effort, revenue, and cost sharing mechanisms

8/7/2019 Effort, Revenue, and Cost Sharing Mechanisms

1/18

MANAGEMENT SCIENCEArticles in Advance, pp. 118issn 0025-1909 eissn 1526-5501

informs

doi 10.1287/mnsc.1090.1010 2009 INFORMS

Effort, Revenue, and Cost Sharing Mechanisms

for Collaborative New Product DevelopmentSreekumar R. Bhaskaran

Cox School of Business, Southern Methodist University, Dallas, Texas 75205,[email protected]

V. KrishnanRady School of Management, University of California, San Diego, La Jolla, California 92093,

[email protected]

The growing sophistication of component technologies and the rising costs and uncertainties of developingand launching new products require firms to collaborate in the development of new products. However, themanagement of new product development that occurs jointly between firms presents a new set of challengesin 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 haveto 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 withdiffering development capabilities and examine the implications of arrangements that go beyond sharing ofrevenues to include sharing of development cost and work. We term these approaches that involve sharing of thedevelopment 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 sharedbetween firms and the type of development project under consideration. Our analysis shows that investmentand innovation sharing are particularly relevant for products with no preexisting revenues, and their benefitsalso depend on the degree to which revenues are shared between the firms. Whereas investment sharing is moreattractive for new-to-the-world product projects with significant timing uncertainty, innovation sharing plays animportant role in environments where projects experience product quality uncertainty, firms are similar in theircapabilities, and the costs of integration of work across firms can be controlled. Our key contribution involvesthe modeling of joint work and decision making between collaborating firms and unearthing the complementaryrole 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-sciencesand electronics industries.

Key words : codevelopment; project alliances; product and supply chain innovation; technology uncertaintyHistory : Received January 11, 2006; accepted January 15, 2009, by Christoph Loch, R&D and product

development. Published online in Articles in Advance.

1. IntroductionNew product innovation has long been a key avenuefor revenue and profit growth, especially in indus-tries such as life sciences and high technology. Withlifetimes of products shrinking, technical complexityincreasing, and daunting odds of success a norm,firms are forced to invest continually and ever-growing amounts to maintain their competitive edge.In response, some firms have looked beyond their fourwalls to manage the costs and risks of new productdevelopment (Quinn 2000). Pharmaceutical compa-nies increasingly turn to each other and partner withspecialist firms to develop new compounds that costnearly a billion dollars to launch (Grover 1998). Arti-cles trace HPs development of its highly profitableprinter business to joint work with suppliers duringdifferent phases of its business. Even companies with

a strong patent portfolio and research budgets suchas IBM have begun forging joint-development agree-ments to meet the challenges of contemporary prod-ucts and markets (Austin American Statesman 2003).

The increasing uncertainty, complexity of know-how, and costs of product development and distribu-tion require firms to pool their resources and enterinto joint-development contracts involving the sharingof product revenues, development costs, and researchand development (R&D) effort between industrialcustomers and suppliers, which forms the focus ofthis paper. Although intercompany alliance issuessuch as the melding of cultures have received a fairamount of research attention in the strategic manage-ment and organizational behavior literature (reviewedin the next section), the operational challenges ofjoint development are equally complex but relatively

1

notbepostedonany

otherwebsite,includingtheauthor

ssite.Pleasesendanyquestionsregardingthispolicytopermissions

@informs.org.

Published online ahead of print May 14, 2009


2/18

Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product Development2 Management Science, Articles in Advance, pp. 118, 2009 INFORMS

understudied. Unlike manufacturing and assembly,where activities can sometimes be outsourced in anarms-length transfer of prices and quantities, productdevelopment is an uncertain and long-lead-time activ-ity with advance investments and joint work betweenfirms that requires coordination about product fea-

tures and qualities. In consolidating and convergingindustries such as high technology, firms may alsofear that their partners may appropriate a large por-tion of the benefits. The classic product developmentchallenges of managing technology and market uncer-tainty are coupled with agency issues stemming fromopportunistic behavior of development partners.

In this paper, we focus on a specific form of jointdevelopment between two firms that consider sharingthe product revenues, development costs, and devel-opment effort. The product being developed is basedon an uncertain underlying technology/science, andthe firms could differ in their development capabili-

ties as well as their power in the value chain. We nowdiscuss the nuances of joint development in two ofthe industries that we studied.

1.1. Managerial Issues in Collaborative NewProduct Development

Product development plays a crucial role in the bio-pharmaceutical industry, which involves the appli-cation of biotechnology research to the developmentof pharmaceutical compounds. Ever since the inven-tion of recombinant insulin in 1982, the application ofbiotechnology to the creation of therapeutic drugs hasgrown dramatically. Due to the know-how required in

R&D and the large investments needed for commer-cialization, small biotechnology companies often workwith large pharmaceutical companies. To understandcodevelopment in this industry, we surveyed a col-lection of articles and conducted a detailed interview-driven field study of joint development between twofirms, which we will call Alpha Labs (a small publicU.S. biotech company) and Mega Pharmaceuticals (alarge Fortune 100 pharmaceutical company).

In September 2002, Alpha and Mega entered into anagreement to work on the development and commer-cialization of a new innovative class of diabetes drugs.Under the terms of this negotiated agreement, Alpha

and Mega agreed to share equally in the developmentinvestment, with Mega making an upfront investmentbased on an estimate of the development costs. Alphaperformed the drug development, while Mega helpedto commercialize and distribute the product with itsvast global salesforce, and in return the firms agreedto share the revenues (30% of U.S. domestic revenuesand 80% of international revenues accrued to Mega,and the remainder went to Alpha). It is noteworthythat despite its significant development and testingcapabilities and infrastructure, Mega decided to fund

the development work at Alpha rather than sharethe development and testing work. Alpha developerskept detailed records of their time spent in the project,and Alphas alliance manager would report the full-time equivalent of scientist/developer time investedin the project every quarter, upon which Mega would

reimburse Alpha for the expense. We call this pat-tern of multifirm product development where the col-laborating companies share the development expensewith one firm doing the bulk of the work, investmentsharing. This pattern has been observed with severalother firms in the industry as illustrated also in therecent example of Actelion and Glaxo (Whalen 2008).

Consider, on the other hand, product developmentat the firm Dell. Although Dell is known widely asa direct and lean distributor of computers, its abilityto leverage the technological capabilities of its com-ponent suppliers is also noteworthy. Dells productdesign processes are deliberately designed to support

and complement supplier innovation with strengths incustomer needs identification, complementary hard-ware/software development, component qualificationand testing, system integration, beta testing, and mar-ket feedback. For example, Dell worked closely withLexmark in 2002 and enhanced Lexmarks printertechnology with an innovative Dell-developed car-tridge replenishment software, and both firms sharedthe revenues (Financial Times 2003). Dell has also builtits other product lines such as laptops, servers, andstorage products through joint-development arrange-ments with supply chain partners, in which Dellshares downstream development and testing work.

Thomke et al. (1999) discuss how Dell worked withSony to bring to market Sonys lithium ion batterytechnology.

Dells joint-development approach is in contrast tothe investment sharing approach used by Mega inthat it partakes in the development work in areasof qualification, testing, and system integration. Werefer to such an approach in which the develop-ment work is shared without an upfront transfer ofmoney, innovation sharing. Although both investmentand innovation sharing approaches seem to be sig-nals to partners about a firms commitment, the ques-tion arises as to what effect they have on the extent

of product development and profits. What impactdoes development uncertainty have on these cost, rev-enue, and effort sharing agreements between joint-development partners? How should firms split thevalue that is created by their collaborative efforts?What role do their bargaining power and outsideoptions play in the structure and execution of thejoint-development process? These are some of thequestions that constitute our focus in this paper.

We seek to model the interaction between two firmswho embark on an opportunity to invest in a devel-opment project that could improve the performance

notbepostedonany



@informs.org.


3/18

Bhaskaran and Krishnan: Effort, Revenue, and Cost Sharing Mechanisms for Collaborative New Product DevelopmentManagement Science, Articles in Advance, pp. 118, 2009 INFORMS 3

quality of a product but faces uncertainties related tounderlying technology which discourage investment.In addition, the firms may have reservations aboutthe opportunistic behavior of their partners, who mayappropriate the gains from the innovation. How thencan a firm address these reservations and achieve

maximal quality improvement and profits? The two-firm decision-making model and the revenue, effort,and cost sharing mechanisms proposed in this paperare intended to help firms manage collaborative prod-uct development under such conditions of technology,market, and partner uncertainty.

