You are on page 1of 18

Published online ahead of print May 14, 2009


Articles in Advance, pp. 1–18 doi 10.1287/mnsc.1090.1010

issn 0025-1909  eissn 1526-5501 © 2009 INFORMS
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

Effort, Revenue, and Cost Sharing Mechanisms

for Collaborative New Product Development
Sreekumar R. Bhaskaran
Cox School of Business, Southern Methodist University, Dallas, Texas 75205,
V. Krishnan
Rady School of Management, University of California, San Diego, La Jolla, California 92093,

T he growing sophistication of component technologies and the rising costs and uncertainties of developing
and launching new products require firms to collaborate in the development of new products. However, the
management of new product development that occurs jointly between firms presents a new set of challenges
in sharing the costs and benefits of innovation. Although collaboration enables each firm to focus on what it
does best, it also introduces new issues associated with the alignment of decisions and incentives that have
to be managed alongside conventional performance and timing uncertainties of new product development.
In this paper, we conceptualize and formulate the joint development of products involving two firms with
differing development capabilities and examine the implications of arrangements that go beyond sharing of
revenues to include sharing of development cost and work. We term these approaches that involve sharing of the
development cost and sharing of the development work investment sharing and innovation sharing, respectively.
These cost and effort sharing mechanisms have subtle interactions with the degree to which revenues are shared
between firms and the type of development project under consideration. Our analysis shows that investment
and innovation sharing are particularly relevant for products with no preexisting revenues, and their benefits
also depend on the degree to which revenues are shared between the firms. Whereas investment sharing is more
attractive for new-to-the-world product projects with significant timing uncertainty, innovation sharing plays an
important role in environments where projects experience product quality uncertainty, firms are similar in their
capabilities, and the costs of integration of work across firms can be controlled. Our key contribution involves
the modeling of joint work and decision making between collaborating firms and unearthing the complementary
role of revenue, cost, and innovative effort sharing mechanisms for new product development. We translate our
analytical findings into a managerial framework and illustrate the results with examples from the life-sciences
and electronics industries.
Key words: codevelopment; project alliances; product and supply chain innovation; technology uncertainty
History: Received January 11, 2006; accepted January 15, 2009, by Christoph Loch, R&D and product
development. Published online in Articles in Advance.

1. Introduction a strong patent portfolio and research budgets such

New product innovation has long been a key avenue as IBM have begun forging joint-development agree-
for revenue and profit growth, especially in indus- ments to meet the challenges of contemporary prod-
tries such as life sciences and high technology. With ucts and markets (Austin American Statesman 2003).
lifetimes of products shrinking, technical complexity The increasing uncertainty, complexity of know-
increasing, and daunting odds of success a norm, how, and costs of product development and distribu-
firms are forced to invest continually and ever- tion require firms to pool their resources and enter
growing amounts to maintain their competitive edge. into joint-development contracts involving the sharing
In response, some firms have looked beyond their four of product revenues, development costs, and research
walls to manage the costs and risks of new product and development (R&D) effort between industrial
development (Quinn 2000). Pharmaceutical compa- customers and suppliers, which forms the focus of
nies increasingly turn to each other and partner with this paper. Although intercompany alliance issues
specialist firms to develop new compounds that cost such as the melding of cultures have received a fair
nearly a billion dollars to launch (Grover 1998). Arti- amount of research attention in the strategic manage-
cles trace HP’s development of its highly profitable ment and organizational behavior literature (reviewed
printer business to joint work with suppliers during in the next section), the operational challenges of
different phases of its business. Even companies with joint development are equally complex but relatively
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
2 Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS

understudied. Unlike manufacturing and assembly, the development work at Alpha rather than share
where activities can sometimes be outsourced in an the development and testing work. Alpha developers
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

arms-length transfer of prices and quantities, product kept detailed records of their time spent in the project,
development is an uncertain and long-lead-time activ- and Alpha’s alliance manager would report the full-
ity with advance investments and joint work between time equivalent of scientist/developer time invested
firms that requires coordination about product fea- in the project every quarter, upon which Mega would
tures and qualities. In consolidating and converging reimburse Alpha for the expense. We call this pat-
industries such as high technology, firms may also tern of multifirm product development where the col-
fear that their partners may appropriate a large por- laborating companies share the development expense
tion of the benefits. The classic product development with one firm doing the bulk of the work, investment
challenges of managing technology and market uncer- sharing. This pattern has been observed with several
tainty are coupled with agency issues stemming from other firms in the industry as illustrated also in the
opportunistic behavior of development partners. recent example of Actelion and Glaxo (Whalen 2008).
In this paper, we focus on a specific form of joint Consider, on the other hand, product development
development between two firms that consider sharing at the firm Dell. Although Dell is known widely as
the product revenues, development costs, and devel- a direct and lean distributor of computers, its ability
opment effort. The product being developed is based to leverage the technological capabilities of its com-
on an uncertain underlying technology/science, and ponent suppliers is also noteworthy. Dell’s product
the firms could differ in their development capabili- design processes are deliberately designed to support
ties as well as their power in the value chain. We now and complement supplier innovation with strengths in
discuss the nuances of joint development in two of customer needs identification, complementary hard-
ware/software development, component qualification
the industries that we studied.
and testing, system integration, beta testing, and mar-
1.1. Managerial Issues in Collaborative New ket feedback. For example, Dell worked closely with
Product Development Lexmark in 2002 and enhanced Lexmark’s printer
Product development plays a crucial role in the bio- technology with an innovative Dell-developed car-
pharmaceutical industry, which involves the appli- tridge replenishment software, and both firms shared
cation of biotechnology research to the development the revenues (Financial Times 2003). Dell has also built
its other product lines such as laptops, servers, and
of pharmaceutical compounds. Ever since the inven-
storage products through joint-development arrange-
tion of recombinant insulin in 1982, the application of
ments with supply chain partners, in which Dell
biotechnology to the creation of therapeutic drugs has
shares downstream development and testing work.
grown dramatically. Due to the know-how required in
Thomke et al. (1999) discuss how Dell worked with
R&D and the large investments needed for commer-
Sony to bring to market Sony’s lithium ion battery
cialization, small biotechnology companies often work
with large pharmaceutical companies. To understand
Dell’s joint-development approach is in contrast to
codevelopment in this industry, we surveyed a col-
the investment sharing approach used by Mega in
lection of articles and conducted a detailed interview- that it partakes in the development work in areas
driven field study of joint development between two of qualification, testing, and system integration. We
firms, which we will call Alpha Labs (a small public refer to such an approach in which the develop-
U.S. biotech company) and Mega Pharmaceuticals (a ment work is shared without an upfront transfer of
large Fortune 100 pharmaceutical company). money, innovation sharing. Although both investment
In September 2002, Alpha and Mega entered into an and innovation sharing approaches seem to be sig-
agreement to work on the development and commer- nals to partners about a firm’s commitment, the ques-
cialization of a new innovative class of diabetes drugs. tion arises as to what effect they have on the extent
Under the terms of this negotiated agreement, Alpha of product development and profits. What impact
and Mega agreed to share equally in the development does development uncertainty have on these cost, rev-
investment, with Mega making an upfront investment enue, and effort sharing agreements between joint-
based on an estimate of the development costs. Alpha development partners? How should firms split the
performed the drug development, while Mega helped value that is created by their collaborative efforts?
to commercialize and distribute the product with its What role do their bargaining power and outside
vast global salesforce, and in return the firms agreed options play in the structure and execution of the
to share the revenues (30% of U.S. domestic revenues joint-development process? These are some of the
and 80% of international revenues accrued to Mega, questions that constitute our focus in this paper.
and the remainder went to Alpha). It is noteworthy We seek to model the interaction between two firms
that despite its significant development and testing who embark on an opportunity to invest in a devel-
capabilities and infrastructure, Mega decided to fund opment project that could improve the performance
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS 3

quality of a product but faces uncertainties related to Weiss 1997, Amaldoss et al. 2000), or limits itself to
underlying technology which discourage investment. institutional alliances and the challenge of integrat-
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

