UNDERSTANDING EXPLORATION OF NEW CAPABILITIES USING REAL OPTIONS AND - - PDF document

SLIDE 1

UNDERSTANDING EXPLORATION OF NEW CAPABILITIES USING REAL OPTIONS AND DIVERSIFICATION LENSES*

Roberto S. Vassolo Krannert Graduate School of Management Purdue University West Lafayette, IN 47907-1310, USA Tel: (765) 746-5444 Fax: (765) 796-3483 Email: roberto_vassolo@mgmt.purdue.edu Jaideep Anand Corporate Strategy and International Business University of Michigan Business School Ann Arbor, MI 48109-1234, USA Tel: (734) 764-2310 Fax: (734) 764-2557 Email: jayanand@umich.edu Timothy B. Folta Krannert Graduate School of Management Purdue University West Lafayette, IN 47907-1310, USA Tel: (765) 494-9252 Fax: (765) 494-9658 Email: foltat@mgmt.purdue.edu

* We are grateful to Gautam Ahuja, Will Mitchell, Dan Schendel and Mark Shanley for guidance and helpful

suggestions.

SLIDE 2

2 ABSTRACT Real options formulations and diversification models represent the two dominant ways of thinking on how firms cope with technological and market uncertainties. Yet, they have developed independent of each other and the literature has missed interesting opportunities for cross-fertilization between them. In this paper we begin with the observation that firms often simultaneously invest in multiple options and develop a model of the interaction among such multiple options within a single firm. We integrate two conceptual lenses from the diversification literature into a real option model. First, we investigate the effects of correlations between the outcomes in different options. Second, we analyze the effects of investments that are fungible across project options. We show that under different conditions multiple options can be sub-additive (due to redundancies in outcomes) or super-additive (due to fungible inputs). We test the implications of our model with data from the biotech industry and find supporting

• evidence. Our model and results have some interesting implications for the real options lens in

general, and for investments in multiple technological and market-entry projects in particular.

SLIDE 3

3 A growing body of research has developed to enhance our understanding of how firms make investments for accessing new capabilities in high velocity environments. Recently, Kogut (1991), Bowman and Hurry (1992), Kim and Kogut (1996), McGrath (1997), and Folta (1998), among others, suggest the use of real option theory to explain how strategic flexibility influences firms’ pursuit of new capabilities. Strategic flexibility can be characterized as “real options,” that can be initiated through minimal or partial commitments (Sanchez, 1993), while strategic commitment involves full investment in a course of action. The standard result from this view is that uncertainty raises the value of holding the option. The real option literature to date has tended to focus on individual options (i.e., one

• ption at a time). However, strategic flexibility in most firms typically takes the form of a

collection of real options. We extend previous work in this area in two important ways. First, we consider that a firm’s investment opportunities should not be evaluated in isolation. When firms have multiple real options that interact with one another, their individual values may be non-additive. The implication is that the timing or likelihood of exercise of a single option may be influenced by the presence of correlated options in the firm’s option portfolio. The nature of such interactions and the conditions under which they may be small or large, as well as negative

• r positive, may not be trivial. Second, we use advances in the resourced-based view of the firm

to isolate the conditions where interactions are non-trivial. As such, we provide an important step in marrying diversification literature with real option theory. This also provides an important departure from prior work in finance that does not consider firms are asymmetrically positioned to initiate and exercise their real options. The remainder of the paper is organized as follows. The next section presents the model and derives propositions relating to the factors that determine the value of firm’s options in the presence of multiple overlapping options. Section three applies the model to the specific context, where firms’ collection of equity-based alliances is characterized as a portfolio of real options on new technological opportunities. In this sense, we follow prior work that has characterized equity-based strategic alliances as real options (Folta, 1998; Folta and Miller, 2002), but extends prior work by focusing on how interactions within the portfolio influence option exercise. Section four introduces a sample and measures to test our hypotheses. Finally, results are discussed and future directions for research are offered. MODEL AND PROPOSITIONS Option theory is useful for valuing the flexibility inherent in managers’ investment