The remainder of this paper is organized as follows.After reviewing the related work in 2, we presenta model of two-firm product development in 3. Wemethodically analyze the different mechanisms forcoordinating multifirm product development in 4,identifying key properties and evaluating their abilityto deal with different types of development projects.

In 5 and 6, we develop the main results of this paper,comparing and contrasting the different mechanismsand demarcating the domains of appropriateness ofthe revenue, cost, and effort sharing mechanisms.Finally, in 7, we summarize the managerial implica-tions using a conceptual framework, relate to exam-ples presented above, and point to directions forfuture work.

2. Relevant LiteratureNew product development has been a topic of sig-nificant research interest in the operations literature

in recent years, but most of the literature on prod-uct development is single-firm-centric with its focusat the level of a project or, at best, a product line. Morerecently, there has been an emerging stream of workon the interactions between product and supply chaindesign decisions (Fine and Whitney 1996, Andersonand Parker 2002, Novak and Eppinger 2001, Ulrichand Ellison 2005, Grahovac and Parker 2002). Eventhis literature focuses only on how a single firmshould make decisions involving its suppliers and noton the interaction between the decisions of firms. Eratand Kavadias (2006) study the development of prod-ucts in an industrial context, but their focus is on

intertemporal discrimination of a technology supplierthrough partial adoption of technology and not on thejoint development of products by multiple firms. Ourwork builds on the model of product developmentunder technology uncertainty from the new prod-uct operations literature (Loch and Terwiesch 1998),but breaks new ground in the area of joint productdevelopment.

The technology management literature considersinstitutional alliances but ignores the role of uncer-tainty in development decision making (Dutta and

Weiss 1997, Amaldoss et al. 2000), or limits itself toinstitutional alliances and the challenge of integrat-ing cultures in such ventures (Doz and Hamel 1998).Despite the challenges of contractual joint devel-opment, many products do not warrant the hugeinvestments in time and money required for institu-

tional alliances, such as joint ventures. Managementof institutional alliances often requires the establish-ment of a separate organization with investmentsof hundreds of millions of dollars as well as theblending of established organizational cultures (Dozand Hamel 1998). Although there are a few high-profile success stories of joint ventures, their mixedrecord (as evidenced by examples such as the Iridiumalliance or the IBM-Motorola RISC technology ven-ture) and their high investments make firms pause(before spending large amounts of money in institu-tional alliances) and look for alternative approaches tojoint development. Contractual joint development

that begins with the recognition that a supplier thatinvests in component technological innovation cannottake for granted the possibility of both the develop-ment succeeding and other value-chain participantsmaking mutually aligned decisions if the innovationmaterializesseeks to achieve the benefits of collab-oration without the significant fixed costs of institu-tional alliances (Gerwin and Ferris 2004). Customerfirms that wish to tap into the technological know-how of their suppliers have to develop mechanismsto address these supplier concerns, stimulate inno-vation, and thereby realize the full potential of thecollaborative development process. Gerwin and Ferris

(2004) provide a qualitative discussion of a taxonomyof alliances in product development projects, high-lighting the role of contractual alliances. Dyer (2003)paints a rich portrait of the weaknesses of both ver-tical integration and arms-length supplier manage-ment using examples from Chrysler and Toyota, butfocuses more broadly on the issue of trust than on thespecifics of sharing effort and development.

Several papers in economics, marketing, and supplychain operations that study the interaction betweenvertical firms have proposed mechanisms to alle-viate tactical (price-quantity) coordination problems(Jeuland and Shugan 1983, Lee and Staelin 1997,McGuire and Staelin 1983, Aghion and Tirole 1994,Grossman and Hart 1986, Cachon and Lariviere 2005,Spencer and Brander 1992, Cvsa and Gilbert 2002).Attempts have also been made to use similar mod-els to analyze the effect of innovation of one of thefirms on its channel partners. Gupta and Loulou (1998)study how interactions between firms in a channelaffect innovation. Gilbert and Cvsa (2003) analyze theeffect of strategic commitment to price by a supplierto stimulate downstream innovation in a supply chain.However, this stream of literature deals mostly with

notbepostedonany



@informs.org.


4/18


process innovations and tactical decisions like pricesand quantities, and ignores the effect of technologi-cal uncertainty on firms decision making. Our paper,focused on collaborative product innovation, borrowssome modeling elements from this stream, but dif-fers significantly in that it emphasizes uncertainty of

technologies and goes beyond interfirm coordinationbased on prices and quantities. We evaluate productdevelopment decision making that focuses on productquality improvements through cost or effort sharingarrangements between firms and examine how firmsshould incorporate the underlying technological andmarket parameters of an industry while entering intopartnerships.

3. The ModelIn this section, we model the collaborative prod-uct development process between two firms thatmust decide about sharing the revenues and costs

of an innovation. This innovation along a certaintechnological dimension could improve the perfor-mance quality of the focal product and has thepotential to enhance end-customer demand. How-ever, the innovation itself is fraught with risks andrequires investments in time and money. We modelthe development/innovation process as comprisingof two distinct phasesone involving the develop-ment of a key underlying technology, and the secondinvolving the packaging and integration of this tech-nology into a new product/systemhandled by twoindependent firms; a focal firm (FF) and a partner firm(PF). Note that the complementary nature of theseactivities and specialized skills that these activitiesentail imply that the two firms have to come togetherto partner in development to realize the value ofthe innovation. To be general, we have chosen neu-tral terminology, although in specific situations (suchas the examples discussed earlier), one of the firmsmay have an expertise in certain aspects of technol-ogy/product development. The two firms could sharethe revenues as well as the development expense orthe effort, which we refer to as investment and inno-vation sharing, respectively.

3.1. Model of InnovationThe innovation we model is such that the improve-ment in quality does not increase a firms marginalcosts. Abbott (1953) terms such improvements inno-vational quality dimensionsthe introduction of anovel quality that is judged superior by most or allbuyers and costs no more to produce, thus mak-ing the older quality obsolete. However, the innova-tion entails costs: a firm decides the level of qualityimprovement (innovation) that it wants to achieve,and it incurs both a fixed cost of development and avariable time-dependent cost of resource deployment

to achieve this quality improvement. In addition,when the development effort is distributed betweenthe firms, integration costs are incurred as discussedlater in this section.

The fixed cost of development is an upfront invest-ment and is a function of the level of quality improve-

ment .1

We model that this cost is of the form I

,where I is an investment parameter. We assumethat > 1, implying that the cost of innovation isconvex with respect to (w.r.t.) . Convexity is not adriving factor for much of our analysis (specificallynot impacting revenue sharing and investment shar-ing), and it does have empirical backing in literature;see, for example, Cohen and Klepper (1992, 1996) andKamien and Schwartz (1992). Convex costs are oftenattributed to diminishing returns from R&D expendi-tures or to diseconomies of scale that, in practice, canbe linked to bureaucracies in a larger firm that stiflecreativity and impede innovation. Also, as the level

of technological innovation increases, diseconomies inoffering employment contracts could also lead to dis-economies of scale in R&D (Zenger 1994). Throughoutthe body of this paper, we assume that the convexityparameter = 2. However, directionality of the resultsis preserved for any > 1.

In addition to the fixed development cost, a firmwould also have to dedicate resources to the tech-nology development process. To accommodate thefact that higher levels of quality improvement requiremore resources, we assume that these costs are pro-portional to the level of quality improvement .Also, because the firm would have to dedicate these

resources throughout the course of the developmentprocess, we assume that these costs are proportionalto the development time. Specifically, we model thesecosts as ct , where c is a constant and t is the totaltime taken for technology and product development.Thus, the total fixed and variable costs of develop-ment (not including integration costs) for a firm thatimproves the quality by would be

T C= I 2 + ct (1)

Investments in quality improvement enable the firmto command higher margins and increase profitability,

which we capture by assuming that there is a one-to-one correspondence between the quality of the finalproduct and the subsequent revenues that firms areable to generate in the end-product market. In partic-ular, we model that an innovation level generates atotal additional value of V= r . If the residual valueof the product before the innovation is , then the

1 Because there is a one-to-one correspondence between the levelof innovation and the level of quality improvement, we use boththese terms interchangeably.

notbepostedonany



@informs.org.