In addition, the firms may have reservations about ing cultures in such ventures (Doz and Hamel 1998).
the opportunistic behavior of their partners, who may Despite the challenges of contractual joint devel-
appropriate the gains from the innovation. How then opment, many products do not warrant the huge
can a firm address these reservations and achieve investments in time and money required for institu-
maximal quality improvement and profits? The two- tional alliances, such as joint ventures. Management
firm decision-making model and the revenue, effort, of institutional alliances often requires the establish-
and cost sharing mechanisms proposed in this paper ment of a separate organization with investments
are intended to help firms manage collaborative prod- of hundreds of millions of dollars as well as the
uct development under such conditions of technology, blending of established organizational cultures (Doz
market, and partner uncertainty. and Hamel 1998). Although there are a few high-
The remainder of this paper is organized as follows. profile success stories of joint ventures, their mixed
After reviewing the related work in §2, we present record (as evidenced by examples such as the Iridium
a model of two-firm product development in §3. We alliance or the IBM-Motorola RISC technology ven-
methodically analyze the different mechanisms for ture) and their high investments make firms pause
coordinating multifirm product development in §4, (before spending large amounts of money in institu-
identifying key properties and evaluating their ability tional alliances) and look for alternative approaches to
to deal with different types of development projects. joint development. Contractual joint development—
In §§5 and 6, we develop the main results of this paper, that begins with the recognition that a supplier that
comparing and contrasting the different mechanisms invests in component technological innovation cannot
and demarcating the domains of appropriateness of take for granted the possibility of both the develop-
the revenue, cost, and effort sharing mechanisms. ment succeeding and other value-chain participants
Finally, in §7, we summarize the managerial implica- making mutually aligned decisions if the innovation
tions using a conceptual framework, relate to exam- materializes—seeks to achieve the benefits of collab-
ples presented above, and point to directions for oration without the significant fixed costs of institu-
future work. tional alliances (Gerwin and Ferris 2004). Customer
firms that wish to tap into the technological know-
how of their suppliers have to develop mechanisms
2. Relevant Literature to address these supplier concerns, stimulate inno-
New product development has been a topic of sig- vation, and thereby realize the full potential of the
nificant research interest in the operations literature collaborative development process. Gerwin and Ferris
in recent years, but most of the literature on prod- (2004) provide a qualitative discussion of a taxonomy
uct development is single-firm-centric with its focus of alliances in product development projects, high-
at the level of a project or, at best, a product line. More lighting the role of contractual alliances. Dyer (2003)
recently, there has been an emerging stream of work paints a rich portrait of the weaknesses of both ver-
on the interactions between product and supply chain tical integration and arms-length supplier manage-
design decisions (Fine and Whitney 1996, Anderson ment using examples from Chrysler and Toyota, but
and Parker 2002, Novak and Eppinger 2001, Ulrich focuses more broadly on the issue of trust than on the
and Ellison 2005, Grahovac and Parker 2002). Even specifics of sharing effort and development.
this literature focuses only on how a single firm Several papers in economics, marketing, and supply
should make decisions involving its suppliers and not chain operations that study the interaction between
on the interaction between the decisions of firms. Erat vertical firms have proposed mechanisms to alle-
and Kavadias (2006) study the development of prod- viate tactical (price-quantity) coordination problems
ucts in an industrial context, but their focus is on (Jeuland and Shugan 1983, Lee and Staelin 1997,
intertemporal discrimination of a technology supplier McGuire and Staelin 1983, Aghion and Tirole 1994,
through partial adoption of technology and not on the Grossman and Hart 1986, Cachon and Lariviere 2005,
joint development of products by multiple firms. Our Spencer and Brander 1992, Cvsa and Gilbert 2002).
work builds on the model of product development Attempts have also been made to use similar mod-
under technology uncertainty from the new prod- els to analyze the effect of innovation of one of the
uct operations literature (Loch and Terwiesch 1998), firms on its channel partners. Gupta and Loulou (1998)
but breaks new ground in the area of joint product study how interactions between firms in a channel
development. affect innovation. Gilbert and Cvsa (2003) analyze the
The technology management literature considers effect of strategic commitment to price by a supplier
institutional alliances but ignores the role of uncer- to stimulate downstream innovation in a supply chain.
tainty in development decision making (Dutta and However, this stream of literature deals mostly with
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
4 Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS

process innovations and tactical decisions like prices to achieve this quality improvement. In addition,
and quantities, and ignores the effect of technologi- when the development effort is distributed between
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

cal uncertainty on firms’ decision making. Our paper, the firms, integration costs are incurred as discussed
focused on collaborative product innovation, borrows later in this section.
some modeling elements from this stream, but dif- The fixed cost of development is an upfront invest-
fers significantly in that it emphasizes uncertainty of ment and is a function of the level of quality improve-
technologies and goes beyond interfirm coordination ment .1 We model that this cost is of the form I  ,
based on prices and quantities. We evaluate product where I is an investment parameter. We assume
development decision making that focuses on product that  > 1, implying that the cost of innovation is
quality improvements through cost or effort sharing convex with respect to (w.r.t.) . Convexity is not a
arrangements between firms and examine how firms driving factor for much of our analysis (specifically
should incorporate the underlying technological and not impacting revenue sharing and investment shar-
market parameters of an industry while entering into ing), and it does have empirical backing in literature;
partnerships. see, for example, Cohen and Klepper (1992, 1996) and
Kamien and Schwartz (1992). Convex costs are often
3. The Model attributed to diminishing returns from R&D expendi-
In this section, we model the collaborative prod- tures or to diseconomies of scale that, in practice, can
uct development process between two firms that be linked to bureaucracies in a larger firm that stifle
must decide about sharing the revenues and costs creativity and impede innovation. Also, as the level
of an innovation. This innovation along a certain of technological innovation increases, diseconomies in
technological dimension could improve the perfor- offering employment contracts could also lead to dis-
mance quality of the focal product and has the economies of scale in R&D (Zenger 1994). Throughout
potential to enhance end-customer demand. How- the body of this paper, we assume that the convexity
ever, the innovation itself is fraught with risks and parameter  = 2. However, directionality of the results
requires investments in time and money. We model is preserved for any  > 1.
the development/innovation process as comprising In addition to the fixed development cost, a firm
of two distinct phases—one involving the develop- would also have to dedicate resources to the tech-
ment of a key underlying technology, and the second nology development process. To accommodate the
involving the packaging and integration of this tech- fact that higher levels of quality improvement require
nology into a new product/system—handled by two more resources, we assume that these costs are pro-
independent firms; a focal firm (FF) and a partner firm portional to the level of quality improvement .
(PF). Note that the complementary nature of these Also, because the firm would have to dedicate these
activities and specialized skills that these activities resources throughout the course of the development
entail imply that the two firms have to come together process, we assume that these costs are proportional
to partner in development to realize the value of to the development time. Specifically, we model these
the innovation. To be general, we have chosen neu- costs as ct, where c is a constant and t is the total
tral terminology, although in specific situations (such time taken for technology and product development.
as the examples discussed earlier), one of the firms Thus, the total fixed and variable costs of develop-
may have an expertise in certain aspects of technol- ment (not including integration costs) for a firm that
ogy/product development. The two firms could share improves the quality by  would be
the revenues as well as the development expense or
the effort, which we refer to as investment and inno- T C = I 2 + ct (1)
vation sharing, respectively.
Investments in quality improvement enable the firm
3.1. Model of Innovation to command higher margins and increase profitability,
The innovation we model is such that the improve- which we capture by assuming that there is a one-to-
ment in quality does not increase a firm’s marginal one correspondence between the quality of the final
costs. Abbott (1953) terms such improvements “inno- product and the subsequent revenues that firms are
vational quality dimensions”—the introduction of a able to generate in the end-product market. In partic-
novel quality that is judged superior by most or all ular, we model that an innovation level  generates a
buyers and costs no more to produce, thus mak- total additional value of V = r. If the residual value
ing the older quality obsolete. However, the innova- of the product before the innovation is , then the
tion entails costs: a firm decides the level of quality
improvement (innovation)  that it wants to achieve, 1
Because there is a one-to-one correspondence between the level
and it incurs both a fixed cost of development and a of innovation and the level of quality improvement, we use both
variable time-dependent cost of resource deployment these terms interchangeably.
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS 5

total value after the innovation is successful can be Figure 1 Decision Timeline of the Codevelopment Process
represented as T V =  + r. When  = 0, the final
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

