4(2), 305-336.

Geyskens, I., Steenkamp, J. B., & Kumar, N. (2006). Make, ally, or buy? A meta-analysis
of transaction cost theory. Academy of Management Journal, 49(3), 519-543.

Griffin, A., & Page, A. L. (1996). PDMA success measurement project: Recommended
measures for product development success and failure. Journal of Product
Innovation Management, 13(6), 478-496.

Grilliches, Z. (1990). Patent statistics as economic indicators: A survey. Journal of
Economic Literature, 28(4), 1661-1707.

Hall, B., J affe, A., & Trajtenberg, M. (2001). The NBER patent citation data file:
Lessons, insights and methodological tools. NBER Working Paper 8498.

Hall, B., J affe, A., & Trajtenberg, M. (2006). The NBER US. patent citations data file.
Retrieved August, 2007, from hflp://www.nber.org/patents/

111

Hamilton, G. (1999). Property rights and transaction costs in marriage: Evidence from
prenuptial contracts. The Journal of Economic History, 59(1), 68-103.

Hauser, J ., Tellis, G. J ., & Griffin, A. (2006). Research on innovation: A review and
agenda for Marketing Science. Marketing Science, 25(6), 587-717.

Heide, J. B., & John, G. (1990). Alliances in industrial purchasing: The determinants of
joint action in buyer-supplier relationships. Journal of Marketing Research, 2 7,
24-36.

Heide, J. B., & Stump, R L. (1995). Performance implications of buyer-supplier
relationships in industrial markets - a transaction cost explanation. Journal of
Business Research, 32(1), 57-66.

Howells, J. (1999). Research and technology outsourcing and innovation systems: An
exploratory analysis. Industry and Innovation, 6(1), 111-129.

Howells, J ., James, A., & Malik, K. (2003). The sourcing of technological knowledge:
Distributed innovation processes and dynamic change. R&D Management, 33(4),
395-409.

Hultink, E. J ., & Robben, H. S. J. (1995). Measuring new product success: The difference
that time perspective makes. Journal of Product Innovation Management, 12(5),
392.

John, G. (1984). An empirical investigation of some antecedents of opportunism in a
marketing channel. Journal of Marketing Research, 21, 27 8-289.

John, G., & Weitz, B. A. (1988). Forward integration into distribution: An empirical test
of transaction cost analysis. Journal of Law, Economics, & Organization, 4(2),
337-355.

Kerr, W. R., & F u, S. (2006). The industry R&D survey: Patent database link project.
Center for Economic Studies Working Paper 06-28.

Kessler, E. H., Bierly, P. E., & Gopalakrishnan, S. (2000). Internal vs. external learning
in new product development: Effects on speed, costs and competitive advantage.
R&D Management, 30(3), 213-223.

Kim, Y. (2004). An analysis of the incentives to licensing in US. information
technology. Journal of American Academy of Business, 5(1), 31-36.

Lane, P. J ., & Lubatkin, M. (1998). Relative absorptive capacity and interorganizational
learning. Strategic Management Journal, 19(5), 461-477.

Leiblein, M. J. (2003). The choice of organizational governance form and performance:
Predictions from transaction cost, resource-based, and real options theories.
Journal of Management, 29(6), 937-961.

112

Leiblein, M. J ., & Miller, D. J. (2003). An empirical examination of transaction- and
firm-level influences on the vertical boundaries of the firm. Strategic
Management Journal, 24(9), 839-859.

Leiblein, M. J ., Reuer, J. J ., & Dalsace, F. (2002). Do make or buy decisions matter? The
influence of organizational governance on technological performance. Strategic
Management Journal, 23(9), 817-833.

Levy, D. T. (1985). The transactions cost approach to vertical integration: An empirical
examination. The Review of Economics and Statistics, 67(3), 438-445.

Lohtia, R., Brooks, C. M., & Krapfel, R. E. (1994). What constitutes a transaction-
specific asset? An examination of the dimensions and types. Journal of Business
Research, 30(3), 261-270.

Love, J. H., & Roper, S. (2002). Internal versus external R&D: A study of R&D choice
with sample selection. International Journal of the Economics of Business, 9(2),
239-255.

Love, J. H., & Roper, S. (2005). Economists‘ perceptions versus managers' decision: An
experiment in transaction-cost analysis. Cambridge Journal of Economics, 29(1),
19-36.

Luca, L. M. D., & Atuahene-Gima, K. (2007). Market knowledge dimensions and cross-
functional collaboration: Examining the different routes to product innovation
performance. Journal of Marketing, 71(1), 95.