5/18


total value after the innovation is successful can berepresented as T V = + r . When = 0, the finalvalue of the product depends entirely on the inno-vation, and we refer to these projects as new-revenueprojects. In contrast, when > 0, the value of theproject does not depend entirely on the innovation

that is undertaken. Such projects can be considered asreplacement-revenue projects.

3.2. Model of Interfirm InteractionFirms can engage in joint development in numer-ous ways. To begin making a contribution in mod-eling joint development, we start with specific formsof revenue and cost sharing between firms inspiredby industrial examples presented earlier. To main-tain focus on how product development issues influ-ence the collaboration between firms, we consider abaseline case in which the focal and partner firmsare currently in a revenue sharing agreement wherein

the focal firm gets a fraction of the total rev-enue (which can be decided endogenously or can bean exogenous parameter), whereas the partner firmreceives (1) of the revenue. Revenue sharing hasbeen shown to be a useful mechanism for coordi-nation in the supply chain literature that deals pri-marily with postlaunch price-quantity decisions; ourinterest here is in understanding how it combineswith prelaunch cost sharing and influences qualityand development investment decisions. In addition,to a pure revenue sharing agreement, firms can alsoenter into an investment sharing or an innovation shar-ing agreement. Under investment sharing, firms share

the cost of product development, whereas under inno-vation sharing, they share the innovation itself, withpart of the work being done by each of the firms.

At first, we analyze the case in which the revenuesharing parameter is an exogenous parameter,which could be the case in mature markets wherepreexisting relationships or channel power equationscould form the basis of how revenues are sharedbetween collaborating firms. In contrast, negotiationbetween innovating firms could form the basis ofsuch division of revenues in emerging markets. Toaccommodate such markets, we also consider the casein which the equilibrium revenue sharing parame-

ter is determined as a function of the bargain-ing between collaborating firms. Subsequently, weexamine the impact of this bargaining frameworkon pure revenue, investment, and innovation sharingagreements. The complementary roles of FF and PFimply that the firms have to work together to realizethe value of the project. The central questions thenbecome what mechanism should be used to opera-tionalize codevelopment and how the value realizedfrom codevelopment should be shared between firms.In all of these analyses, we assume that firms agree

Figure 1 Decision Timeline of the Codevelopment Process

Focal firm decides the mode ofcodevelopment

Firms determine cost/revenue sharing termsthrough Nash bargaining

Optimal investment levels are determinedand undertaken

Technology uncertainty resolved and projectvalue realized

to use the bargaining structure proposed by Nash(1950, 1953) to determine the optimal contract terms.In a Nash bargaining game, two players cooperativelydecide to how to split the surplus that occurs as a

result of their interaction. The manner in which thesurplus is split depends both on the utility functionsof both players as well as the value of their out-side option (the value they are able to obtain if theychoose not bargain with each other). Please refer toOwen (1995) for more details of the bargaining pro-cess. Initially, we assume that the outside option ofboth firms are the same and, without loss of gener-ality, normalize this value to zero. Subsequently, werelax this assumption and numerically evaluate theeffect of different outside options for FF and PF onthe optimal mode of codevelopment.

The sequence of decision making, represented in

Figure 1, is as follows. At first, FF decides which of thecodevelopment mechanisms should be used to con-duct the development project. The firms then coop-eratively determine how the revenues and costs ofthe innovation should be shared. Subsequently, theinvestment for the development project is made, thevalue of which is realized at the end of the develop-ment process once technology uncertainty is resolved.

To understand how each of these agreementsimpacts the technology development process anda firms costs and revenues, let us examine theseagreements more closely. When firms enter into aninvestment sharing agreement, one firm conducts the

development work while the other agrees to bear partof the development costs. Let k be the fraction ofdevelopment costs borne by the focal firm. Similarto pure revenue sharing, the technology developmentitself could be conducted by the focal or partner firm,which we analyze in detail in the next section.

Innovation sharing entails the splitting of the devel-opment work across both firms, so its form and func-tion differ in a subtle fashion from investment sharing.Specifically, a part of the development work F is doneby the focal firm (who incurs a fixed cost I 2F), and a

notbepostedonany



@informs.org.


6/18


7/18


both investment and innovation sharing, more deci-sions about investments have to be made in conjunc-tion with the innovation levels, as discussed below.Without loss of generality, we assume that the focalfirm makes the decision on which mechanism for jointdevelopment has to be chosen. Depending upon this

choice, the decisions and implementation of contrac-tual development could take different forms.Under an investment sharing agreement, the deci-

sion on the level or degree of investment sharing (k)should be made before deciding the level of inno-vation (). Thus, when a firm determines its inno-vation level, it knows the share of costs it wouldreceive from its partner, and hence its decision onthe level of innovation would accommodate this costsharing level. (It is trivial to note that determiningk after will not have any impact of the innova-tion level or profits.) Either the focal or partner firmcould conduct the innovation while letting its part-

ner share a part of the development costs, so therecan be two forms of investment sharing dependingon who determines the level of investment sharing kand who determines the product quality . In con-trast, under innovation sharing, the level of invest-ments F and P can be determined simultaneously.Although there are other forms of innovation shar-ing in which these investment levels are determinedsequentially, it can be easily shown that under theassumption that the integration cost is of the form asin Equation (2), simultaneous and sequential forms ofinnovation sharing yield identical investment levelsand profits.4

Our model is driven by the assumption that firmscan infer product development investments fromquality improvements in the end product. Approachessuch as the function-point method in software devel-opment and project journaling make estimation ofproduct development effort possible for a certain levelof product quality and are being increasingly usedin joint development projects. It is worth pointingout the manner in which collaboration was formal-ized and contracted in the joint development relation-ship between Alpha and Mega cited earlier. A jointteam from both firms kept track of the man-hoursspent by Alpha on every aspect of the project, and

Mega compensated Alpha on quarterly basis for theeffort according to a contractual agreement. In addi-tion, we assume that firms do not have informationabout whether the new technology would succeed,but they do know their own capabilities and thoseof their partners. The implication of these assump-tions is that firms can contract on the level of innova-tion and in turn the extent of development cost thatthey undertake without incurring any agency costs.

4 Proof is available from the authors upon request.

Although contractual agreements on these parame-ters might be imperfect in some cases, it is certainlypossible in many of the industrial settings we haveconsidered in this paper, where quality improvementsare observable and measurable, and development costcurves of firms have known trajectories. Specifically,

this is applicable in cases where quality improvementscould be broken down to attributes such as battery life,processor speed, and drug efficacy. There is an increas-ing tendency for firms to engage outside firms (in par-ticular, industrial design firms such as IDEO andcontract development organizations such as CharlesRiver Labs), which suggests that product developmentwork may be contractible. In summary, the model-ing assumptions we make about information structureand decision sequence are meant to derive first-orderinsights and are motivated by industrial practice.

In all of our analyses, we assume that the launchcosts and marginal costs of production are constant,

and for ease of exposition we normalize these coststo zero. Although the assumption of linear produc-tion cost is simplistic, the fundamental interactionbetween the different product development strategiesconsidered requires only that these costs be nonde-creasing. We also assume that firms are risk neu-tral and are interested in maximizing their expectedprofits. In addition, they have to make their invest-ment into the technology before uncertainty associ-ated with the technology is resolved. This would berealistic in conditions where lead times of innovationsare much higher than normal production times, whichis the case for technology and knowledge intensive

industries.

4. AnalysisIn this section, we evaluate the different productdevelopment and collaboration choices of firms. Atfirst, we look at the case in which the firms onlyhave a revenue sharing agreement and evaluatehow this agreement influences product developmentinvestments. Subsequently, we derive the effect ofinvestment and innovation sharing agreements on theoptimal development investment. Finally, we endog-enize the choice of revenue sharing whose structure

embodies a bargaining framework between the twocollaborating firms.