• decisions. Compared to traditional valuation methods, such as net present value (NPV), it more

accurately accounts for the value of flexibility when investment decisions involve some irreversibility and the outcome of an investment is uncertain. Trigeorgis (1987) suggested that the value of a firm’s investment is a function of it traditional NPV and its option value. With few exceptions, a characteristic of previous studies in the real option literature is a focus on isolated options that are independent of other options held in a firm’s portfolio. Hereafter, the focus of this study is on the implications of this assumption using the example of two investments, α and β, by a single firm. The assumption of independence suggests that the value of α and β, are: ,

β β β α α α

OV NPV V OV NPV V + = + = [1] where NPV refers to the passive NPV and OV to the (flexibility) option value. Under the independence assumption, the value of both investments can simply be added together:

SLIDE 4

4

β β α α β α

OV NPV OV NPV V V + + + = =

+

. [2] However, under certain strategic conditions (e.g., R&D activities), it is common to

• bserve that firms make multiple investments seeking the same output (Madhok, 1997). In other

words, they duplicate their commitment to increase the odds of achieving a first-mover

• advantage. Also, many of the assets required for carrying out these investments are intangible or

tacit and, therefore, fungible. In other words, they are public goods within the firm (Penrose, 1959). The consideration of these investments as real options introduces particular complexities that affect the additivity nature of their values and, therefore, violates the independence

• assumption. Hereafter, these violations are mentioned as the portfolio effect (PE), and the

equation of overall valuation becomes

αβ β β α α β α

PE OV NPV OV NPV V V + + + + = =

+

. [3] The feature that the output of several investments is correlated and competitive generates sub-additivity in their total value. This is because the “duplication” of the investment effort, since succeeding in one investment significantly erodes the strategic value of the remaining ones. The presence of correlation, therefore, diminishes the total value of the portfolio of strategic commitments, that is ) ( < =

αβ αβ

ρ f PE . [4] In this type of model (competitive options) it is not possible to have a positive effect of uncertainty correlation. Stulz (1982) and Johnson (1987) were the first to examine the valuation implications of correlated options held simultaneously by an investor. In their view, the option holder has the ability to switch to the option with the highest value among the correlated options. Since switching to the highest valued option erodes the value of the remaining options, there is a decreasing marginal return to holding correlated options. Firms rich with intangible or tacit resources have few capacity constraints on these resources and can apply them readily across the organization. More relevant to the research question, however, is that these resources often create greater fungibility – meaning they can be used more readily as public goods within the firm and its investments. Fungibility represents a firm-level capability that enables a firm to benefit from economies of scope, reducing the cost of each investment. If the firm can leverage its assets into the current strategic investments, the total value of the portfolio will present super-additivity. The correlation here is not between the investments but between the firm (F) and its investments. This relationship can be expressed as ) , ( > =

β α αβ

ρ ρ

F F

f PE . [5] The presence of these two types of phenomena introduces an intriguing trade-off in the valuation of the strategic investments of the type described above. The relative size of the positive and negative valuation effects will determine whether the portfolio effect is sub-additive

• r super-additive. Figure 1 assesses the expected relationship under different situations.

*** Insert Figure 1 about here *** To summarize, the propositions are two fold: Proposition 1: When a firm invests in multiple projects, correlations among the outcomes of the projects lead to a sub-additive value of the combined options. Proposition 2: When a firm invests in multiple projects, fungibility of shared resources with the projects leads to a super-additive value of the combined options. In summary, uncertainty per-se is not enough to determine the potential value created in the portfolio of options; it is also necessary to include the level of correlation among options and the degree of fungibility between the option and the focal firm. A failure to consider either