value of the product depends entirely on the inno- Focal firm decides the mode of
vation, and we refer to these projects as new-revenue
projects. In contrast, when  > 0, the value of the
project does not depend entirely on the innovation Firms determine cost/revenue sharing terms
through Nash bargaining
that is undertaken. Such projects can be considered as
replacement-revenue projects.
Optimal investment levels are determined
3.2. Model of Interfirm Interaction and undertaken
Firms can engage in joint development in numer-
ous ways. To begin making a contribution in mod- Technology uncertainty resolved and project
eling joint development, we start with specific forms value realized
of revenue and cost sharing between firms inspired
by industrial examples presented earlier. To main-
tain focus on how product development issues influ- to use the bargaining structure proposed by Nash
ence the collaboration between firms, we consider a (1950, 1953) to determine the optimal contract terms.
baseline case in which the focal and partner firms In a Nash bargaining game, two players cooperatively
are currently in a revenue sharing agreement wherein decide to how to split the surplus that occurs as a
the focal firm gets a fraction  of the total rev- result of their interaction. The manner in which the
enue (which can be decided endogenously or can be surplus is split depends both on the utility functions
an exogenous parameter), whereas the partner firm of both players as well as the value of their out-
receives (1 − ) of the revenue. Revenue sharing has side option (the value they are able to obtain if they
been shown to be a useful mechanism for coordi- choose not bargain with each other). Please refer to
nation in the supply chain literature that deals pri- Owen (1995) for more details of the bargaining pro-
marily with postlaunch price-quantity decisions; our cess. Initially, we assume that the outside option of
interest here is in understanding how it combines both firms are the same and, without loss of gener-
with prelaunch cost sharing and influences quality ality, normalize this value to zero. Subsequently, we
and development investment decisions. In addition, relax this assumption and numerically evaluate the
to a pure revenue sharing agreement, firms can also effect of different outside options for FF and PF on
enter into an investment sharing or an innovation shar- the optimal mode of codevelopment.
ing agreement. Under investment sharing, firms share The sequence of decision making, represented in
the cost of product development, whereas under inno- Figure 1, is as follows. At first, FF decides which of the
vation sharing, they share the innovation itself, with codevelopment mechanisms should be used to con-
part of the work being done by each of the firms. duct the development project. The firms then coop-
At first, we analyze the case in which the revenue eratively determine how the revenues and costs of
sharing parameter  is an exogenous parameter, the innovation should be shared. Subsequently, the
which could be the case in mature markets where investment for the development project is made, the
preexisting relationships or channel power equations value of which is realized at the end of the develop-
could form the basis of how revenues are shared ment process once technology uncertainty is resolved.
between collaborating firms. In contrast, negotiation To understand how each of these agreements
between innovating firms could form the basis of impacts the technology development process and
such division of revenues in emerging markets. To a firm’s costs and revenues, let us examine these
accommodate such markets, we also consider the case agreements more closely. When firms enter into an
in which the equilibrium revenue sharing parame- investment sharing agreement, one firm conducts the
ter  is determined as a function of the bargain- development work while the other agrees to bear part
ing between collaborating firms. Subsequently, we of the development costs. Let k be the fraction of
examine the impact of this bargaining framework development costs borne by the focal firm. Similar
on pure revenue, investment, and innovation sharing to pure revenue sharing, the technology development
agreements. The complementary roles of FF and PF itself could be conducted by the focal or partner firm,
imply that the firms have to work together to realize which we analyze in detail in the next section.
the value of the project. The central questions then Innovation sharing entails the splitting of the devel-
become what mechanism should be used to opera- opment work across both firms, so its form and func-
tionalize codevelopment and how the value realized tion differ in a subtle fashion from investment sharing.
from codevelopment should be shared between firms. Specifically, a part of the development work F is done
In all of these analyses, we assume that firms agree by the focal firm (who incurs a fixed cost IF2 ), and a
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
6 Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS

part of the development work P is conducted by the whether that incremental investment would trans-
partner firm (who incurs the fixed cost IP2 ). Note that late into an innovative product and higher margins
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

convexity of the fixed development costs allows us to is uncertain. As v increases, a firm faces less uncer-
capture the effect that the division of innovation across tainty about its ability to create a capable product.
the two firms could lead to lower development costs We refer to this uncertainty as translational uncertainty
compared with the case in which this innovation were because it represents a firm’s ability to translate an
to be done in a single firm (which is what we would idea into a commercially viable product. High trans-
expect, because such division of work would enable lational uncertainty is characteristic of new-to-the-
firms to do what they do best). However, innovation market product projects (such as the launch of the
sharing entails distributed development, which could improved version of a product).
result in firms having to incur additional integration In addition, the time required for technology devel-
costs to merge quality improvements achieved across opment is uncertain and depends on the firm capa-
firms. Realistically, these costs would depend on the bility as well as product quality, which we refer
level of innovation that is undertaken by each of the to as timing uncertainty and model, following prior
firms. Specifically, when FF and PF invest F and P , research, as an exponentially distributed random vari-
respectively, we consider several models of integration able with a probability density function
t .
costs, the simplest of which has the following func- Timing uncertainty is characteristic of new-to-the-
tional form: world product projects (such as the new category
of diabetes drugs discussed earlier in this paper).
CF P  = KF2 + P2  (2) Note that the characterization of timing uncertainty
using an exponential distribution allows us to cap-
which implies that the integration cost is increasing
ture the effect that both mean and variance of devel-
in the level of innovation undertaken by both firms.
opment time are positively correlated, which would
Other forms of integration costs are considered in
be the case for such highly innovative projects.3 The
Technical Appendix II (provided in the e-companion)2
development rate
, which determines the time
and do not affect our results drastically. To avoid triv-
taken for technology development, is assumed to be
ial cases, we assume that K is not too high relative to
a function of the firm’s innate development capa-
I and c; specifically, we assume that (I
i + c ≥
i K).
i —directly impacting the speed and consis-
3.3. Model of Technology Uncertainty tency of the development process—and the level of
Prior research has shown that technology uncertainty quality improvement () undertaken by that firm. To
has a significant influence on the product develop- account for the fact that the time required for devel-
ment process (Iansiti 1995, Krishnan and Bhattacharya oping a higher quality product would be higher, we
2002). Consider the case of firms developing a assume that
 is decreasing in . In particular, we
pharmaceutical drug from a lab-discovered molecule use the functional form
i  −1 . To consider the
(technology). The commercializing firm faces several difference in development capabilities between the
types of uncertainty, which limit their chances of suc- two firms, we model that when technology devel-
cessfully getting regulatory and market acceptance. opment is conducted by the focal firm,
i =
F , and
These include factors such as the molecule’s toxic- when the development occurs at the partner firm,
ity, carcinogenicity, and efficacy, as well as endoge-
i =
P . We use = 
F /
P  to represent the rela-
nous factors such as firm capability in managing the tive capabilities of these firms, which will be found
project. to play a key role in future results. Our analysis
We model that at the beginning of technology shows how the choice of codevelopment mechanism
development process, when firms make their invest- is contingent on the nature of uncertainty facing the
ment decisions, they only have an estimate of the effi- project.
cacy of the new product project, which determines the
extent to which a firm can exploit the new innova- 3.4. Information Structure and Decision Sequence
tion. In particular, when the efficacy is ṽ, the resultant An important part of joint development, as we have
effective quality of the product would be ṽ, where ṽ modeled in this paper, is its implementation and,
is uniformly distributed between v and 1. This implies in particular, the sequence of decision making and the
that although the firms could increase the quality informational assumptions that underlie the model.
of the final product through increased investments, In the baseline case of pure revenue sharing, only the
level of innovation has to be determined, but under
An electronic companion to this paper is available as part of the
online version that can be found at http://mansci.journal.informs. We are grateful to the associate editor for pointing out this aspect
org/. of the model.
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS 7

both investment and innovation sharing, more deci- Although contractual agreements on these parame-
sions about investments have to be made in conjunc- ters might be imperfect in some cases, it is certainly
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

tion with the innovation levels, as discussed below. possible in many of the industrial settings we have
Without loss of generality, we assume that the focal considered in this paper, where quality improvements
firm makes the decision on which mechanism for joint are observable and measurable, and development cost
development has to be chosen. Depending upon this curves of firms have known trajectories. Specifically,
choice, the decisions and implementation of contrac- this is applicable in cases where quality improvements
tual development could take different forms. could be broken down to attributes such as battery life,
Under an investment sharing agreement, the deci- processor speed, and drug efficacy. There is an increas-
sion on the level or degree of investment sharing (k) ing tendency for firms to engage outside firms (in par-
should be made before deciding the level of inno- ticular, industrial design firms such as IDEO and
vation (). Thus, when a firm determines its inno- contract development organizations such as Charles
vation level, it knows the share of costs it would River Labs), which suggests that product development
receive from its partner, and hence its decision on work may be contractible. In summary, the model-
the level of innovation would accommodate this cost ing assumptions we make about information structure
sharing level. (It is trivial to note that determining and decision sequence are meant to derive first-order
k after  will not have any impact of the innova- insights and are motivated by industrial practice.
tion level or profits.) Either the focal or partner firm In all of our analyses, we assume that the launch
could conduct the innovation while letting its part- costs and marginal costs of production are constant,
ner share a part of the development costs, so there and for ease of exposition we normalize these costs
can be two forms of investment sharing depending to zero. Although the assumption of linear produc-
on who determines the level of investment sharing k tion cost is simplistic, the fundamental interaction
and who determines the product quality . In con- between the different product development strategies
trast, under innovation sharing, the level of invest- considered requires only that these costs be nonde-
ments F and P can be determined simultaneously. creasing. We also assume that firms are risk neu-
Although there are other forms of innovation shar- tral and are interested in maximizing their expected
ing in which these investment levels are determined profits. In addition, they have to make their invest-
sequentially, it can be easily shown that under the ment into the technology before uncertainty associ-
assumption that the integration cost is of the form as ated with the technology is resolved. This would be
in Equation (2), simultaneous and sequential forms of realistic in conditions where lead times of innovations
innovation sharing yield identical investment levels are much higher than normal production times, which
and profits.4 is the case for technology and knowledge intensive
Our model is driven by the assumption that firms industries.
can infer product development investments from
quality improvements in the end product. Approaches
4. Analysis
such as the function-point method in software devel-
In this section, we evaluate the different product
opment and project journaling make estimation of
development and collaboration choices of firms. At
product development effort possible for a certain level
first, we look at the case in which the firms only
of product quality and are being increasingly used
have a revenue sharing agreement and evaluate
in joint development projects. It is worth pointing
how this agreement influences product development
out the manner in which collaboration was formal-
investments. Subsequently, we derive the effect of
ized and contracted in the joint development relation-
investment and innovation sharing agreements on the
ship between Alpha and Mega cited earlier. A joint
optimal development investment. Finally, we endog-
team from both firms kept track of the man-hours
enize the choice of revenue sharing whose structure
spent by Alpha on every aspect of the project, and
embodies a bargaining framework between the two
Mega compensated Alpha on quarterly basis for the
collaborating firms.
effort according to a contractual agreement. In addi-
tion, we assume that firms do not have information 4.1. Codevelopment Under Pure Revenue Sharing
about whether the new technology would succeed, Let us first consider the case in which investment lev-
but they do know their own capabilities and those els are decided exclusively through revenue sharing
of their partners. The implication of these assump- without any additional cost or effort sharing between
tions is that firms can contract on the level of innova- the firms (“pure revenue sharing case”). Only one of
tion and in turn the extent of development cost that the firms does the development work, and the deci-
they undertake without incurring any agency costs. sion on the quality level  is also made by that firm.
In this setup, the value offered by the other firm that
Proof is available from the authors upon request. justifies revenue sharing could be in complementary
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
8 Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS

value-chain activities such as supply base or channel Result 1. (a) There exists a threshold R on the devel-
management. The case in which the focal firm does opment capability ratio such that if > R , it is optimal
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

development work is referred to as FF revenue shar- for the focal firm to conduct innovation where
ing, and the case in which partner firm does develop-
1 − c
ment work is referred to as PF revenue sharing. R =  (9)
Let us first consider the case when the focal firm 2c − I
P 1 − 3
does the development work. Once the technology has
been successfully developed, the revenues are divided (b) The threshold R is decreasing in .
between the firms according to a revenue sharing Proof. All proofs are provided in Technical
agreement, . The focal and partner firm’s profits for Appendix I (provided in the e-companion).
a given innovation level  can then be represented as For starters, Result 1 simply confirms our intuitive
F  =  + ṽr (3) understanding that who conducts innovation in pure
revenue sharing depends on the relative development
P  = 1 −  + ṽr (4) capability of the firms. If the focal firm’s develop-
ment capability is sufficiently higher than the partner,
Note that this is the profit of the focal firm if the it prefers doing the innovation itself. The focal firm is
new product has succeeded. However, the decision also better served by conducting the innovation itself
on  has to be made before it is known how well when the revenue sharing parameter  increases, and
the innovation can be translated to a product or as a result R is decreasing in .
how much time the development process would take. We now examine the more complex case when
After accounting for these elements of technology the revenue sharing parameter  is endogenously
uncertainty, the expected profit function of FF can be decided by the firms. Recall that we use the bargain-
represented as follows: ing structure proposed by Nash (1950, 1953) to deter-
F  = E  + ṽr − ctF − I 2  (5) mine the equilibrium . For ease of exposition, we
define ∗R to be equilibrium share of revenues of the
2 + 1 + vr c 2
firm who does the development work (also called the
= − − I 2  (6)
F innovating firm).
Differentiating the profit function w.r.t. , we can now Proposition 1. (a) There exists a solution to the Nash
calculate the optimal investment level F∗ of the focal bargaining problem such that ∗R ∈  12 43 .
firm. First-order conditions (sufficient due to profit (b) ∗R is decreasing in the product’s residual value ()
function concavity) yield and increasing in the development capability of the inno-
vating firm.
1 + vr
F (c) When  = 0, ∗R = 43 .
F∗ =  (7)
F + c
Proposition 1 is interesting in that it shows that the
In a similar fashion, we can also find the optimal firm that does the development work is able to retain
level of innovation P∗ under partner revenue sharing a larger share (greater than 50%) of the total revenue.
in which innovation decision is made by the partner This implies that doing the development work implic-
firm instead of focal firm: itly provides the innovating firm a better bargain-
1 − 1 + vr
P ing position in its negotiation with its partner, which
P∗ =  (8) it is able to parlay into a larger share of the total
P + c
profits for itself. The share of revenues R is high-
A casual observation of Equations (7) and (8) shows est when  = 0 (new-revenue projects), in which case
that the investment levels under pure revenue sharing ∗R = 43 . As  increases, R decreases, which illustrates
are lower than under centralized decision making.5 the interaction between the nature of the project and
This distortion is because part of the benefits of the consequent bargaining between firms. To understand
innovation accrues to a firm’s partner, which reduces this interaction, note that the value of innovation is
the marginal value of the development investment. most when  is small (not much preexisting revenue).
We begin by evaluating the profits of both firms when This is because for smaller values of , most of the
the revenue sharing parameter is determined exoge- value that firms attempt to share between them is as
nously. Result 1 below characterizes when and how a result of the innovation. Naturally, it is most valu-
firms engage in pure revenue sharing under this case. able to provide incentives to the innovating firm to
increase investment when  is small. This is achieved
Under centralized decision making, it can be see that i∗ = through offering the innovating firm a larger share of
1 + vr
i /4I
i + c (proof available from the authors upon the total revenue. However, as  increases, the inno-
request). vation is relatively less valuable and consequently, it
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS 9

becomes less important to share a larger fraction of certain ranges of revenue sharing ratio . In partic-
the total revenue to the innovating firm. As a result, ular, when  is high, the focal firm obtains a larger
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

∗R is decreasing in . share of the total profits compared to the partner firm,
It can also be seen that the development capability which limits the incentive of the partner to cofinance
of the firm that does the innovation also affects how the development by the focal firm. When  > F , the
the revenues are shared between the firms. As the incentive of the partner to cofinance the focal firm’s
development capability increases, a firm increases investment costs erodes to such an extent that the
the amount of investment it undertakes (because optimal investment sharing level goes to zero. We
the marginal value of investment increases). Because see a similar effect in PF investment sharing as well
this increases the relative value of the innovation when  is low. However, the advantage of investment
vis-à-vis , it is able to retain a larger share of the sharing lies in that by sharing a part of the develop-
total revenue. As is to be expected, the equilibrium ment costs, a firm is able to incentivize its partner to
revenue sharing value ∗R is increasing in the devel- invest more into the innovation, and this can be seen
opment capability of the innovating firm. by examining the profit function under investment
Having characterized the effect of revenue sharing sharing in the proof of Proposition 2 (in Technical
agreements on investment levels and profitability, we Appendix I). Result 2 also shows how the choice for
now turn our attention to the cases when firms also a specific form of investment sharing depends on the
engage in cost or effort sharing agreements in addi- relative development capability of firms. As expected,
tion to revenue sharing. First we look at investment a firm would prefer to have its partner conduct devel-
sharing and its impact on investments and profits, fol- opment work while cofinancing this investment only
lowed by innovation sharing. if the capability of its partner is sufficiently high.
When the capability of its partner is low compared to
4.2. Investment Sharing itself, allowing the partner to undertake development
Under investment sharing, development work is done results in lower levels of quality improvement, and
at one of the firms, with its partner sharing the invest- for sufficiently low , a firm is better off doing the
ment cost. Two different forms of investment sharing development itself.
emerge. The case in which the focal firm conducts Similar to revenue sharing, the choice between the
innovation is referred to as FF investment sharing. different forms of investment sharing is also influ-
When the partner firm does development work and enced by how the revenues are split between the
the focal firm in turn cofinances this development, we firms. However, unlike under pure revenue sharing,
refer to it as PF investment sharing. It follows that we find that the threshold on capability above which
there are two decisions to be made under investment a firm would like to conduct the innovation on its
sharing, namely, the level of investment sharing and own is decreasing in  only if  > 12 . Note that when
the level of quality improvement. Let k be the share of  > 12 , the focal firm receives a larger share of the total
total development costs that the focal firm would bear revenue vis-à-vis its partner. A further reduction in 
and  be the development work undertaken by the in this range ( > 12 ) reduces the partner firm’s profits
innovating firm. As under revenue sharing analysis, even more, and hence reduces its incentives to share
the optimal decisions of both firms can be character- the development costs. Because of this, the focal firm
ized by backward induction. The following starting is better served by assuming Stackelberg leadership
result characterizes the effect (an exogenous) revenue and does so by choosing PF investment sharing. It
sharing parameter  and the development capabil- follows that IS is decreasing in  only if  > 12 .
ity have on the relative attractiveness of these two Although the above analysis is useful in under-
mechanisms. standing interactions between development invest-
Result 2. (a) FF investment sharing is feasible iff  ≤ ments and cost sharing in established markets where
F and PF investment sharing is feasible iff  ≥ P . existing contractual agreements could form the basis
(b) There exists a threshold IS on the development of codevelopment choices, negotiations between col-
capability ratio such that if > IS , then FF investment laborating parties could determine the revenue shar-
sharing dominates PF investment sharing where ing parameter in emerging product markets. We now
turn our attention to such markets and examine how
2c1 − 2  these mechanisms are affected when  is endogenized
IS =  (10)
c2 − 2 +
P I2 + 32 − 4 using the Nash bargaining structure proposed earlier.
In contrast to revenue sharing, note that the level of
(c) The threshold IS is decreasing in  iff  > 12 .
investment sharing (k) must also be decided, mak-
It can be seen above that revenue sharing ratio  ing this decision part of the bargaining structure. Let
strongly impacts the viability of investment sharing. ∗F and ∗P represent the equilibrium levels of rev-
We find that investment sharing is feasible only for enue sharing under FF and PF investment sharing,
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
10 Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS

and let kT∗ and kP∗ represent the corresponding invest- We see that innovation sharing is feasible only for
ment sharing levels. intermediate values of . This is because innovation
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

sharing entails both firms to invest in development

Proposition 2. The equilibrium level of revenue shar-
and subsequently incur the integration costs. As a
ing and investment sharing are as follows:
result, if the integration cost parameter K is very high,
∗F = ∗P = 12 the corresponding share of revenues that the firm
would receive would not be sufficient to compensate
kF∗ = kP∗ = 12  the firm for the development costs that it would incur.
Naturally, it will choose to invest in development only
Proposition 2 is interesting in that when  and k are if its share of revenues is sufficiently high. If  is
determined through bargaining between firms, both too low, then the focal firm does not get compen-
costs and revenues are shared equally. This is quite sated adequately for its investment, and if  is too
unlike under revenue sharing where a larger share high, then the partner does not get compensated ade-
of the profits gets captured by the firm that does the quately. In both these cases, innovation sharing would
development work. By engaging in investment shar- be infeasible.
ing and splitting the development costs with its part- Having determined the effect of  on the firms’
ner, a firm is able to achieve two objectives: It is able incentive to engage in innovation sharing, we now
to induce its partner to invest in higher quality lev- examine what happens when  is endogenized. In
els. In addition, it is able to improve its bargaining the following proposition, we characterize the equi-
position in its negotiation with its partner and retain librium revenue sharing level under innovation shar-
a larger share of the total profits. The fact that rev- ing and how this level is affected by the development
enues and costs are shared equally is also interesting. capability of firms.
This is because the equilibrium  is set in a manner
that provides the partnering firms adequate incen- Proposition 3. The equilibrium level of revenue shar-
tives to both invest and cofinance. In fact, when costs ing under innovation sharing ∗IN can be represented as
and revenues are shared in the same ratio, the distor- follows:
tions created due to multiparty decision making are (a) If = 1, then ∗IN = 12 .
eliminated, as a result of which sharing the costs and (b) If < 1, then there exists a solution to the Nash
revenues equally becomes the equilibrium solution. bargaining problem such that ∗IN ∈ 41 12 .
It is also worthwhile to point out that the equilib- (c) If > 1, then there exists a solution to the Nash
rium revenue sharing ratio is independent of both  bargaining problem such that ∗IN ∈  21 43 .
and the development capability of the innovating
In contrast to investment sharing, the equilib-
firm, which is another point of departure from rev-
rium  under innovation sharing is not always 12 .
enue sharing. This is again because of the fact that any
In fact, ∗ = 12 only when the development capa-
changes in both  and development capability are felt
bilities of both firms are the same, i.e., = 1. This
equally by both firms because they share not only the
is because, unlike under investment sharing or rev-
revenues of the innovation but also the correspond-
enue sharing, both firms are involved directly in
ing costs. Moreover, this difference also points to the
the development work when they engage in inno-
ability of investment sharing as mechanism to align
vation sharing. Coincidentally, the extent of bargain-
incentives of both firms as opposed to a pure revenue
ing power—and the corresponding share of revenues
sharing agreement where only revenues are shared.
that this power provides—depends on the amount of
4.3. Innovation Sharing development work undertaken by a firm. When the
Under innovation sharing, a part of the development development capabilities of firms are the same, they
work is conducted by the focal firm and the rest is invest exactly the same amount, and hence the rev-
conducted by its partner. Let F and P be the devel- enues are also shared equally. However, when the
opment work conducted by focal firm and partner development capabilities of the firms are not the same,
firm, respectively. Recall that we assume that these the firm with higher capability seems to have a greater
decisions on investment levels are made simultane- incentive to undertake more development work than
ously. Once the innovation materializes, firms incur its partner, thereby also earning a higher share of
an integration cost CF P  = KF2 + P2  to bring revenue.
together the quality improvements distributed across
both firms. As before, we first examine the conditions 5. Comparison of Codevelopment
under which innovation sharing is a feasible mecha-
nism for the partnering firms.
Thus far, we have focused our efforts on identifying
Result 3. Innovation sharing is feasible iff  ≤  ≤ . how cost and effort sharing mechanisms should be
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS 11

implemented. However, as we saw in the industrial we will refer to PF investment sharing as investment
examples discussed earlier, firms in different contexts sharing and FF revenue sharing as revenue sharing.
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

typically use one of these mechanisms to manage We now examine the effect of timing uncertainty on a
the joint development process. This raises the impor- firm’s preference for cost and effort sharing. Because
tant question as to whether there are conditions—both
T and
P represent a firm’s ability to manage tim-
market and project related—that make these mecha- ing uncertainty, examining the choice among different
nisms appropriate for different conditions. To answer mechanisms as a function of these parameters allows
this question, we now look at how these mechanisms us to capture the effect of timing uncertainty.
compare against each other and characterize the con-
Proposition 5 (Effect of Development Capabil-
ditions of appropriateness of each of these mecha-
ity and Timing Uncertainty):
nisms in Propositions 4–7.
(a) There exists a threshold on the development capabil-
Proposition 4 (Effect of Endogenous Revenue ity IS such that if < IS , then investment sharing is the
Sharing (Bargaining) on Codevelopment Mecha- dominant mechanism.
nisms): (b) There exists a threshold on development capabil-
(a) PF revenue sharing is dominated by the other rev- ity RB such that if > RB , revenue sharing is the dominant
enue and cost sharing mechanisms and is never optimal at mechanism.
equilibrium. (c) When = 1, there exists a threshold on the inte-
(b) FF investment sharing is dominated by the other gration cost K such that if K < K , innovation sharing
revenue and cost sharing mechanisms and is never optimal dominates investment sharing and revenue sharing.
at equilibrium.
Proposition 5 shows that the preferred form of
Proposition 4 shows that if the focal firm were to codevelopment depends critically on the develop-
choose investment sharing as the mechanism to col- ment capability and associated timing uncertainty.
laborate with its partner, it should let the partner In particular, investment sharing is the dominant
do the innovation and cofinance the development. mechanism when the development capability of the
However, if pure revenue sharing is the mecha- focal firm is low, and revenue sharing is the dominant
nism that is chosen, it should rather conduct the mechanism when its capability is high. When devel-
development itself. This is because of the interac- opment capability falls in the intermediate range,
tion between the type of mechanism that is chosen innovation sharing is the dominant mechanism.
and the resultant bargaining between firms to deter- One of the reasons why revenue sharing is the
mine the equilibrium revenue sharing level . Recall preferred mechanism when the focal firm’s capabil-
that under investment sharing revenues are shared ity is high is because an innovating firm is able
equally, whereas with pure revenue sharing the firm to retain a larger share of the total revenue under
that conducts the innovation is able to retain a larger revenue sharing. Because the level of investment is
portion (50%–75%) of the total revenue. The subop- determined by the firm that gets the larger share of
timality of FF investment sharing and PF revenue revenues (in this case, the focal firm), it does not
sharing is primarily because of this reason. If the result in that much inefficiency/quality deterioration
focal firm were to seek cofinancing from its part- compared with investment sharing. More importantly,
ner when it does the development work, it would because the focal firm’s capability is also sufficiently
have to forgo a significant portion of the revenues in high, there isn’t much of a penalty for lack of col-
the process at the bargaining stage. Thus, although laboration. Although investment sharing would have
investment sharing results in better allocation of costs enabled better allocation of costs, it would require the
and, hence, potentially higher development invest- firm to forgo a large fraction of the revenues, so it
ment, the resultant lower bargaining power makes ceases to be a viable option. However, when the firm’s
a firm less willing to consider it for collaboration. capability is low compared with its partner’s (low ),
As a result, FF revenue sharing always dominates FF it becomes important to ensure that the development
investment sharing. However, when at equilibrium work is conducted by the partner firm. By engaging
it is the partner who does the innovation, it would in investment sharing when is low, a firm is able
be in a firm’s best interests to engage in investment to ensure that its partner’s investment is sufficiently
sharing with its partner. The focal firm achieves two high in addition to obtaining a better bargaining posi-
objectives by doing this: it is able to ensure that the tion to retain a large enough share of the total profits.
partner invests sufficiently high amount; in addition, As a result, investment sharing becomes the optimal
investment sharing provides it the bargaining power mechanism when is low.
to obtain a larger share of the total profits. As a result, More interestingly, we find that innovation shar-
PF investment sharing dominates PF revenue shar- ing dominates both investment sharing and rev-
ing. Due to these dominance results, from here on enue sharing when is in the intermediate range.
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
12 Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS

This is because the economies of specialization that shared equally between the firms (Proposition 2). Dis-
innovation sharing benefits from enable each firm to tortions associated with a multiparty decision-making
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

concentrate on aspects that are closer to its core capa- setup are eliminated because revenue and cost impli-
bility and ensure higher levels of innovation with cations of any investment are borne equally by both
lower investment costs. When capabilities of both firms. This reduces any incentive by the innovating
firms are comparable, it benefits to resort to innovation firm to underinvest, which is the primary reason for
sharing, have the development work split between the inefficiency in a pure revenue sharing mechanism.
both firms, and exploit the economies of specialization As a result, the combined profits of both firms under
that this division provides. However, such a division investment sharing could be the same as that of a cen-
of effort also comes with an explicit cost in the form tralized decision-making setup.
of integration costs. If the integration cost parameter
K is too high, the benefits of specialization do not ade- 6. Sensitivity Analysis of
quately compensate for the higher integration costs,
and innovation sharing ceases to be optimal.
Codevelopment Mechanisms
To understand the sensitivity of the above results
Now let us consider the impact of translational
about the relative attractiveness of investment and
uncertainty on the choice of mechanisms. To this end,
innovation sharing to changes in market and techno-
we let the development capability of both firms
logical parameters, we discuss the results of numer-
P to tend to , or let the resource cost param-
ical analysis, which illustrate the effects of timing
eter c tend to 0. Because firms face no timing uncer-
uncertainty, translational uncertainty, and nature of
tainty either when their development capability is
technology on the codevelopment mechanism a firm
very high or when the resource costs are negligible,
should adopt. This analysis leads to four observations
the case in which
P →  or c = 0 can be used to
that are valid for a large range of technological and
characterize the effects of translational uncertainty.
market parameters. In all of the analyses depicted as
Proposition 6 (Effect of Translational Uncer- figures, the vertical axes of the figures represent the
tainty). When
P → , or  = 0: profits of the focal firm for the different mechanisms
(a) Investment sharing is dominated by innovation and the horizontal axes represent the development
sharing and revenue sharing. capability ratio of the firms ( ).
(b) There exists a threshold on the integration cost Kc In the first observation below, we examine how the
such that is K < Kc , innovation sharing is the dominant preferred codevelopment mechanism is affected by
mechanism. an increase in timing uncertainty. The results of this
The primary reason for the nonoptimality of invest- numerical analysis are represented in the two graphs
ment sharing under translational uncertainty is its that form Figure 2. In the graph on the left side of this
effect on the revenue sharing negotiation between figure, we represent the scenario when the resource
firms. In our model, when there is no timing uncer- costs are low (c = 1), whereas on the right the resource
tainty, the development capability of firms does not costs are higher (c = 2). Because the effects of tim-
impact which mechanism is preferred. As result, rev- ing uncertainty are greater when resource costs are
enue sharing, which allows the focal firm to obtain higher, an increase in the resource costs (c = 1 to c = 2)
a larger share of the total revenue, dominates invest- allows us to characterize a firm’s optimal response to
ment sharing. However, if the integration costs are an increase in timing uncertainty.
sufficiently low, then innovation sharing becomes the Observation 1. An increase in the effect of timing
dominant mechanism. This is because the distributed uncertainty increases the value of codevelopment using
effort and cost function convexity under innovation investment sharing and reduces the relative value of pure
sharing reduces the development cost and provides revenue sharing.
adequate incentives for innovating firms to invest
in greater levels of product quality improvement. As seen in Figure 2, an increase in timing uncer-
Thus, translational uncertainty lends itself to inno- tainty makes codevelopment using either investment
vation sharing, and timing uncertainty favors invest- sharing or innovation sharing more valuable than
ment sharing. a pure revenue sharing arrangement. This can be
inferred by comparing the two graphs in Figure 2 and
Proposition 7. When investment sharing is the dom- considering the range of the development capability
inant mechanism, the combined profits of both firms match for which both investment and innovation sharing are
that of a centralized decision-making setup in which devel- optimal. As we move to the right, the shaded region,
opment work is conducted by a single firm. which corresponds to the region of dominance of rev-
Proposition 7 is useful in that it shows invest- enue sharing, shrinks in size, and the threshold devel-
ment sharing mechanism can replicate a centralized opment capability below which investment sharing
decision-making setup because revenues and costs are is attractive increases (implying that an increase in
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS 13

Figure 2 Effect of Timing Uncertainty

not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may


Investment sharing 0.70

Equilibrium profits

Equilibrium profits
0.75 Innovation sharing
Revenue sharing

0.70 Profits under pure 0.60

revenue sharing
Profits under 0.55
innovation sharing
Profits under
investment sharing 0.50

0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Development capability ratio Development capability ratio

Note.  = 0; I = 4; P = 2; cL = 1; cH = 2; K = 13; v = 05; r = 20/3.

timing uncertainty would make investment sharing by comparing the two graphs in Figure 2, where we
more attractive). Because innovation sharing involves see that the threshold of below which innovation
development work being conducted by both firms, sharing is optimal is lower when the resource costs
the expected development time required under inno- are higher (c = 2). This is because incentive alignment
vation sharing relative to investment could be higher, benefits that innovation sharing provide compared
particularly when the focal firm’s development capa- with revenue sharing results in higher development
bility is low relative to that of its partner. This also investments from both firms. When the resource costs
implies that resource allocation costs for innovation increase, the relative value of such an incentive align-
sharing would be higher compared with investment ment also increases because of the higher marginal
sharing. As a result, an increase in c affects the viabil- costs of development. In addition, the higher devel-
ity of innovation sharing more than it affects invest- opment time makes splitting the fixed costs between
ment sharing. It follows that the optimal response to firms more valuable, and innovation sharing becomes
an increase in c would be to engage in investment more attractive compared with revenue sharing.
sharing even for higher values of . Two other project dimensions that influence a firm’s
This result of increasing attractiveness of invest- preferred mode of codevelopment are the nature of
ment sharing, however, does not carry over to the case the revenues of the development project and the
when the development capability of the focal firm is degree of translational uncertainty. In Figure 3, we
high. Recall from Proposition 5 that when the devel- look at how a change in  (which captures the extent
opment capability of a firm is high, it is optimal to to which a development project generates new rev-
either engage in pure revenue sharing or innovation enues) affects the optimal codevelopment mechanism.
sharing. Under such situations, an increase in c actu- Observation 2. For new-revenue projects, investment
ally pushes a firm toward innovation sharing in place and innovation sharing are more attractive codevelopment
of pure revenue sharing. This can again be observed approaches than pure revenue sharing.

Figure 3 New-Revenue vs. Replacement-Revenue Projects


Investment sharing
Equilibrium profits

Innovation sharing
Equilibrium profits

Revenue sharing 0.80

Profits under pure
revenue sharing
Profits under
0.60 innovation sharing
Profits under
investment sharing 0.70

0.4 0.6 0.8 1.0 1.2 1.4 0.4 0.6 0.8 1.0 1.2 1.4
0.50 Development capability ratio
Development capability ratio

Note. L = 0; H = 025; I = 4; P = 2; c = 2; K = 13; v = 05; r = 20/3.

Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
14 Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS

New-revenue projects have the most to gain total revenue becomes the overarching objective. As a
when firms engage in codevelopment through invest- result, revenue sharing that allows the innovating firm
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

ment and innovation sharing. This is illustrated to retain a larger share of the total revenue to itself
through a comparison of the focal firm’s optimal co- becomes the dominant mechanism.
development strategy for different values of , the We now focus on the effect of translational uncer-
analysis of which is represented in the two graphs tainty on a firm’s preferred mode of codevelopment.
in Figure 3. In the graph on the left,  = 0, which In particular, we examine the shift in a firm’s code-
corresponds to projects with no preexisting revenue velopment strategy as it moves from a project with
(all revenues accrue from this innovation), whereas low translational uncertainty to another project with
in the graph on the right we have that  = 025, high translational uncertainty. Note that the presence
which corresponds to replacement-revenue projects. of timing uncertainty is what makes this analysis dis-
Although the structures of the optimal policy as repre- tinct from Proposition 6.
sented in both of these graphs have a similar form, we
Observation 3. For projects facing both translational
see that the thresholds of that demarcate the regions
and timing uncertainty, investment sharing competes with
of optimality of codevelopment mechanisms are dif-
innovation sharing for new-revenue products and with
ferent depending on whether  is small or large. As
pure revenue sharing for replacement revenue products.
the comparison between these two graphs suggests,
we can see that when  increases from 0 to 025, We find that innovation sharing is relatively less
the range of parameters for which investment sharing attractive compared with investment sharing and rev-
and innovation sharing are optimal decreases. enue sharing for replacement-revenue projects that
Why are the codevelopment mechanisms proposed have both timing and translational uncertainty. This
in this paper (investment and innovation sharing) can be seen by comparing the two graphs in Figure 4,
more valuable for new-revenue projects than for which represent different levels of translational uncer-
replacement-revenue projects? One possible reason is tainty. In the graph on the left, we have that v = 08
that the value of collaboration is greatest for new- (representing low translational uncertainty), and in
revenue projects. When  is small or negligible, the the graph on the right we have that v = 02 (rep-
entire value that firms seek to maximize is a direct con- resenting high translational uncertainty). In contrast,
sequence of the development work that is undertaken. when firms face both timing and translational uncer-
Under such circumstances, it becomes very important tainty for new-revenue projects, we find that invest-
for the firms to choose mechanisms that induce higher ment sharing and innovation sharing are dominant
investment levels. Because both investment and inno- forms of codevelopment. This can be seen by exam-
vation sharing align incentives of firms and increase ining Figure 2 for which  = 0 and comparing it with
their development investments, and, consequently, the Figure 4.
level of innovation, these mechanisms also become As before, we can compare the specific thresholds
most valuable when  is low. Additionally, the reduc- on within which innovation sharing is optimal for
tion in the fixed development costs that occurs under these different levels of v to illustrate how the code-
innovation sharing allows firms to realize higher qual- velopment strategy should be adjusted to the level
ity with lower costs. In contrast, when  is high, of translational uncertainty in a replacement-revenue
the value generated through the new innovation is project. This comparison shows that the range of
lower, and the ability to retain a larger share of the development capability ratio for which both revenue