4.1. Codevelopment Under Pure Revenue SharingLet us first consider the case in which investment lev-els are decided exclusively through revenue sharingwithout any additional cost or effort sharing betweenthe firms (pure revenue sharing case). Only one ofthe firms does the development work, and the deci-sion on the quality level is also made by that firm.In this setup, the value offered by the other firm thatjustifies revenue sharing could be in complementary

notbepostedonany



@informs.org.


8/18


value-chain activities such as supply base or channelmanagement. The case in which the focal firm doesdevelopment work is referred to as FF revenue shar-ing, and the case in which partner firm does develop-ment work is referred to as PF revenue sharing.

Let us first consider the case when the focal firm

does the development work. Once the technology hasbeen successfully developed, the revenues are dividedbetween the firms according to a revenue sharingagreement, . The focal and partner firms profits fora given innovation level can then be represented as

F =+ vr (3)

P = 1+ vr (4)

Note that this is the profit of the focal firm if thenew product has succeeded. However, the decisionon has to be made before it is known how wellthe innovation can be translated to a product or

how much time the development process would take.After accounting for these elements of technologyuncertainty, the expected profit function of FF can berepresented as follows:

F = E+ vr ctF I2 (5)

=2+ 1+ vr

2

c2

F I 2 (6)

Differentiating the profit function w.r.t. , we can nowcalculate the optimal investment level F of the focalfirm. First-order conditions (sufficient due to profitfunction concavity) yield

F =1+ vrF

4I F + c (7)

In a similar fashion, we can also find the optimallevel of innovation P under partner revenue sharingin which innovation decision is made by the partnerfirm instead of focal firm:

P=11+ vrP

4I P+ c (8)

A casual observation of Equations (7) and (8) showsthat the investment levels under pure revenue sharingare lower than under centralized decision making.5

This distortion is because part of the benefits of theinnovation accrues to a firms partner, which reducesthe marginal value of the development investment.We begin by evaluating the profits of both firms whenthe revenue sharing parameter is determined exoge-nously. Result 1 below characterizes when and howfirms engage in pure revenue sharing under this case.

5 Under centralized decision making, it can be see that i =1 + vri/4I i + c (proof available from the authors uponrequest).

Result 1. (a) There exists a threshold R on the devel-opment capability ratio such that if > R, it is optimalfor the focal firm to conduct innovation where

R =1c

2c I P1 3 (9)

(b) The threshold R is decreasing in .

Proof. All proofs are provided in TechnicalAppendix I (provided in the e-companion).

For starters, Result 1 simply confirms our intuitiveunderstanding that who conducts innovation in purerevenue sharing depends on the relative developmentcapability of the firms. If the focal firms develop-ment capability is sufficiently higher than the partner,it prefers doing the innovation itself. The focal firm isalso better served by conducting the innovation itselfwhen the revenue sharing parameter increases, andas a result R is decreasing in .

We now examine the more complex case whenthe revenue sharing parameter is endogenouslydecided by the firms. Recall that we use the bargain-ing structure proposed by Nash (1950, 1953) to deter-mine the equilibrium . For ease of exposition, wedefine R to be equilibrium share of revenues of thefirm who does the development work (also called theinnovating firm).

Proposition 1. (a) There exists a solution to the Nashbargaining problem such that R

12

34

.(b) R is decreasing in the products residual value ()

and increasing in the development capability of the inno-

vating firm.(c) When = 0, R =34

.

Proposition 1 is interesting in that it shows that thefirm that does the development work is able to retaina larger share (greater than 50%) of the total revenue.This implies that doing the development work implic-itly provides the innovating firm a better bargain-ing position in its negotiation with its partner, whichit is able to parlay into a larger share of the totalprofits for itself. The share of revenues R is high-est when = 0 (new-revenue projects), in which caseR =

34

. As increases, R decreases, which illustratesthe interaction between the nature of the project andconsequent bargaining between firms. To understandthis interaction, note that the value of innovation ismost when is small (not much preexisting revenue).This is because for smaller values of , most of thevalue that firms attempt to share between them is asa result of the innovation. Naturally, it is most valu-able to provide incentives to the innovating firm toincrease investment when is small. This is achievedthrough offering the innovating firm a larger share ofthe total revenue. However, as increases, the inno-vation is relatively less valuable and consequently, it

notbepostedonany



@informs.org.


9/18


becomes less important to share a larger fraction ofthe total revenue to the innovating firm. As a result,R is decreasing in .

It can also be seen that the development capabilityof the firm that does the innovation also affects howthe revenues are shared between the firms. As the

development capability increases, a firm increasesthe amount of investment it undertakes (becausethe marginal value of investment increases). Becausethis increases the relative value of the innovationvis--vis , it is able to retain a larger share of thetotal revenue. As is to be expected, the equilibriumrevenue sharing value R is increasing in the devel-opment capability of the innovating firm.

Having characterized the effect of revenue sharingagreements on investment levels and profitability, wenow turn our attention to the cases when firms alsoengage in cost or effort sharing agreements in addi-tion to revenue sharing. First we look at investment

sharing and its impact on investments and profits, fol-lowed by innovation sharing.

4.2. Investment SharingUnder investment sharing, development work is doneat one of the firms, with its partner sharing the invest-ment cost. Two different forms of investment sharingemerge. The case in which the focal firm conductsinnovation is referred to as FF investment sharing.When the partner firm does development work andthe focal firm in turn cofinances this development, werefer to it as PF investment sharing. It follows thatthere are two decisions to be made under investment

sharing, namely, the level of investment sharing andthe level of quality improvement. Let k be the share oftotal development costs that the focal firm would bearand be the development work undertaken by theinnovating firm. As under revenue sharing analysis,the optimal decisions of both firms can be character-ized by backward induction. The following startingresult characterizes the effect (an exogenous) revenuesharing parameter and the development capabil-ity have on the relative attractiveness of these twomechanisms.

Result 2. (a) FF investment sharing is feasible iff F and PF investment sharing is feasible iffP.

(b) There exists a threshold IS on the developmentcapability ratio such that if > IS , then FF investmentsharing dominates PF investment sharing where

IS =2c12

c22 +PI 2+ 32 4

(10)

(c) The threshold IS is decreasing in iff >12

.

It can be seen above that revenue sharing ratio strongly impacts the viability of investment sharing.We find that investment sharing is feasible only for

certain ranges of revenue sharing ratio . In partic-ular, when is high, the focal firm obtains a largershare of the total profits compared to the partner firm,which limits the incentive of the partner to cofinancethe development by the focal firm. When > F, theincentive of the partner to cofinance the focal firms

investment costs erodes to such an extent that theoptimal investment sharing level goes to zero. Wesee a similar effect in PF investment sharing as wellwhen is low. However, the advantage of investmentsharing lies in that by sharing a part of the develop-ment costs, a firm is able to incentivize its partner toinvest more into the innovation, and this can be seenby examining the profit function under investmentsharing in the proof of Proposition 2 (in TechnicalAppendix I). Result 2 also shows how the choice fora specific form of investment sharing depends on therelative development capability of firms. As expected,a firm would prefer to have its partner conduct devel-

opment work while cofinancing this investment onlyif the capability of its partner is sufficiently high.When the capability of its partner is low compared toitself, allowing the partner to undertake developmentresults in lower levels of quality improvement, andfor sufficiently low , a firm is better off doing thedevelopment itself.

Similar to revenue sharing, the choice between thedifferent forms of investment sharing is also influ-enced by how the revenues are split between thefirms. However, unlike under pure revenue sharing,we find that the threshold on capability above whicha firm would like to conduct the innovation on its

own is decreasing in only if > 12 . Note that when > 1

2, the focal firm receives a larger share of the total

revenue vis--vis its partner. A further reduction in in this range ( > 1

2) reduces the partner firms profits

even more, and hence reduces its incentives to sharethe development costs. Because of this, the focal firmis better served by assuming Stackelberg leadershipand does so by choosing PF investment sharing. Itfollows that IS is decreasing in only if >

12

.Although the above analysis is useful in under-

standing interactions between development invest-ments and cost sharing in established markets whereexisting contractual agreements could form the basisof codevelopment choices, negotiations between col-laborating parties could determine the revenue shar-ing parameter in emerging product markets. We nowturn our attention to such markets and examine howthese mechanisms are affected when is endogenizedusing the Nash bargaining structure proposed earlier.In contrast to revenue sharing, note that the level ofinvestment sharing (k) must also be decided, mak-ing this decision part of the bargaining structure. LetF and

P represent the equilibrium levels of rev-

enue sharing under FF and PF investment sharing,

notbepostedonany



@informs.org.