SLIDE 5

5 portfolio effect will lead to a problem of misspecification in the analysis of the determinants of portfolio value, and hence portfolio size. TECHNOLOGY ALLIANCES AS REAL OPTIONS Strategic alliances are viewed as important modes by which exploration takes place. For example, Khanna et al. (1998) focus on how firms develop their technological expertise through strategic alliances. Mitchell and Singh (1992) stress that through strategic alliances in R&D, firms spread risk, increase market power, share resources, and gain organizational learning. Alliances also allow organizations to obtain the desired benefits without the added costs of governance (Williamson, 1985). A separate, but complementary stream has characterized strategic alliances as real

• ptions. The seminal paper in this area was written by Kogut (1991). He suggested that joint

ventures provide growth options for future expansion, while retaining the option to defer complete commitment. The joint venture enable the firms to learn about growth opportunities through close interaction with their partner, and thereby secure upside gains. His results suggest that partners will be more likely to exercise the option to buy out their partner when the target industry experience higher sales than forecasted. This result is consistent with expectations, and led to additional studies testing whether real options theory offers a useful perspective for understanding strategic alliance behavior. For example, like Kogut (1991), Folta and Miller (2002) examine the exercise decision associated with equity-based alliances. Their sample differs in that they exclude joint ventures, and firms could have multiple simultaneous alliances. They demonstrate that even controlling for the presence of other alliances, the key predictions from option theory hold. In the context of the biotechnology industry, Folta (1998) explored whether real options theory could inform the decision to prefer equity-based alliances (both joint ventures and minority direct investments) to outright acquisition biotechnology firms. His work attends to Williamson’s (1988) concern that integration of R&D activities may not be the optimal prescription in the face of uncertainty. Results indicate that firms have a greater propensity to initiate equity alliances in higher uncertainty, and the influence of uncertainty increases when investing in domains of higher asset specificity. Folta’s empirical focus on the propensity of established firm’s to initiate alliances with smaller biotechnology firms is consistent with Chi’s (2000) theoretical conclusion that significant resource asymmetries must exist between partners for the real option to have strategic value. Chang (1995) also considered that joint ventures alliances provide a platform into international domains, and platforms have been identified as a type of growth option that offers expansion opportunities because of path dependency (Kim and Kogut, 1996; Kogut and Kulatilaka, 1995). Other studies have found evidence that non-equity investments can be characterized as growth options (Folta and Leiblein, 1994; Mang, 1998). In summary, the aforementioned work has bridged theory to improve our understanding

• f why alliances may have real option characteristics, and why such a perspective illuminates the

context of strategic alliances. It has also provided evidence consistent with expectations from real options theory. Despite these contributions, there remains considerable opportunity for

• study. The main limitation of the literature with respect to this study is that all previous studies

focus on a single real option in isolation. One exception is Folta and Miller (2002), however they only control for the presence of other alliances. They do not consider the theoretical and empirical implications when firms have multiple options that may be dependent upon one

• another. Luehrman (1998b) explicitly considers strategy as a portfolio of real options, but does

not consider the effect of interactions on the value of each strategic alliance. Since interactions are a very common and important phenomenon in R&D alliances (Madhok, 1997), the above real

SLIDE 6

6

• ption studies are not adequate for understanding the configuration of a portfolio of strategic

alliances. In the presence of interactions, the valuation of a portfolio of simultaneous growth

• ptions is not straightforward. McGrath (1997) suggests that the cross-effect of uncertainty of
• ne strategic alliance on the boundary conditions of other strategic alliances be included.1 Her

work is a substantial contribution to the understanding of how firms configure their portfolio of exploratory strategic alliances. McGrath also emphasizes the need for further studies. Her comments are particularly relevant given Chi’s (2000) assertion about the difficulty of gaining an accurate understanding of a particular strategic situation without explicitly modeling the option

• structure. Even though in this study we do not mathematically model the option, we take care of

framing it in detail. The literature on diversification is also relevant for this study because it deals with firms’