Figure 4 Effect of Translational Uncertainty

Investment sharing
Equilibrium profits

Innovation sharing 0.55

Equilibrium profits

1.05 Revenue sharing

Profits under pure

1.00 revenue sharing
Profits under
0.95 innovation sharing
Profits under
0.90 investment sharing

0.4 0.6 0.8 1.0 1.2 1.4

0.4 0.6 0.8 1.0 1.2 Development capability ratio
Development capability ratio

Note.  = 02; I = 4; P = 2; c = 2; K = 13; vH = 08; vL = 02; r = 20/3.

Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS 15

Figure 5 Effect of Outside Option on Codevelopment

not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

Investment sharing

Equilibrium profits
0.65 Innovation sharing
Equilibrium profits

Revenue sharing
Profits under pure
revenue sharing
Profits under
innovation sharing 0.60
0.50 Profits under
investment sharing

0.2 0.4 0.6 0.8 1.0 1.2 1.4

0.2 0.4 0.6 0.8 1.0 1.2 1.4 Development capability ratio
Development capability ratio
Note.  = 0; I = 4; P = 2; c = 2; K = 13; v = 05; r = 20/3; F = 005; P = 005.

sharing and investment sharing are optimal increases the outside option compares with that of the rival.
as v increases, implying that when firms face both In particular, we find that when the outside option
timing and translational uncertainty, revenue shar- of the partner is higher, innovation sharing becomes
ing and investment sharing are the dominant mecha- the dominant mechanism. In contrast, when the focal
nisms. To understand this result, note that the effects firm’s outside option is higher compared with its part-
of timing uncertainty are most felt by innovation shar- ner’s, pure revenue sharing is sufficient to coordi-
ing, particularly when development capabilities of nate the codevelopment process. To understand why
firms are different (high or low ). As a result, when a higher value of outside option makes innovation
the translational uncertainty associated with a project sharing preferable, recall that the focal firm tends to
is already very high, exacerbating its effect by taking retain a larger fraction of the project value under
upon timing uncertainty as well might not be opti- pure revenue sharing, as a result of which it is more
mal. It follows that both investment sharing and rev- likely for the partner firm to opt out under this agree-
enue sharing, which face less timing uncertainty and ment. Thus, revenue sharing is less likely to align
no integration costs, become dominant mechanisms incentives and hence becomes less attractive. In con-
for larger ranges when translational uncertainty in trast, innovation sharing results in a more equitable
a replacement revenue project is higher. distribution of project value because the revenues
Until now we have assumed that the outside are shared (loosely) according to their development
options of both firms were identical.6 We now exam- investments. This enables better incentive alignment
ine how the structure and operationalization of co- and hence makes it less likely for the partner firm
development is affected when this assumption is to opt out. However, when FF’s outside option is
relaxed. These results are depicted in the two graphs higher, eliminating its incentives to opt out becomes
more important. Naturally, revenue sharing, which
in Figure 5—in the graph on the left, the outside option
allows the focal firm to retain a larger share of the
of PF is 0.05 and that of FF is 0, whereas in the graph
revenues to itself, becomes the preferred mode of
on the right, we have that the outside option of PF
is 0 and that of FF is 0.05. Comparing these graphs
Our numerical analysis also shows that the equilib-
with Figure 2, where we assume that outside options
rium  under investment sharing is no longer 12 . Sim-
of both firms are zero, helps us characterize the impact
ilarly, the presence of outside options can also lead
of these options on codevelopment choices.
to a pure revenue sharing agreement where the equi-
Observation 4. When the value of the outside option librium ∗R is greater than 43 . Thus, the availability
for the partner firm increases, codevelopment using inno- of outside options also provides a better bargaining
vation sharing becomes more valuable. In contrast, when position for a firm with respect to its partner and
the value of focal firm’s outside option increases, revenue allows it to capture a larger share of the realized
sharing becomes more attractive. project value. The implication of these observations is
that the bargaining power with which a firm enters
Whether FF prefers innovation sharing or pure the negotiation process plays an important role in
revenue sharing depends on how its own value of the operational implementation of codevelopment. It
influences both how the revenues are split between
Recall that we normalized the outside option value of the firms firms and determines which mechanism is the opti-
to 0. mal mode of codevelopment.
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
16 Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS

7. Managerial Implications and predominant timing uncertainty. When the project

involves the launch of a product with incremental rev-
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to

Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

In this paper, we conceptualize and model the rev- enues, revenue sharing is sufficient to facilitate the
enue, cost, and effort sharing collaborative arrange- collaboration.
ments between two firms and characterize the optimal The above framework can be used to interpret, in
joint-development approach for various technological part, why some firms resort to sharing the develop-
and market parameters. We find that the preferred ment work (innovation sharing) in multifirm develop-
codevelopment approach should go beyond simple ment, whereas we see the funding of the development
revenue sharing, under which there exists an incen- work in other situations. Consider the computer
tive for an innovating firm to underinvest in quality industry where most products are new to the mar-
improvements. The cost and effort sharing mecha- ket rather than new to the world and are introduced
nisms presented in this paper have the potential to around major industry events and conferences. In this
address the inefficiencies associated with multiparty case, timing is determined by outside events and the
decision making. We found from our analysis that uncertainty is more of the translational kind (quality
investment and innovation sharing are more appro- of the product). Our framework would recommend an
priate for collaboration depending on a variety of innovation sharing approach for such cases. We dis-
conditions. In particular, when firms have distinc- cussed earlier Dell’s efforts to complement its suppli-
tive capabilities, innovation sharing agreements help ers, who perform upstream component development
firms exploit their specialized product development activities, leaving Dell with the system integration and
capabilities and provide firms the appropriate incen- qualification activities. In fact, the supplier selection
tives to ensure greater investments in technology processes at Dell ensure that vendor skills comple-
and product development. However, when develop- ment its core capabilities in operational integration
ment capability is concentrated in one of the firms, (Financial Times 2003, Hachman 2002). This puts it in
the additional integration costs of innovation sharing the top left of the framework in Figure 6.
reduce its attractiveness. Because the quality distortion In contrast, the pharmaceutical industry is char-
effects of multiparty decision making are eliminated acterized by long and highly uncertain lead times
through investment sharing under such conditions, for drug development and, hence, correspondingly
investment sharing helps attain optimal investments high resource allocation costs, high levels of develop-
in development and becomes the optimal mode of ment uncertainty, and development capabilities that
codevelopment. are concentrated in small biotech firms, all of which
The insights from the modeling and analysis of result in high timing uncertainty. More importantly,
the codevelopment problem are distilled into a con- strong regulatory influence (from the likes of the Food
ceptual framework in Figure 6. This framework pro- and Drug Administration in the United States), which
vides managerial insights by demarcating the regions makes it necessary for products to possess a bare
of appropriateness of the different codevelopment minimum quality, forces firms to continue working
approaches. Specifically, the cost sharing approaches until such progress is achieved, making their challenge
are more appropriate for new-revenue projects. Inno- managing the development time rather than the final
vation sharing is more appropriate for projects with product quality. We believe that these industry and
translational uncertainty, and investment sharing out- technology characteristics have made investment shar-
performs the other approaches for projects with ing the popular form of contractual development in
the pharmaceutical industry (top right of the frame-
Figure 6 Optimal Codevelopment Approach for Different work in Figure 6).
Project Types Our analysis, which provides concrete guidelines on
how the codevelopment approach needs to be tailored
to the nature of the project/product, still is quite styl-
ized and represents merely a groundbreaking effort.