10/18


and let kT and kP represent the corresponding invest-

ment sharing levels.

Proposition 2. The equilibrium level of revenue shar-ing and investment sharing are as follows:

F =P=

12

kF = kP=

12

Proposition 2 is interesting in that when and k aredetermined through bargaining between firms, bothcosts and revenues are shared equally. This is quiteunlike under revenue sharing where a larger shareof the profits gets captured by the firm that does thedevelopment work. By engaging in investment shar-ing and splitting the development costs with its part-ner, a firm is able to achieve two objectives: It is ableto induce its partner to invest in higher quality lev-els. In addition, it is able to improve its bargaining

position in its negotiation with its partner and retaina larger share of the total profits. The fact that rev-enues and costs are shared equally is also interesting.This is because the equilibrium is set in a mannerthat provides the partnering firms adequate incen-tives to both invest and cofinance. In fact, when costsand revenues are shared in the same ratio, the distor-tions created due to multiparty decision making areeliminated, as a result of which sharing the costs andrevenues equally becomes the equilibrium solution.

It is also worthwhile to point out that the equilib-rium revenue sharing ratio is independent of both and the development capability of the innovating

firm, which is another point of departure from rev-enue sharing. This is again because of the fact that anychanges in both and development capability are feltequally by both firms because they share not only therevenues of the innovation but also the correspond-ing costs. Moreover, this difference also points to theability of investment sharing as mechanism to alignincentives of both firms as opposed to a pure revenuesharing agreement where only revenues are shared.

4.3. Innovation SharingUnder innovation sharing, a part of the developmentwork is conducted by the focal firm and the rest is

conducted by its partner. Let F and P be the devel-opment work conducted by focal firm and partnerfirm, respectively. Recall that we assume that thesedecisions on investment levels are made simultane-ously. Once the innovation materializes, firms incuran integration cost CF P = K

2F +

2P to bring

together the quality improvements distributed acrossboth firms. As before, we first examine the conditionsunder which innovation sharing is a feasible mecha-nism for the partnering firms.

Result 3. Innovation sharing is feasible iff .

We see that innovation sharing is feasible only forintermediate values of . This is because innovationsharing entails both firms to invest in developmentand subsequently incur the integration costs. As aresult, if the integration cost parameter K is very high,the corresponding share of revenues that the firm

would receive would not be sufficient to compensatethe firm for the development costs that it would incur.Naturally, it will choose to invest in development onlyif its share of revenues is sufficiently high. If istoo low, then the focal firm does not get compen-sated adequately for its investment, and if is toohigh, then the partner does not get compensated ade-quately. In both these cases, innovation sharing wouldbe infeasible.

Having determined the effect of on the firmsincentive to engage in innovation sharing, we nowexamine what happens when is endogenized. Inthe following proposition, we characterize the equi-

librium revenue sharing level under innovation shar-ing and how this level is affected by the developmentcapability of firms.

Proposition 3. The equilibrium level of revenue shar-ing under innovation sharing IN can be represented asfollows:

(a) If = 1, then IN =12

.(b) If < 1, then there exists a solution to the Nash

bargaining problem such that IN 14

12

.(c) If > 1, then there exists a solution to the Nash

bargaining problem such that IN 12

34

.

In contrast to investment sharing, the equilib-

rium under innovation sharing is not always 12 .In fact, = 1

2only when the development capa-

bilities of both firms are the same, i.e., = 1. Thisis because, unlike under investment sharing or rev-enue sharing, both firms are involved directly inthe development work when they engage in inno-vation sharing. Coincidentally, the extent of bargain-ing powerand the corresponding share of revenuesthat this power providesdepends on the amount ofdevelopment work undertaken by a firm. When thedevelopment capabilities of firms are the same, theyinvest exactly the same amount, and hence the rev-enues are also shared equally. However, when thedevelopment capabilities of the firms are not the same,the firm with higher capability seems to have a greaterincentive to undertake more development work thanits partner, thereby also earning a higher share ofrevenue.

5. Comparison of CodevelopmentMechanisms

Thus far, we have focused our efforts on identifyinghow cost and effort sharing mechanisms should be

notbepostedonany



@informs.org.


11/18


implemented. However, as we saw in the industrialexamples discussed earlier, firms in different contextstypically use one of these mechanisms to managethe joint development process. This raises the impor-tant question as to whether there are conditionsbothmarket and project relatedthat make these mecha-

nisms appropriate for different conditions. To answerthis question, we now look at how these mechanismscompare against each other and characterize the con-ditions of appropriateness of each of these mecha-nisms in Propositions 47.

Proposition 4 (Effect of Endogenous RevenueSharing (Bargaining) on Codevelopment Mecha-nisms):

(a) PF revenue sharing is dominated by the other rev-enue and cost sharing mechanisms and is never optimal atequilibrium.

(b) FF investment sharing is dominated by the otherrevenue and cost sharing mechanisms and is never optimalat equilibrium.

Proposition 4 shows that if the focal firm were tochoose investment sharing as the mechanism to col-laborate with its partner, it should let the partnerdo the innovation and cofinance the development.However, if pure revenue sharing is the mecha-nism that is chosen, it should rather conduct thedevelopment itself. This is because of the interac-tion between the type of mechanism that is chosenand the resultant bargaining between firms to deter-mine the equilibrium revenue sharing level . Recallthat under investment sharing revenues are shared

equally, whereas with pure revenue sharing the firmthat conducts the innovation is able to retain a largerportion (50%75%) of the total revenue. The subop-timality of FF investment sharing and PF revenuesharing is primarily because of this reason. If thefocal firm were to seek cofinancing from its part-ner when it does the development work, it wouldhave to forgo a significant portion of the revenues inthe process at the bargaining stage. Thus, althoughinvestment sharing results in better allocation of costsand, hence, potentially higher development invest-ment, the resultant lower bargaining power makesa firm less willing to consider it for collaboration.As a result, FF revenue sharing always dominates FFinvestment sharing. However, when at equilibriumit is the partner who does the innovation, it wouldbe in a firms best interests to engage in investmentsharing with its partner. The focal firm achieves twoobjectives by doing this: it is able to ensure that thepartner invests sufficiently high amount; in addition,investment sharing provides it the bargaining powerto obtain a larger share of the total profits. As a result,PF investment sharing dominates PF revenue shar-ing. Due to these dominance results, from here on

we will refer to PF investment sharing as investmentsharing and FF revenue sharing as revenue sharing.We now examine the effect of timing uncertainty on afirms preference for cost and effort sharing. BecauseT and P represent a firms ability to manage tim-ing uncertainty, examining the choice among different

mechanisms as a function of these parameters allowsus to capture the effect of timing uncertainty.

Proposition 5 (Effect of Development Capabil-ity and Timing Uncertainty):

(a) There exists a threshold on the development capabil-ity BIS such that if <

BIS , then investment sharing is the

dominant mechanism.(b) There exists a threshold on development capabil-

ity BR such that if > BR, revenue sharing is the dominant

mechanism.(c) When = 1, there exists a threshold on the inte-

gration cost K such that if K < K, innovation sharingdominates investment sharing and revenue sharing.

Proposition 5 shows that the preferred form ofcodevelopment depends critically on the develop-ment capability and associated timing uncertainty.In particular, investment sharing is the dominantmechanism when the development capability of thefocal firm is low, and revenue sharing is the dominantmechanism when its capability is high. When devel-opment capability falls in the intermediate range,innovation sharing is the dominant mechanism.