• portfolios. The degree of relatedness (interaction from a portfolio perspective) has been

identified as a source of competitive advantage (Penrose, 1959). Therefore, relatedness between a firm and its alliances can impart substantial advantages. In the real options framework, such relatedness and resource sharing effectively lowers the entry fee (costs) of these options. To the extent firm resources are redeployable and fungible, they can exhibit public good properties at least within the firm. Firms can be effective in transferring intangible and tacit resources among different projects (Kogut and Zander, 1992). Hypotheses The standard result in option pricing theory is that the value of an option increases with an increase in the volatility of the underlying asset. For example, Folta and Miller (2002) demonstrate that in the context of strategic alliances firms are less likely to exercise their options in the presence of high sub-field value. However, this hypothesis does not take into account the interaction among the options with the portfolio of a firm. We begin with a base case of real options without considering the effect of interactions among them. The firm faces the choice of whether to explore in uncertain environments by fully committing resources through in-house development or acquisitions, or by partially committing resources by adopting a hybrid form of organization like a strategic alliance or equity agreement. Since real options emerge in response to technological or market uncertainty, there should be a positive relationship between uncertainty and the use of options. If the value of an asset is held constant, greater variance (i.e., uncertainty) of potential outcome implies higher option value (McGrath, 1999). More uncertainty implies higher expected upside gains while downside loses are not affected. Even more, in highly unpredictable situations, the best way to respond effectively to future challenges is to deploy not one but patterns of options (McGrath and MacMillan, 2000). In particular, equity agreements are suitable for environments characterized by rapid innovation and geographical dispersion in the sources of know-how (Teece, 1992). It makes more sense to split the investment into several small ventures than to making a single big bet on a single opportunity. Therefore, in terms of governance choice, the real options tradition assumes a negative relationship between uncertainty and exercise of the ongoing options, i.e., acquisition of alliances (Folta, 1998; Kogut, 1991). Higher technological uncertainty will cause firms to keep their agreements instead of terminating them since the termination of an option reduces the set of possible avenues for the firm. Under conditions of uncertainty, firms would tend to prefer to enhance rather than reduce this opportunity set. The implicit assumption of

1 Trigeorgis (1993) also develops a formal model for interactions but he considers compound and not simultaneous

effects

SLIDE 7

7 previous studies is that what holds for a single option should hold for the portfolio, independently of the degree of relatedness. Therefore: H1a: The higher the technological uncertainty, the lower the rate at which the firm divests alliances. H1b: The higher the technological uncertainty, the lower the rate at which the firm acquires alliances. The above hypothesis does not take into account the interactions among the options within the portfolio of a firm. As noted above, potential miss-specification can result from not considering such potential interrelations. We now turn to implications of Proposition 1 for such a

• portfolio. This proposition deals with the presence of interactions among the strategic alliances

that seek a first-mover advantage, and argues that correlations among outcomes of options reduce their combined value. In particular, a new option that seeks a first mover competitive advantage and is related to the other ongoing options of the firm, adds value to the portfolio in a decreasing fashion (sub-additivity property). Correlation between an ongoing options portfolio diminishes the expected gain of each of the options due to the option to switch. If a firm carries such related options in its portfolio, it will divest the least attractive of such related alliances, retaining the best potential alliances as possible candidates for future acquisition. Consequently, we will observe a positive relationship between uncertainty correlation and divestitures. Therefore, H2: The higher the technological overlap between an alliance and the portfolio, the greater the expected rate at which the firm divests the alliance. Proposition 2 assesses the effect of fungibility of input resources on the portfolio strategic