Type of project revenue


Innovation Investment
sharing Collaborative product development is a rich topic with
a whole host of issues that need to be studied to deter-
mine how relationships between firms must be struc-

Pure revenue Pure revenue/ Investment

tured and nurtured. Complex interactions between
sharing investment sharing sharing firms in a product development context seem to limit
the degree to which inefficiencies stemming from lack
Only translational Both translational Predominantly of coordination can be reduced compared to a supply
uncertainty and timing timing chain/distribution context where primarily prices and
uncertainty uncertainty
quantities are being negotiated. Further analytical and
Type of project uncertainty empirical work would enhance our understanding of
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS 17

these complex yet important issues. In the first phase Investment sharing is a better mechanism when firm
of the work, we made a number of stylized assump- capabilities are dissimilar and projects face significant
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

tions to obtain compact analytical expressions. In addi- timing uncertainty. However, when firms are simi-
tion to assuming specific forms for development cost lar in terms of their capabilities and collaborate on
functions similar to the ones used in the literature, new-to-market product projects with significant trans-
we also ignored agency costs associated with coop- lational uncertainty, innovation sharing is better suited
erative effort. A firm’s investments normally extend to leverage the specialized skills of individual innovat-
beyond money into human resources, intellectual cap- ing firms. We believe the modeling and the framework
ital, and a host of other factors that are not verifiable derived from the analytical results in this paper pro-
across firms. Although, expressing the development vide a richer understanding of collaborative product
investment in terms of quality improvement (quality development (beyond the current general discussion
could be quantified in terms of battery life, processor found in the business press) and lay the ground-
speed, or memory capacity in most technology inten- work for more advances on this increasingly impor-
sive industries) brings in a degree of verifiability to tant topic.
the process, several other soft issues like trust make
the problem difficult to quantify. Although we recog- 8. Electronic Companion
nize the existence and importance of these factors, we An electronic companion to this paper is available as
believe that modeling these are beyond the scope of part of the online version that can be found at http://
this research, and we leave those as an avenue for
future research.
We also modeled the case when firms make deci- Acknowledgments
sions with complete information about each other’s The authors gratefully acknowledge the department edi-
costs and incentives. Although this would be true tor, associate editor, and three anonymous referees for their
helpful suggestions and comments during the review pro-
for advanced development projects where collaborat-
cess. The second author also acknowledges the financial
ing firms have to work very closely (particularly if support from the Sheryl and Harvey White endowment.
they collaborated successfully in the past), it might
not be appropriate for R&D initiatives of firms that
do not have a history of collaboration with each References
other. Additional work needs to be conducted to Abbott, L. 1953. Vertical equilibrium under pure quality competi-
understand how asymmetric information about mar- tion. Amer. Econom. Rev. 43(5) 826–845.
Aghion, P., J. Tirole. 1994. The management of innovation. Quart. J.
ket demand will impact development decisions. We Econom. 109(4) 1185–1209.
also ignored manufacturing and launch costs, which Amaldoss, W., R. Meyer, J. S. Raju, A. Rapoport. 2000. Collaborating
could be important in some industries. Interactions to compete. Marketing Sci. 19(2) 105–126.
between these costs and product design choices could Anderson, E. G., G. G. Parker. 2002. Project sourcing decisions in
have a significant impact on joint-development deci- the presence of learning and integration costs. Working paper,
University of Texas, Austin.
sions, and modeling these issues is an avenue for Austin American Statesman. 2003. Alliances, consulting let IBM share
future research. More broadly, we also restricted our research costs. (April 21).
analysis to the case of a single technology and prod- Cachon, G. P., M. A. Lariviere. 2005. Supply chain coordination with
uct development firm. In reality, multiple technol- revenue sharing contracts: Strengths and limitations. Manage-
ment Sci. 51(1) 30–44.
ogy suppliers go to market through multiple product
Cohen, C. M., S. Klepper. 1996. A reprise of size and R&D.
development firms, and understanding competitive Econom. J. 106(437) 925–951.
interactions between multiple suppliers and product Cohen, W. M., S. Klepper. 1992. The anatomy of industry R&D
development firms would further enrich this discus- intensity distributions. Amer. Econom. Rev. 82(4) 773–799.
sion. A more detailed model of technology uncer- Cvsa, V., S. M. Gilbert. 2002. Strategic commitment versus post-
ponement in a two-tier supply chain. Eur. J. Oper. Res. 141(3)
tainty and distributed product development would 526–543.
expand the applicability of this research. Doz, Y., G. Hamel. 1998. Alliance Advantage: The Art of Creating Value
In summary, conceptualization, model formulation, Through Partnering. Harvard Business School Press, Boston.
analysis, and discussion of multifirm product devel- Dutta, S., A. M. Weiss. 1997. The relationship between a firm’s level
opment has produced a number of interesting insights. of technological innovativeness and its pattern of partnership
agreements. Management Sci. 43(3) 343–356.
First, it is important to consider technology and mar- Dyer, J. H. 2003. Collaborative Advantage: Winning Through Extended
ket characteristics in conjunction with economic and Enterprise Supplier Networks. Oxford University Press, Oxford,
operational issues when making product development UK.
decisions. Second, investment and innovation sharing, Erat, S., S. Kavadias. 2006. Introduction of new technologies to com-
peting industrial customers. Management Sci. 52(11) 1675–1688.
proposed as codevelopment mechanisms on top of
Financial Times. 2003. Dell takes on printer market. (March 25).
revenue sharing mechanisms, can help product devel- Fine, C. H., D. E. Whitney. 1996. Is the make-buy decision
opment firms coordinate investments and achieve bet- process a core competence? Working Paper 140-96, Sloan
ter quality products and higher profitability for firms. School of Management, Massachusetts Institute of Technology,
Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development
18 Management Science, Articles in Advance, pp. 1–18, © 2009 INFORMS

Cambridge, MA. Lee, E., R. Staelin. 1997. Vertical strategic interaction: Implications
html. for channel pricing strategy. Marketing Sci. 16(3) 185–207.
not be posted on any other website, including the author’s site. Please send any questions regarding this policy to
Copyright: INFORMS holds copyright to this Articles in Advance version, which is made available to institutional subscribers. The file may

Gerwin, D., J. S. Ferris. 2004. Organizing new product development Loch, C. H., C. Terwiesch. 1998. Communication and uncertainty
projects in strategic alliances. Organ. Sci. 15(1) 22–37. in concurrent engineering. Management Sci. 44(8) 1032–1048.
Gilbert, S. M., V. Cvsa. 2003. Strategic commitment to price to stim- McGuire, T. W., R. Staelin. 1983. An industry equilibrium anal-
ulate downstream innovation in a supply chain. Eur. J. Oper. ysis of downstream vertical integration. Marketing Sci. 2(2)
Res. 150(3) 617–639. 161–191.
Grahovac, J., G. G. Parker. 2002. Component modularity, external Nash, Jr., J. 1950. The bargaining problem. Econometrica 18(2)
economies of scale and outsourcing decisions in the supply 155–162.
chain. Working paper, Tulane University, New Orleans. Nash, Jr., J. 1953. Two-person cooperative games. Econometrica 21(1)
Grossman, S. J., O. D. Hart. 1986. The costs and benefits of own- 128–140.
ership: A theory of vertical and lateral integration. J. Political Novak, S., S. Eppinger. 2001. Sourcing by design: Product complex-
Econom. 94(4) 691–719. ity and the supply chain. Management Sci. 47(1) 189–204.
Grover, J. M. 1998. The evolving role of collaboration in biotechnol- Owen, G. 1995. Game Theory, 3rd ed. Academic Press, New York.
ogy. Nature Biotechnology 16 31–32. Quinn, J. B. 2000. Outsourcing innovation: The new engine for
Gupta, S., R. Loulou. 1998. Process innovation, product differenti- growth. Sloan Management Rev. 41(4) 13–28.
ation, and channel structure: Strategic incentives in a duoploy. Spencer, B., J. Brander. 1992. Pre-commitment and flexibility: Appli-
Marketing Sci. 17(4) 301–316. cations to oligopoly theory. Eur. Econom. Rev. 36(8) 1601–1626.
Hachman, M. 2002. Dell to enter printer market with Lexmark. Thomke, S. H., V. Krishnan, A. Nimgade. 1999. Product devel-
Extreme Tech (September 24), opment at Dell Computer Corporation. HBS Case 9-699-010,
article2/0,3973,551678,00.asp. Harvard Business School, Boston.
Iansiti, M. 1995. Technology integration: Managing technological Ulrich, K. T., D. Ellison. 2005. Beyond make-buy: Internalization
evolution in a complex environment. Res. Policy 24 521–542. and integration of design and production. Production Oper.
Jeuland, A. P., S. M. Shugan. 1983. Managing channel profits. Mar- Management 14(4) 315–330.
keting Sci. 2(3) 239–272. Whalen, J. 2008. Glaxo gives actelion sweet deal for drug. Wall
Kamien, M. I., N. L. Schwartz. 1992. Market Structure and Innovation: Street Journal (July 15) B9.
A Survey. Cambridge University Press, Cambridge, UK. Zenger, T. R. 1994. Explaining organizational diseconomies of scale
Krishnan, V., S. Bhattacharya. 2002. Technology selection and com- in R&D: Agency problems and the allocation of engineering
mitment in new product development: The role of uncertainty talent, ideas, and effort by firm size. Management Sci. 40(6)
and design flexibility. Management Sci. 48(3) 313–327. 708–729.