One of the reasons why revenue sharing is thepreferred mechanism when the focal firms capabil-ity is high is because an innovating firm is able

to retain a larger share of the total revenue underrevenue sharing. Because the level of investment isdetermined by the firm that gets the larger share ofrevenues (in this case, the focal firm), it does notresult in that much inefficiency/quality deteriorationcompared with investment sharing. More importantly,because the focal firms capability is also sufficientlyhigh, there isnt much of a penalty for lack of col-laboration. Although investment sharing would haveenabled better allocation of costs, it would require thefirm to forgo a large fraction of the revenues, so itceases to be a viable option. However, when the firmscapability is low compared with its partners (low ),it becomes important to ensure that the developmentwork is conducted by the partner firm. By engagingin investment sharing when is low, a firm is ableto ensure that its partners investment is sufficientlyhigh in addition to obtaining a better bargaining posi-tion to retain a large enough share of the total profits.As a result, investment sharing becomes the optimalmechanism when is low.

More interestingly, we find that innovation shar-ing dominates both investment sharing and rev-enue sharing when is in the intermediate range.

notbepostedonany



@informs.org.


12/18


This is because the economies of specialization thatinnovation sharing benefits from enable each firm toconcentrate on aspects that are closer to its core capa-bility and ensure higher levels of innovation withlower investment costs. When capabilities of bothfirms are comparable, it benefits to resort to innovation

sharing, have the development work split betweenboth firms, and exploit the economies of specializationthat this division provides. However, such a divisionof effort also comes with an explicit cost in the formof integration costs. If the integration cost parameterK is too high, the benefits of specialization do not ade-quately compensate for the higher integration costs,and innovation sharing ceases to be optimal.

Now let us consider the impact of translationaluncertainty on the choice of mechanisms. To this end,we let the development capability of both firms Fand P to tend to , or let the resource cost param-eter c tend to 0. Because firms face no timing uncer-

tainty either when their development capability isvery high or when the resource costs are negligible,the case in which T P or c = 0 can be used tocharacterize the effects of translational uncertainty.

Proposition 6 (Effect of Translational Uncer-tainty). When T P, or = 0:

(a) Investment sharing is dominated by innovationsharing and revenue sharing.

(b) There exists a threshold on the integration cost Kcsuch that is K < Kc, innovation sharing is the dominantmechanism.

The primary reason for the nonoptimality of invest-

ment sharing under translational uncertainty is itseffect on the revenue sharing negotiation betweenfirms. In our model, when there is no timing uncer-tainty, the development capability of firms does notimpact which mechanism is preferred. As result, rev-enue sharing, which allows the focal firm to obtaina larger share of the total revenue, dominates invest-ment sharing. However, if the integration costs aresufficiently low, then innovation sharing becomes thedominant mechanism. This is because the distributedeffort and cost function convexity under innovationsharing reduces the development cost and providesadequate incentives for innovating firms to invest

in greater levels of product quality improvement.Thus, translational uncertainty lends itself to inno-vation sharing, and timing uncertainty favors invest-ment sharing.

Proposition 7. When investment sharing is the dom-inant mechanism, the combined profits of both firms matchthat of a centralized decision-making setup in which devel-opment work is conducted by a single firm.

Proposition 7 is useful in that it shows invest-ment sharing mechanism can replicate a centralizeddecision-making setup because revenues and costs are

shared equally between the firms (Proposition 2). Dis-tortions associated with a multiparty decision-makingsetup are eliminated because revenue and cost impli-cations of any investment are borne equally by bothfirms. This reduces any incentive by the innovatingfirm to underinvest, which is the primary reason for

the inefficiency in a pure revenue sharing mechanism.As a result, the combined profits of both firms underinvestment sharing could be the same as that of a cen-tralized decision-making setup.

6. Sensitivity Analysis ofCodevelopment Mechanisms

To understand the sensitivity of the above resultsabout the relative attractiveness of investment andinnovation sharing to changes in market and techno-logical parameters, we discuss the results of numer-ical analysis, which illustrate the effects of timinguncertainty, translational uncertainty, and nature of

technology on the codevelopment mechanism a firmshould adopt. This analysis leads to four observationsthat are valid for a large range of technological andmarket parameters. In all of the analyses depicted asfigures, the vertical axes of the figures represent theprofits of the focal firm for the different mechanismsand the horizontal axes represent the developmentcapability ratio of the firms ().

In the first observation below, we examine how thepreferred codevelopment mechanism is affected byan increase in timing uncertainty. The results of thisnumerical analysis are represented in the two graphsthat form Figure 2. In the graph on the left side of thisfigure, we represent the scenario when the resourcecosts are low (c = 1), whereas on the right the resourcecosts are higher (c = 2). Because the effects of tim-ing uncertainty are greater when resource costs arehigher, an increase in the resource costs (c = 1 to c = 2)allows us to characterize a firms optimal response toan increase in timing uncertainty.

Observation 1. An increase in the effect of timinguncertainty increases the value of codevelopment usinginvestment sharing and reduces the relative value of purerevenue sharing.

As seen in Figure 2, an increase in timing uncer-tainty makes codevelopment using either investmentsharing or innovation sharing more valuable thana pure revenue sharing arrangement. This can beinferred by comparing the two graphs in Figure 2 andconsidering the range of the development capabilityfor which both investment and innovation sharing areoptimal. As we move to the right, the shaded region,which corresponds to the region of dominance of rev-enue sharing, shrinks in size, and the threshold devel-opment capability below which investment sharingis attractive increases (implying that an increase in

notbepostedonany



@informs.org.


13/18


Figure 2 Effect of Timing Uncertainty

0.80

0.75

0.70

0.65

0.75

0.70

0.65

0.55

0.60

0.50

Equilibriumprofits

Equilibriumprofits

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

Investment sharing

Innovation sharing

Revenue sharing

Profits under purerevenue sharing

Profits underinnovation sharing

Profits underinvestment sharing

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

timing uncertainty would make investment sharingmore attractive). Because innovation sharing involvesdevelopment work being conducted by both firms,

the expected development time required under inno-vation sharing relative to investment could be higher,particularly when the focal firms development capa-bility is low relative to that of its partner. This alsoimplies that resource allocation costs for innovationsharing would be higher compared with investmentsharing. As a result, an increase in c affects the viabil-ity of innovation sharing more than it affects invest-ment sharing. It follows that the optimal response toan increase in c would be to engage in investmentsharing even for higher values of .

This result of increasing attractiveness of invest-ment sharing, however, does not carry over to the case

when the development capability of the focal firm ishigh. Recall from Proposition 5 that when the devel-opment capability of a firm is high, it is optimal toeither engage in pure revenue sharing or innovationsharing. Under such situations, an increase in c actu-ally pushes a firm toward innovation sharing in placeof pure revenue sharing. This can again be observed

Figure 3 New-Revenue vs. Replacement-Revenue Projects

0.75

0.70

0.85

0.80

0.75

0.70

0.65

0.55

0.60

0.50

Equilibriumprofits

Equilibriumprofits

0.4 0.6 0.8 1.0 1.2 1.4

Development capability ratio

0.4 0.6 0.8 1.0 1.2 1.4


Investment sharing

Innovation sharing

Revenue sharing




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

by comparing the two graphs in Figure 2, where wesee that the threshold of below which innovationsharing is optimal is lower when the resource costs

are higher (c = 2). This is because incentive alignmentbenefits that innovation sharing provide comparedwith revenue sharing results in higher developmentinvestments from both firms. When the resource costsincrease, the relative value of such an incentive align-ment also increases because of the higher marginalcosts of development. In addition, the higher devel-opment time makes splitting the fixed costs betweenfirms more valuable, and innovation sharing becomesmore attractive compared with revenue sharing.

Two other project dimensions that influence a firmspreferred mode of codevelopment are the nature ofthe revenues of the development project and the

degree of translational uncertainty. In Figure 3, welook at how a change in (which captures the extentto which a development project generates new rev-enues) affects the optimal codevelopment mechanism.

Observation 2. For new-revenue projects, investmentand innovation sharing are more attractive codevelopmentapproaches than pure revenue sharing.

notbepostedonany



@informs.org.