• alliances. A resource is fungible or redeployable if it can be used by the target, either by a

physical transfer of the resource or by a resource sharing without physical movement (Capron et al., 1998). Since the value of the alliance does not depend exclusively on uncertainty but also on the underlying asset value, firms can possess strategic alliances with high levels of fungibility between the firm and the alliance target. When the investment is partially redeployable or fungible among alliance agreements, the cost of a related option diminishes, making them more valuable (Folta et al., 2001). In this line, proposition 2 postulates a positive relationship between fungibility and the value of the portfolio (super-additivity property). Greater levels of fungibility increase the probability of acquisition. Upon acquisition, such related alliances, requiring common resource inputs can be consolidated with the existing capabilities base of the firm leading to economic gain. Greater fungibility reduces the costs of acquiring these options. Therefore, H3: The higher the fungibility of technological resources between the firm and an alliance, the greater the expected rate at which the firm acquires the alliance. Hypothesis 1 summarizes the link between uncertainty and real options based on previous research and extend previous arguments to a portfolio of alliances. Hypothesis 2 is based on proposition 1 regarding correlated outcomes, and hypothesis 3 is based on proposition 2 regarding fungible resources. Note that each of the hypothesis is based on a different dependent

• variable. Hypothesis 1 predicts termination including both divestitures and acquisitions,

hypothesis 2 only predicts divestitures, while hypothesis 3 only predicts acquisitions. We now turn to a description of the empirical context in which we test these hypotheses.

SLIDE 8

8 EMPIRICAL ANALYSIS The Biotech Industry Context Like previous studies that characterized equity alliances as real options (Folta, 1998; Folta and Miller, 2002), the context of our study is the biotechnology industry. The emergence

• f the modern biotechnology represents a technological discontinuity that has heavily challenged

the pharmaceutical industry configuration. The biotechnology revolution refers to a technique that comes from a scientific advance –the advent of molecular genetics and recombinant DNA. By 1960, messenger RNA had been discovered. This and other discoveries allowed, between 1972 and 1973, for the first experiments involving the introduction of foreign DNA into a bacteria, cutting and splicing gene fragmentation. This class of experiments was named “recombinant DNA”, expressing the fact that genetic information (DNA) is recombined in the test tube (in vitro). By significantly widening the opportunity for carrying out applied research, this technique has allowed for important commercial applications in different fields (e.g., medicine). The dynamic relationship between scientific discoveries and industrial evolution occurred in different stages. At the beginning, the new industry benefited significantly from strong federal support and university participation in both technology development and technology transfer. Technological transfer enabled different industries to interact with the source of knowledge

• generation. As a consequence, the initial links critically served to fill the gap between the

science and the market. Small startup companies did the initial stages of applied research and commercial developments. Between 1973 and 1987, 493 new biotechnology firms were created (Krimsky, 1991). The new companies developed a unique knowledge endowment that challenged pharmaceutical incumbents (Teitelman, 1989). As a strategic response to the lack of expertise in the new technology, large pharmaceutical corporations developed alliances with one

• r more biotech companies. The larger companies exchanged financial support and established
• rganizational capabilities in clinical research, regulatory affairs, manufacturing, and marketing

for the biotech startups’ expertise and patents (Galambos and Sturchio, 1998). Pharmaceutical companies approached the new technology through two main pathways. Some companies started by developing highly specific expertise and then attempted to generalize it across a range of different therapeutic categories. Another group skipped this stage and attempted to acquire and build upon general capabilities very early in the process of establishing licensing, research, and equity relationships with biotech labs (Galambos and Sturchio, 1998). However, regardless of which path pharmaceutical companies chose, the phenomenon of strategic alliances, buyouts, and acquisitions determined the evolution of the industry until today. Vertical relationships with biotechnological companies have been, for the pharmaceuticals, a twofold strategic action. First, they have wanted to block competitors in case certain biotech labs discover a valuable drug (Teitelman, 1989). In addition, they have used the strategic alliances to substitute for developing internal expertise judged of marginal value (Zucker and Daily, 1997). Stated differently, pharmaceutical companies have mainly explored through their strategic alliances. As a salient feature, these alliances have often been held with labs doing competitive research. At the end of the 1990s, the shape of the pharmaceutical-biotechnological industry was slightly different than that of the seventies. Pharmaceutical companies had established significant

