|
Article Excerpt
Introduction
How do firms strategically manage their research and development (R & D) pipelines? Consider the pharmaceutical company Merck & Co., Inc. that was in an enviable position in the mid-1980s. Several blockbuster drugs had been recently commercialized and its R & D pipeline was chock-full of more potential blockbusters. With this success Roy Vagelos, Merck's CEO, shifted his firm's policy and admonished scientists to drop projects in the pipeline that did not have blockbuster potential (Nichols 1994, p. 107). Envy diminished by 1993 as Merck's pipeline grew thin: The pipeline did not promise to replace revenue losses from drugs coming off patent later in the decade. The thinness of the R & D pipeline became critical by 1998 when blockbusters began coming off patent (Seiden 1998). Internal development produced few blockbuster drugs, which led Merck to another shift in policy that increased its reliance on in-licensing, joint ventures, and collaborations for replenishing its pipeline. (1) With the pipeline remaining precariously thin (Barrett 2001), CEO Raymond Gilmartin again shifted policy by reorganizing the firm in 2000 to greatly increase its ability to purchase projects from the technology market (Pisano 2002). Merck's pipeline received unexpected additional pressure when in 2005 it withdrew the drug Vioxx from the market. In response, Merck announced a new policy that would put an even greater emphasis on bringing in compounds from the outside as well as accelerating the development of several internally discovered compounds (Bellucci 2005).
Merck's policy shifts for managing its R & D pipeline hints at the potential of a complex interaction between the state of its pipeline, the financial thresholds it used to evaluate project advancement and termination, the nature of R & D projects it undertook, and its choice to invest in an internal R & D versus acquiring projects originating outside the firm. This history of policy shifts suggests at a minimum that Merck's financial hurdles for advancing projects fluctuated over the past 20 years--shifting from high thresholds selecting only to develop potential blockbuster drugs, to lower thresholds that included Merck's willingness to commercialize its less valuable projects and to acquire other firms' discoveries. This history may also reflect changes in preference for projects--Merck may have shifted from high-risk and high-return internal projects to lower-risk and lower-return projects, which might further explain the company's difficulty in replacing revenue from products that came off patent.
Merck's experience is not unique suggesting that such interactions may be pervasive. (2) Indeed, any firm from any industry where the value of commercialized projects ultimately expires faces similar conundrums: What financial thresholds should a firm use for evaluating projects and should these thresholds vary over time? Should a firm buy or sell innovations in technology markets or rely on internal development and commercialization? To what extent should a firm's risk preferences for projects vary over time? More generally, how should a firm's policies for strategically managing R & D change in response to the state of its pipeline?
A variety of academic literature considers some of these questions but does so in piecemeal and incomplete ways. The literature on project finance, for instance, proposes that firms should advance any project whose expected value is greater than zero. This prescription does not seem to accord with actions taken by some firms like Merck that terminated apparently positive-expected-value projects, as well as changed thresholds over time, because it fails to account for the impact of a single project on the whole pipeline of projects. (3) Product-development literature in marketing and operations that discusses R & D pipelines recommends a stage-and-gate system for assessing which projects to advance or terminate. This literature argues that balancing the portfolio is an important strategic goal (Cooper et al. 2001); however, it provides no systematic analysis of how to balance (Cooper 1986) nor discusses the possibility that thresholds may change over time. This literature neither provides a comprehensive framework for thinking about how technology markets play a role in strategically managing the R & D pipeline. Streams of literature in economics, management, and organization theory discuss licensing, joint ventures, and partnerships; (4) but, these studies rarely come in contact with issues concerning the managing of the R & D pipeline nor do they inform the financial thresholds and risk preferences used to make pipeline management decisions.
In this paper, we explore the strategic management of R & D pipelines in the sense that policies and their associated pattern of resource-allocation decisions are critical to a firm's long-run performance (Barney 1997). Unlike the previous literature, we do not model product and technology market competition or cooperation, but we do provide an analytic assessment of how to dynamically manage the R & D pipeline. One of the key insights on which we build our theory is our observation that some firms invest in downstream cospecialized activities (Williamson 1985) such as manufacturing, distribution channels, and brands. We assume that firms would incur substantial adjustment costs if R & D efforts are unsuccessful causing the firm to run out of projects to commercialize that utilize these cospecialized assets. For instance, some pharmaceutical firms make cospecialized investments in "drug reps" (i.e., drug sales representatives) who receive firm-specific training and invest in idiosyncratic relationships with select physicians who are most likely to prescribe the pharmaceutical products the firm sells. Should a firm fail to commercialize new products, it would forgo returns from these cospecialized investments and personnel would leave for better opportunities or would be laid off to avoid incurring their carrying cost. Although these two types of costs differ, they both have the same economic impact in that the firm loses the opportunity of continued returns from its past investments; so, we refer to them collectively as adjustment costs. To include adjustment costs in a typical expected-value analysis for each project is complicated by the fact that these costs are contingent on the state of the pipeline. For example, an incumbent firm's decision to terminate one project raises the probability of incurring future adjustment costs; however, this probability hinges on the performance of other projects in the pipeline now and in the future. (5) We argue that these adjustment costs can have substantial implications for innovative behavior and firm performance; hence, they should be evaluated through investigating the dynamic interdependencies among project decisions.