14/18


New-revenue projects have the most to gainwhen firms engage in codevelopment through invest-ment and innovation sharing. This is illustratedthrough a comparison of the focal firms optimal co-development strategy for different values of , theanalysis of which is represented in the two graphs

in Figure 3. In the graph on the left, = 0, whichcorresponds to projects with no preexisting revenue(all revenues accrue from this innovation), whereasin the graph on the right we have that = 025,which corresponds to replacement-revenue projects.Although the structures of the optimal policy as repre-sented in both of these graphs have a similar form, wesee that the thresholds of that demarcate the regionsof optimality of codevelopment mechanisms are dif-ferent depending on whether is small or large. Asthe comparison between these two graphs suggests,we can see that when increases from 0 to 025,the range of parameters for which investment sharing

and innovation sharing are optimal decreases.Why are the codevelopment mechanisms proposed

in this paper (investment and innovation sharing)more valuable for new-revenue projects than forreplacement-revenue projects? One possible reason isthat the value of collaboration is greatest for new-revenue projects. When is small or negligible, theentire value that firms seek to maximize is a direct con-sequence of the development work that is undertaken.Under such circumstances, it becomes very importantfor the firms to choose mechanisms that induce higherinvestment levels. Because both investment and inno-vation sharing align incentives of firms and increase

their development investments, and, consequently, thelevel of innovation, these mechanisms also becomemost valuable when is low. Additionally, the reduc-tion in the fixed development costs that occurs underinnovation sharing allows firms to realize higher qual-ity with lower costs. In contrast, when is high,the value generated through the new innovation islower, and the ability to retain a larger share of the

Figure 4 Effect of Translational Uncertainty

0.55

0.50

0.80

0.90

0.95

1.05

1.15

1.00

1.10

Equilibriumprofits

Equilibriumprofits

0.4 0.6 0.8 1.0 1.2


0.4 0.6 0.8 1.0 1.2 1.4


Investment sharing

Innovation sharing

Revenue sharing




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

total revenue becomes the overarching objective. As aresult, revenue sharing that allows the innovating firmto retain a larger share of the total revenue to itselfbecomes the dominant mechanism.

We now focus on the effect of translational uncer-tainty on a firms preferred mode of codevelopment.

In particular, we examine the shift in a firms code-velopment strategy as it moves from a project withlow translational uncertainty to another project withhigh translational uncertainty. Note that the presenceof timing uncertainty is what makes this analysis dis-tinct from Proposition 6.

Observation 3. For projects facing both translationaland timing uncertainty, investment sharing competes withinnovation sharing for new-revenue products and withpure revenue sharing for replacement revenue products.

We find that innovation sharing is relatively lessattractive compared with investment sharing and rev-

enue sharing for replacement-revenue projects thathave both timing and translational uncertainty. Thiscan be seen by comparing the two graphs in Figure 4,which represent different levels of translational uncer-tainty. In the graph on the left, we have that v = 08(representing low translational uncertainty), and inthe graph on the right we have that v = 02 (rep-resenting high translational uncertainty). In contrast,when firms face both timing and translational uncer-tainty for new-revenue projects, we find that invest-ment sharing and innovation sharing are dominantforms of codevelopment. This can be seen by exam-ining Figure 2 for which = 0 and comparing it with

Figure 4.As before, we can compare the specific thresholds

on within which innovation sharing is optimal forthese different levels of v to illustrate how the code-velopment strategy should be adjusted to the levelof translational uncertainty in a replacement-revenueproject. This comparison shows that the range ofdevelopment capability ratio for which both revenue

notbepostedonany



@informs.org.


15/18


Figure 5 Effect of Outside Option on Codevelopment

0.75

0.65

0.55

0.70

0.60

0.80

0.70

0.65

0.55

0.45

0.60

0.50

Equilibr

iumprofits

Equilibrium

profits

0.40.2 0.6 0.8 1.0 1.2 1.4


0.40.2 0.6 0.8 1.0 1.2 1.4


Investment sharing

Innovation sharing

Revenue sharing




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 increasesas v increases, implying that when firms face bothtiming and translational uncertainty, revenue shar-ing and investment sharing are the dominant mecha-nisms. To understand this result, note that the effectsof timing uncertainty are most felt by innovation shar-ing, particularly when development capabilities offirms are different (high or low ). As a result, whenthe translational uncertainty associated with a projectis already very high, exacerbating its effect by takingupon timing uncertainty as well might not be opti-mal. It follows that both investment sharing and rev-enue sharing, which face less timing uncertainty andno integration costs, become dominant mechanismsfor larger ranges when translational uncertainty in

a replacement revenue project is higher.Until now we have assumed that the outsideoptions of both firms were identical.6 We now exam-ine how the structure and operationalization of co-development is affected when this assumption isrelaxed. These results are depicted in the two graphsin Figure 5in the graph on the left, the outside optionof PF is 0.05 and that of FF is 0, whereas in the graphon the right, we have that the outside option of PFis 0 and that of FF is 0.05. Comparing these graphswith Figure 2, where we assume that outside optionsof both firms are zero, helps us characterize the impactof these options on codevelopment choices.

Observation 4. When the value of the outside optionfor the partner firm increases, codevelopment using inno-vation sharing becomes more valuable. In contrast, whenthe value of focal firms outside option increases, revenuesharing becomes more attractive.

Whether FF prefers innovation sharing or purerevenue sharing depends on how its own value of

6 Recall that we normalized the outside option value of the firmsto 0.

the outside option compares with that of the rival.In particular, we find that when the outside optionof the partner is higher, innovation sharing becomesthe dominant mechanism. In contrast, when the focalfirms outside option is higher compared with its part-ners, pure revenue sharing is sufficient to coordi-nate the codevelopment process. To understand whya higher value of outside option makes innovationsharing preferable, recall that the focal firm tends toretain a larger fraction of the project value underpure revenue sharing, as a result of which it is morelikely for the partner firm to opt out under this agree-ment. Thus, revenue sharing is less likely to alignincentives and hence becomes less attractive. In con-trast, innovation sharing results in a more equitabledistribution of project value because the revenuesare shared (loosely) according to their developmentinvestments. This enables better incentive alignmentand hence makes it less likely for the partner firmto opt out. However, when FFs outside option ishigher, eliminating its incentives to opt out becomesmore important. Naturally, revenue sharing, whichallows the focal firm to retain a larger share of therevenues to itself, becomes the preferred mode ofcodevelopment.

Our numerical analysis also shows that the equilib-rium under investment sharing is no longer 1

2. Sim-

ilarly, the presence of outside options can also leadto a pure revenue sharing agreement where the equi-librium R is greater than

34

. Thus, the availabilityof outside options also provides a better bargainingposition for a firm with respect to its partner andallows it to capture a larger share of the realizedproject value. The implication of these observations isthat the bargaining power with which a firm entersthe negotiation process plays an important role inthe operational implementation of codevelopment. Itinfluences both how the revenues are split betweenfirms and determines which mechanism is the opti-mal mode of codevelopment.

notbepostedonany



@informs.org.


16/18


7. Managerial Implications andConclusions

In this paper, we conceptualize and model the rev-enue, cost, and effort sharing collaborative arrange-ments between two firms and characterize the optimaljoint-development approach for various technological

and market parameters. We find that the preferredcodevelopment approach should go beyond simplerevenue sharing, under which there exists an incen-tive for an innovating firm to underinvest in qualityimprovements. The cost and effort sharing mecha-nisms presented in this paper have the potential toaddress the inefficiencies associated with multipartydecision making. We found from our analysis thatinvestment and innovation sharing are more appro-priate for collaboration depending on a variety ofconditions. In particular, when firms have distinc-tive capabilities, innovation sharing agreements helpfirms exploit their specialized product development

capabilities and provide firms the appropriate incen-tives to ensure greater investments in technologyand product development. However, when develop-ment capability is concentrated in one of the firms,the additional integration costs of innovation sharingreduce its attractiveness. Because the quality distortioneffects of multiparty decision making are eliminatedthrough investment sharing under such conditions,investment sharing helps attain optimal investmentsin development and becomes the optimal mode ofcodevelopment.

The insights from the modeling and analysis ofthe codevelopment problem are distilled into a con-

ceptual framework in Figure 6. This framework pro-vides managerial insights by demarcating the regionsof appropriateness of the different codevelopmentapproaches. Specifically, the cost sharing approachesare more appropriate for new-revenue projects. Inno-vation sharing is more appropriate for projects withtranslational uncertainty, and investment sharing out-performs the other approaches for projects with

Figure 6 Optimal Codevelopment Approach for Different

Project Types

Innovationsharing

Pure revenue

sharing

Pure revenue/

investment sharing

Investment

sharing

Innovation/

investment

sharing

InvestmentsharingN

ew

revenue

Replacement

revenue

Only translationaluncertainty

Both translationaland timing

uncertainty

Predominantlytiming

uncertainty

Type of project uncertainty

Typeofprojectrevenue

predominant timing uncertainty. When the projectinvolves the launch of a product with incremental rev-enues, revenue sharing is sufficient to facilitate thecollaboration.