SLIDE 9

9 capabilities in the new field. However, even the largest firms were challenged to fund both basic and developmental research across the wide range of opportunities, and these problems, together with the necessity of achieving economies of scale in manufacturing and distribution, reinforced experiments with strategic alliances (Galambos and Sturchio, 1998).2 In summary, the attraction of this industry lies in the high levels of strategic alliance formation and transformation taking place, with pharmaceutical firms holding simultaneous alliances in similar areas of research. At the same time, most of these alliances are easy to identify as relating to exploration activities. Further, the strategic alliances appear to be largely driven by the incentive to gain a first mover advantage in the form of being the first to patent a new drug. As such, the pharmaceutical-biotechnological industry offers a logical choice as a setting to test the relationship between real options thinking and diversification in exploration contexts. Data The starting point for studying how pharmaceutical alliance decisions are influenced by the presence of correlations in its portfolio of strategic alliances, is to generate a sample of pharmaceutical firms to study. We use BioScan and the North Carolina Biotechnology Industry databases to establish a sample of equity alliances involving pharmaceutical firms with biotechnology firms. Rather than use the whole set of pharmaceutical firms with equity alliances in the databases, we selected the thirty firms with the largest number of equity alliances between the period 1989-1999. By focusing on firms with the largest portfolios we are able to aptly study

• ur phenomenon of interest. These thirty firms come from several different countries, including

USA, England, France, Germany, and Switzerland. During our sample period, conglomeration in the industry reduced our set of firms to seventeen. We account for mergers and acquisitions by modeling the firms separately until the merger or acquisition took place. Our sample includes 363 equity agreements initiated between pharmaceutical firms and biotechnology partners between 1989-1999. We tracked the equity agreements to understand whether they were terminated or maintained, in the following way. If the October 1999 issue of Bioscan listed the equity partnership as ongoing, the alliance was coded as right censored. Otherwise, a systematic search was undertake to understand the nature of the termination using sources in addition to those listed above, including Ernst & Young Biotechnology Industry Reports, Predicast F&S Index of Corporate Change, Lexis/Nexus, Dow Jones News Service, and SEC Schedule 13D

• filings. This effort enabled us to verify that there were 76 terminations: 14 instances where the

pharmaceutical firm bought out the biotechnology partner, and 62 divestitures. 183 different biotechnology firms were involved in 363 equity alliances. Dependent Variables The hypotheses are concerned about determinants of option exercise. Option exercise is defined as the timing of the exercise event – either acquisition or divestiture. The dependent variable examines the exercise decision (termination) surrounding existing equity alliances. As noted earlier, of the 363 equity alliances in our sample, we identified those as being terminated through acquisition or divestiture. Acquisitions and divestitures were coded “1” if the focal firm acquires or divests the target, and “0” otherwise. Independent Variables

2 The following data help to assess the magnitude of vertical relationships. In 1995, pharmaceutical companies

spent about U$S3.5 billion to acquire biotech firms, approximately U$S1.6 billion for R&D and licensing agreements with biotech firms, and from U$S1.2 to U$S7.5 billion on in-house biotechnological R&D (Davidson, 1996).

SLIDE 10

10 Technological Distance. This measure indicates the technological overlapping, which is the ratio

• f the common technological domain overlapping of two companies divided by the length of the

total technological domain of each of the companies. This study uses BioScan’s technological classification. Examples of technological domains that appear in BioScan are Aids Therapeutics, Bone Therapeutics, and DNA Probes. Several antecedents using similar technological measures exist (e.g., Folta and Miller, 2001). Most of these studies use a few broad categories of technologies. For example, BioScan classifies firms’ research in more than 150 technological domains within six major groups: Agriculture, Biotechnology, Food Industry, Human Diagnostics, Industrial Biotechnology, and