Luke, D. A. (2004). Multilevel Modeling. Thousand Oaks: Sage Publications.

Macher, J. T., & Richman, B. D. (2006). Transaction cost economics: An assessment of
empirical research in the social sciences. Georgetown University Working Paper.

Mahoney, J. T. (1992). The choice of organizational form: Vertical financial ownership
versus other methods of vertical integration. Strategic Management Journal,
13(8), 559-584.

Masten, S. E., Meehan Jr, J. W., & Snyder, E. A. (1991). The costs of organization.
Journal of Law, Economics, & Organization, 7(1), 1-25.

Mody, A. (1993). Learning through alliances. Journal of Economic Behavior and
Organization, 20(2), 151-170.

Nagaraj an, A., & Mitchell, W. (1998). Evolutionary diffusion: Internal and external
methods used to acquire encompassing, complementary, and incremental
technological changes in the lithotripsy industry. Strategic Management Journal,
19(11), 1063-1077.

113

Nakamura, K., 8; Odagiri, H. (2005). R&D boundaries of the firm: An estimation of the
double-hurdle model on commissioned R&D, joint R&D, and licensing in Japan.
Economics of Innovation and New Technology, 14(7), 583-615.

Narula, R. (2001). Choosing between internal and non-intemal R&D activities: Some
technological and economic factors. Technology Analysis & Strategic
Management, 13(3), 365-387.

Neter, J ., Kutner, M. H., Nachtsheim, C. J ., & Wasserrnan, W. (1996). Applied Linear
Regression Models, 3rd Edition. Chicago: Irwin.

Nicholls-Nixon, C. L., & Woo, C. Y. (2003). Technology sourcing and output of
established firms in a regime of encompassing technological change. Strategic
Management Journal, 24(7), 651-666.

Nickerson, J. A., & Silverman, B. S. (2003). Why firms want to organize efficiently and
what keeps them from doing so: Inappropriate governance, performance, and
adaptation in a deregulated industry. Administrative Science Quarterly, 48(3),
43 3—465. .

Noordewier, T. G., John, G., & Nevin, J. R. (1990). Performance outcomes of purchasing
arrangements in industrial buyer-vendor relationships. Journal of Marketing,
54(4), 80-93.

Odagiri, H. (2003). Transaction costs and capabilities as determinants of the R&D
boundaries of the firm: A case study of the ten largest pharmaceutical firms in
Japan. Managerial and Decision Economics, 24(2-3), 187-211.

Olson, E. M., Slater, S. F ., & Hult, G. T. M. (2005). The performance implications of fit
among business strategy, marketing organization structure, and strategic behavior.
Journal of Marketing, 69(3), 49-65.

Penner-Hahn, J ., & Shaver, J. M. (2005). Does international research and development
increase patent output? An analysis of Japanese pharmaceutical firms. Strategic
Management Journal, 26(2), 121-140.

Piachaud, B. (2005). Outsourcing technology. Research Technology Management, 48(3),
40-46.

Pisano, G. P. (1990). The R&D boundaries of the firm: An empirical analysis.
Administrative Science Quarterly, 35(1), 153-176.

Powell, W. W., Koput, K. W., & Smith-Doerr, L. (1996). Interorganizational
collaboration and the locus of innovation: Networks of learning in biotechnology.
Administrative Science Quarterly, 41(1), 116-145.

Quinn, J. B. (2000). Outsourcing innovation: The new engine of growth. Sloan
Management Review, 41(4), 13-28.

114

Raz, T., Shenhar, A. J ., & Dvir, D. (2002). Risk management, project success, and
technological uncertainty. R&D Management, 32(2), 101-109.

Rindfleisch, A., & Heide, J. B. (1997). Transaction cost analysis: Past, present, and future
applications. Journal of Marketing, 61(4), 30-54.

Robertson, T. S., & Gatignon, H. (1998). Technology development mode: A transaction
cost conceptualization. Strategic Management Journal, 19(6), 515-531.

Rosenkopf, L., & Nerkar, A. (2001). Beyond local search: Boundary-spanning,
exploration, and impact in the optical disk industry. Strategic Management
Journal, 22(4), 287-306.

Rothaermel, F. T., Hitt, M. A., & Jobe, L. A. (2006). Balancing vertical integration and
strategic outsourcing: effects on product portfolio, product success, and firm
performance. Strategic Management Journal, 27(1 1), 1033-1056.