The above framework can be used to interpret, inpart, why some firms resort to sharing the develop-

ment work (innovation sharing) in multifirm develop-ment, whereas we see the funding of the developmentwork in other situations. Consider the computerindustry where most products are new to the mar-ket rather than new to the world and are introducedaround major industry events and conferences. In thiscase, timing is determined by outside events and theuncertainty is more of the translational kind (qualityof the product). Our framework would recommend aninnovation sharing approach for such cases. We dis-cussed earlier Dells efforts to complement its suppli-ers, who perform upstream component developmentactivities, leaving Dell with the system integration and

qualification activities. In fact, the supplier selectionprocesses at Dell ensure that vendor skills comple-ment its core capabilities in operational integration(Financial Times 2003, Hachman 2002). This puts it inthe top left of the framework in Figure 6.

In contrast, the pharmaceutical industry is char-acterized by long and highly uncertain lead timesfor drug development and, hence, correspondinglyhigh resource allocation costs, high levels of develop-ment uncertainty, and development capabilities thatare concentrated in small biotech firms, all of whichresult in high timing uncertainty. More importantly,strong regulatory influence (from the likes of the Food

and Drug Administration in the United States), whichmakes it necessary for products to possess a bareminimum quality, forces firms to continue workinguntil such progress is achieved, making their challengemanaging the development time rather than the finalproduct quality. We believe that these industry andtechnology characteristics have made investment shar-ing the popular form of contractual development inthe pharmaceutical industry (top right of the frame-work in Figure 6).

Our analysis, which provides concrete guidelines onhow the codevelopment approach needs to be tailoredto the nature of the project/product, still is quite styl-ized and represents merely a groundbreaking effort.Collaborative product development is a rich topic witha whole host of issues that need to be studied to deter-mine how relationships between firms must be struc-tured and nurtured. Complex interactions betweenfirms in a product development context seem to limitthe degree to which inefficiencies stemming from lackof coordination can be reduced compared to a supplychain/distribution context where primarily prices andquantities are being negotiated. Further analytical andempirical work would enhance our understanding of

notbepostedonany



@informs.org.


17/18


these complex yet important issues. In the first phaseof the work, we made a number of stylized assump-tions to obtain compact analytical expressions. In addi-tion to assuming specific forms for development costfunctions similar to the ones used in the literature,we also ignored agency costs associated with coop-

erative effort. A firms investments normally extendbeyond money into human resources, intellectual cap-ital, and a host of other factors that are not verifiableacross firms. Although, expressing the developmentinvestment in terms of quality improvement (qualitycould be quantified in terms of battery life, processorspeed, or memory capacity in most technology inten-sive industries) brings in a degree of verifiability tothe process, several other soft issues like trust makethe problem difficult to quantify. Although we recog-nize the existence and importance of these factors, webelieve that modeling these are beyond the scope ofthis research, and we leave those as an avenue for

future research.We also modeled the case when firms make deci-sions with complete information about each otherscosts and incentives. Although this would be truefor advanced development projects where collaborat-ing firms have to work very closely (particularly ifthey collaborated successfully in the past), it mightnot be appropriate for R&D initiatives of firms thatdo not have a history of collaboration with eachother. Additional work needs to be conducted tounderstand how asymmetric information about mar-ket demand will impact development decisions. Wealso ignored manufacturing and launch costs, which

could be important in some industries. Interactionsbetween these costs and product design choices couldhave a significant impact on joint-development deci-sions, and modeling these issues is an avenue forfuture research. More broadly, we also restricted ouranalysis to the case of a single technology and prod-uct development firm. In reality, multiple technol-ogy suppliers go to market through multiple productdevelopment firms, and understanding competitiveinteractions between multiple suppliers and productdevelopment firms would further enrich this discus-sion. A more detailed model of technology uncer-tainty and distributed product development would

expand the applicability of this research.In summary, conceptualization, model formulation,analysis, and discussion of multifirm product devel-opment has produced a number of interesting insights.First, it is important to consider technology and mar-ket characteristics in conjunction with economic andoperational issues when making product developmentdecisions. Second, investment and innovation sharing,proposed as codevelopment mechanisms on top ofrevenue sharing mechanisms, can help product devel-opment firms coordinate investments and achieve bet-ter quality products and higher profitability for firms.

Investment sharing is a better mechanism when firmcapabilities are dissimilar and projects face significanttiming uncertainty. However, when firms are simi-lar in terms of their capabilities and collaborate onnew-to-market product projects with significant trans-lational uncertainty, innovation sharing is better suited

to leverage the specialized skills of individual innovat-ing firms. We believe the modeling and the frameworkderived from the analytical results in this paper pro-vide a richer understanding of collaborative productdevelopment (beyond the current general discussionfound in the business press) and lay the ground-work for more advances on this increasingly impor-tant topic.

8. Electronic CompanionAn electronic companion to this paper is available aspart of the online version that can be found at http://mansci.journal.informs.org/.

AcknowledgmentsThe authors gratefully acknowledge the department edi-tor, associate editor, and three anonymous referees for theirhelpful suggestions and comments during the review pro-cess. The second author also acknowledges the financialsupport from the Sheryl and Harvey White endowment.

References

Abbott, L. 1953. Vertical equilibrium under pure quality competi-tion. Amer. Econom. Rev. 43(5) 826845.

Aghion, P., J. Tirole. 1994. The management of innovation. Quart. J.Econom. 109(4) 11851209.

Amaldoss, W., R. Meyer, J. S. Raju, A. Rapoport. 2000. Collaboratingto compete. Marketing Sci. 19(2) 105126.

Anderson, E. G., G. G. Parker. 2002. Project sourcing decisions inthe presence of learning and integration costs. Working paper,University of Texas, Austin.

Austin American Statesman. 2003. Alliances, consulting let IBM shareresearch costs. (April 21).

Cachon, G. P., M. A. Lariviere. 2005. Supply chain coordination withrevenue sharing contracts: Strengths and limitations. Manage-ment Sci. 51(1) 3044.

Cohen, C. M., S. Klepper. 1996. A reprise of size and R&D.Econom. J. 106(437) 925951.

Cohen, W. M., S. Klepper. 1992. The anatomy of industry R&Dintensity distributions. Amer. Econom. Rev. 82(4) 773799.

Cvsa, V., S. M. Gilbert. 2002. Strategic commitment versus post-ponement in a two-tier supply chain. Eur. J. Oper. Res. 141(3)526543.

Doz, Y., G. Hamel. 1998. Alliance Advantage: The Art of Creating ValueThrough Partnering. Harvard Business School Press, Boston.

Dutta, S., A. M. Weiss. 1997. The relationship between a firms levelof technological innovativeness and its pattern of partnershipagreements. Management Sci. 43(3) 343356.

Dyer, J. H. 2003. Collaborative Advantage: Winning Through ExtendedEnterprise Supplier Networks. Oxford University Press, Oxford,UK.

Erat, S., S. Kavadias. 2006. Introduction of new technologies to com-peting industrial customers. Management Sci. 52(11) 16751688.

Financial Times. 2003. Dell takes on printer market. (March 25).

Fine, C. H., D. E. Whitney. 1996. Is the make-buy decisionprocess a core competence? Working Paper 140-96, SloanSchool of Management, Massachusetts Institute of Technology,

notbepostedonany



@informs.org.


18/18


Cambridge, MA. http://ideas.repec.org/p/mit/sloanp/2605.html.

Gerwin, D., J. S. Ferris. 2004. Organizing new product developmentprojects in strategic alliances. Organ. Sci. 15(1) 2237.

Gilbert, S. M., V. Cvsa. 2003. Strategic commitment to price to stim-ulate downstream innovation

effort, revenue, and cost sharing mechanisms

Documents