• Veterinary. Previous studies stay at the major group level, while our study specifies each of the

sub-technologies. The variable of technological distance is similar to the one used by Stuart and Podolny (1996). However, while Stuart and Podolny (1996) generate measures of similarity based on patent citations, the current study compares technological domains (i.e., BioScan’s specific subjects). Patent citation has been widely used in the field for analyzing technological

• verlapping (e.g., Mowery et al., 1998, Stuart and Podolny, 1996). Patent citations use

information on research that was conducted many years before the company decides to start a new external exploration activity. The distance between a technological discovery and a patent in this field is, at least, ten years. In this sense, it does not really capture future technological

• expertise. Strictly speaking, patents are only the result of past capabilities. Differently, the

technological domains show what sort of research the firm is carrying out at any given time. The sort of research that firms are carrying out is forward looking, since it identifies the potential products that can arise from this activity. Given that the value of a company is determined by its expected cash flows, this measure is more accurate in determining companies’ future value than patents. A simple example will clarify how this measure works. Firm A is researching one technological domain, T1, and firm B is researching two technological domains, T1 and T2. Figure 2 shows the pattern of technological overlapping and the matrix that results from this

• pattern. Notice that the technological overlapping between two companies is not necessarily
• symmetric. In this example, the overlapping of A with B is 100% while the overlapping of B

with A is 50%. In other words, all the technologies of A overlap with B’s technologies while

• nly 50% of B’s technologies overlap with A’s technologies. This is an important distinction for

understanding how this measure works. Additionally, the only interesting results are the non- diagonal values. Therefore, the values of the diagonal of the matrix are set equal to zero. The Appendix sketches the mathematical derivation of the measure. For more details, refer to Stuart and Podolny (1996). Insert Figure 1 about here For testing the hypotheses, we use two different dyads of technological distance: (1) the technological distance between the incumbent and the target, (2) the technological distance between the incumbent’s portfolio and the target. The value for each of these variables is

• btained by applying the methodology described above. The former measure is used to capture

fungibility between firm and its alliances, while the latter is used to measure the correlation between outcomes of a new alliance and the existing portfolio of alliances. Technological Value. Since almost 50% of the biotechnology companies are not public, the variable that measures technological value is industry returns, which is based on the

SLIDE 11

11 biotechnology industry index adjusted by the risk-free interest rate (interest).3 This index includes 10 public biotechnology companies that existed during all of the period, and it tries to keep a balance in the number of companies within each particular major subject grouping. The initial sample of the companies was taken from Folta’s (1998) index, complemented by other companies in order to ensure that all the companies existed during the period of analysis. Unlike Folta (1998), the index is capitally weighted instead of equally weighted. This change seems to better reflect market perceptions, since Standard and Poor’s uses the cap-weighted methodology. The index was also adjusted each time any of the companies launched new stocks to the market. This adjustment is necessary to avoid spurious upside gains (Standard and Poor’s manual). Technological Uncertainty. Industry volatility approximates technological uncertainty. It is the monthly standard deviation of the returns of the index. Technological Correlation. At the center of this study is the claim that traditional approaches to strategic alliances using real option tools fail to consider interrelations between alliances. This interrelation fell under the construct ‘target value correlation.’ The methodology for measuring this construct comes from applying the equations of technological distance to the lab and the remaining labs of the pharmaceutical firm. In particular, it measures the technological distance with the closest lab. Therefore, the measure will indicate the minimal technological distance between the target and the remaining targets of the focal firm. The variable was also calculated for the year prior to the transaction.