Sampson, R. C. (2004). The cost of misaligned governance in R&D alliances. Journal of
Law, Economics, and Organization, 20(2), 484-526.

Shaver, J. M. (1998). Accounting for endogeneity when assessing strategy performance:
Does entry mode choice affect FDI survival? Management Science, 44(4), 571-
585.

Silverman, B. S., Nickerson, J. A., & Freeman, J. (1997). Profitability, transactional
alignment, and organizational mortality in the US trucking industry. Strategic
Management Journal, 18(Summer Special Issue), 31-52.

Sorescu, A. B., Chandy, R. K., & Prabhu, J. C. (2003). Sources and financial
consequences of radical innovation: Insights from pharmaceuticals. Journal of
Marketing, 67(4), 82-101.

Steensma, H. K., & Corley, K. G. (2000). On the performance of technology-sourcing
partnerships: The interaction between partner interdependence and technology
attributes. Academy of Management Journal, 43(6), 1045-1067.

Steensma, H. K., & Corley, K. G. (2001). Organizational context as a moderator of
theories on firm boundaries for technology sourcing. Academy of Management
Journal, 44(2), 271-291.

Swan, K. S., & Allred, B. B. (2003). A product and process model of the technology-
sourcing decision. Journal of Product Innovation Management, 20(6), 485-496.

Tapon, F. (1989). A transaction costs analysis of innovations in the organization of
pharmaceutical R&D. Journal of Economic Behavior and Organization, 12(2),
1 97-21 3.

115

Teece, D. J. (1986). Profiting from technological innovation: Implications for integration,
collaboration, licensing and public policy. Research Policy, 15(6), 285-305.

Thiébaut, R., & J acqmin-Gadda, H. (2004). Mixed models for longitudinal left-censored
repeated measures. Computer Methods and Programs in Biomedicine, 74(3), 25 5-
260.

Toubia, O. (2006). Idea generation, creativity, and incentives. Marketing Science, 25(5),
411.

Trajtenberg, M. (1990). A penny for your quotes: Patent citations and the value of
innovations. The RAND Journal of Economics, 21(1), 172-187.

Trajtenberg, M., Henderson, R., & J affe, A. (1992). Ivory tower versus corporate lab: An
empirical study of basic research and appropriability. NBER Working Paper 4146.

Ulset, S. (1996). R&D outsourcing and contractual governance: An empirical study of
commercial R&D projects. Journal of Economic Behavior & Organization, 30(1),
63-82.

Veall, M. R., & Zimmermann, K. F. (1996). Pseudo-R2 measures for some common
limited dependent variable models. Journal of Economic Surveys, 10(3), 241-259.

Veugelers, R. (1997). Internal R&D expenditures and external technology sourcing.
Research Policy, 26(3), 303-316.

Veugelers, R., & Cassiman, B. (1999). Make and buy in innovation strategies: evidence
from Belgian manufacturing firms. Research Policy, 28(1), 63-80.

von Hippel, E. (1986). Lead users: a source of novel product concepts. Management
Science, 32(7), 791-805.

von Hippel, E. (2005). Democratizing Innovation. Cambridge MA: MIT Press.

Walker, G., & Weber, D. (1984). A transaction cost approach to make-or—buy decisions.
Administrative Science Quarterly, 29(3), 373-391.

Wathne, K., & Heide, J. B. (2000). Opportunism in interfirm relationships: Forms,
outcomes and solutions. Journal of Marketing, 64, 36-51.

Williamson, O. E. (1975). Markets and Hierarchies: Analysis and Antitrust Implications.
New York: The Free Press.

Williamson, 0. E. (1985). The Economic Institutions of Capitalism: Firms, Markets,
Relational Contracting. New York: The Free Press.

Williamson, 0. E. (1991). Comparative economic organization: The analysis of discrete
structural alternatives. Administrative Science Quarterly, 36(2), 269-296.

116

Williamson, 0. E. (1996). The Mechanisms of Governance. Oxford: Oxford University
Press.

Wolfe, R. M. (1999). Research and Development in Industry: 1995-1996. Arlington VA:
National Science Foundation.

Wolfe, R. M. (2005). Research and Development in Industry: 2001. Arlington VA:
National Science Foundation.

Zhao, H., Luo, Y., & Suh, T. (2004). Transaction cost determinants and ownership-based
entry mode choice: a meta-analytical review. Journal of International Business
Studies, 35(6), 524-544.

117

 

 
 

11111111111111le

02956 66

    

11111111121911le