Another key insight is that firms in some instances may be able to avoid incurring adjustment costs if projects for commercialization can be purchased in the technology market. The benefit of this option depends on the magnitude of the costs of transacting in this market. Transferring technology can face substantial transaction costs (Caves et al. 1983). For example, project knowledge can be tacit (Teece 1977, 1981), which can lead to costly transfers as well as adverse selection and moral hazard problems. If transaction costs are low, firms may be able to avoid adjustment costs by purchasing external R & D projects from the technology market. This option becomes increasingly less likely for higher transaction costs, which may impact the likelihood of incurring adjustment costs, thresholds, and risk preference.
Building on these insights and utilizing the stage-gate literature (for a summary see Calantone and De Benedetto 1990, Cohen et al. 1998), we introduce a dynamic-programming model that assumes the R & D pipeline is comprised of three stages (i.e., research, development, and commercialization) and two gates (i.e., advance to development and advance to commercialization). We assume the possibility of different pipeline states depending on the projects present in development and commercialization stages. We refer to different combinations of pipeline states that symbolize different types of incumbents and entrants, as different "types" of firms, and we explore how managers dynamically adjust their policies for selecting projects in these states. We parameterize both the possibility of adjustment costs, which are incurred if the firm has no project with which to replace an expired product, and transaction costs in the technology market, where firms can purchase projects to replenish their pipeline or sell projects to avoid investing in cospecialized assets. We then derive the corresponding optimal project-selection thresholds and the buy-or-sell price thresholds for each firm type. Our model complements recent research on R & D portfolio management. Cassiman and Ueda (2006), for instance, study why an established firm might choose not to commercialize an innovation whereas a start-up would chose to do so. Although complementary to our approach, the firms in their model face excess supply of R & D projects and their model assumes a frictionless technology market whereas our model studies a technology market with positive transaction costs and introduces the notion of adjustment costs to R & D pipeline management.
Our model leads us to three sets of results. First, our model's analytic results demonstrate that the thresholds for advancing (as well as for buying and selling) projects differ by the state of the firm's pipeline, the magnitude of transaction costs in the technology market, and the magnitude of adjustment costs. Incumbent firms--those firm types already possessing cospecialized assets--choose projects with positive expected value, whether developed internally or purchased in the market, so long as transaction costs for buying and selling in the technology market are negligible. This policy changes dramatically as transaction costs increase in the technology market. Incumbent firms may choose internal projects--even those with negative expected value--in order to avoid incurring either adjustment or transaction costs. We simulate these policies in the context of pharmaceuticals to estimate the relative size of effects.
Second, simulations based on our model show that preferences over project risk differ by the state of the firm's pipeline, the magnitude of transaction costs in the technology market, and the magnitude of adjustment costs. We show that, in general, firms are risk loving in their R & D activities because managers can limit loss by terminating unprofitable projects. But entrants--those firm types not possessing cospecialized assets--are comparatively more risk loving than incumbents implying differences in project risk and hence portfolio management. We also find that changes in adjustment costs and transaction costs can effect preferences for project risk; both incumbents and entrants are increasingly less risk loving the greater adjustment or transaction costs are. Interestingly, efforts to lower adjustment costs and to reduce transaction costs may be substitutes: Relatively low adjustment costs neutralize the effect of greater transaction costs (low adjustment costs reduce the needs to trade in the market) whereas relatively low transaction costs neutralize the impact of greater adjustment cost (the existence of a technology market helps firms to avoid adjustment costs).
Third, our model implies that entrants are more likely to sell their projects to incumbents as transaction costs decline. For sufficiently high adjustment costs or sufficiently low transaction costs, entrants specialize in R & D activities and never commercialize projects. Our model predicts that incumbents and entrants pursue R & D projects with more similar risk characteristics as either adjustment or transaction costs decline, but in the former case entrants are increasingly more likely to commercialize. These results yield interesting public-policy implications as well as generate incentives for managers to shape both transaction costs in the market and adjustment costs within their firms.
The paper proceeds by introducing our dynamic R & D management model and summarizing its main analytic results. The discussion section uses this model as well as several simulations to present our three sets of implications for the strategic management of R & D pipelines. We discuss the extent to which our model informs Merck's behavior and we identify the strengths and weaknesses of our model, which helps us to discuss its generality to industries beyond pharmaceuticals, and identify several additional research questions associated with R & D portfolio management that flow from our implications.
Model
We introduce a dynamic model in which the focal firm, in each period, initiates a project that goes through two stages (in two periods) before generating a possible new product. Stage I is the research stage in which a potential product is first...
|