• Fungibility. Firms develop expertise in certain R&D fields together with improving their

absorptive capacity regarding these fields of expertise (Cohen and Levinthal, 1990). This is particularly important for firm alliance-related learning (Sarkar et al., 2001). Therefore, it is reasonable to expect that a pharmaceutical firm can better evaluate those biotechnology labs that carry out research in the pharmaceutical firm’s areas of expertise. In addition, the incumbent will be in a better position for achieving economies of scope (Rothaermel, 2001). Therefore, the technological overlapping between the incumbent and the target indicates higher possibilities for the incumbent to redeploy assets to the target, diminishing its cost. This study proposes the technological distance between the lab and the pharmaceutical firm for measuring the degree of resource fungibility from the incumbent to the target. The closer the technological endowment of the focal firm to the technological endowment of the target, the better is the relative position of this focal firm regarding its competitors to transfer resources to the target and, therefore, diminish its cost. The variable was calculated for the year prior to the transaction applying the equations of technological distance to the dyad pharmaceutical-lab. Control Variables Control Variables for the Partnership. Three indicator variables approximate the degree of appropriability of the partnership: license, option, and foreign transaction. The presence of a license or an option clause guarantees to the pharmaceutical firm the appropriability over the

• discovery. The purpose of these clauses is to provide an ex-post deterrent to opportunism (Deeds

and Hill, 1998). In other ways, the presence of a complete contract controls for behavioral uncertainty (Williamson, 1991). License is equal to one when the equity agreement includes a clause that gives to the pharmaceutical firm the right to commercially exploit a discovery. Option is coded one when the equity agreement either explicitly or implicitly indicates an option

• clause. It is explicit when the contract gives to the pharmaceutical firm the right but not the

3 A better alternative would be to use the stock market of each of the biotech companies. Unfortunately, this

information was not available for all the cases.

SLIDE 12

12

• bligation to exploit a discovery (explicit call option). It is implicit when different payments are

tied to the accomplishment of certain milestones by the biotechnology firm (implicit call option). Even though the biotechnology lab may fail in achieving the milestone, the presence of this clause signals both potential option gains and the right to exercise a compound option. This information was gathered directly from the press releases, when available. Otherwise, it was taken from BioScan and ReCap Foreign transaction is coded “1” if the pharmaceutical firm and the target belong to different countries and “0” otherwise. Firms entering into alliances with companies that belong to other countries should take into consideration institutional differences and institutional governance factors (Bishop, 1994), among others. Depending on the country, monitoring devices change. This difference makes it more difficult to control investments and, therefore, increases the risk of opportunistic behavior. In this sense, foreign transaction captures the presence of behavioral uncertainty that cannot be incorporated within a contract. Control Variables for the Incumbent. This study uses a measure of organization size for control for the financial resource position: total pharmaceutical annual sales, in US$ millions. Organizational size is a previously used predictor of alliance formation (Burgers et al., 1993; Gulati, 1995a), since larger firms may be more able to establish an equity agreement. The values for these variables were taken from Compustat, Lexis-Nexis, Global Access, and the annual reports of the firms. For most of the cases, the information was available under US standards, ensuring the compatibility of the measure across countries. Different sources were used for early

• bservations. In particular, substantial work was needed to get early observations for non-US
• companies. In some cases, English versions of the Annual Reports were not available, so that it

was necessary to consult the originals in French and German. We control for organizational innovation using the number of technological domains in which the incumbent is investigating, since broader research scope seems to indicate a higher commitment to innovation. Control Variable for the Target. Even though our measures of the industry index and returns may accurately measure the value of the field, they do not assess the value of a particular target

• firm. This may introduces a bias in governance decisions, since incumbents are more willing to

buyout targets with higher value. In order to cope with these potential biases, this study proposes a count variable that measures the total number of a target’s technological domains (Rothaermel, 2001). It is a proxy of the knowledge endowment of the lab and, therefore, of its value. It has been argued that firms active in more technological domains have larger growth options associated with them (Folta and Miller, 2001). Descriptive Statistics Table 1 contains descriptive statistics. It can be seen from this table that the technological distance between the pharmaceutical incumbent and the biotechnological target is much bigger than that between the biotechnological target and the remaining targets of the portfolio. In other words, pharmaceutical firms explore in distant technological domains, but they do that within a fairly tight portfolio. Table 2 shows the Pearson correlation coefficients. Note the low correlation between the two measures of technological distance (0.02). This information provides strong evidence for concluding that the two measures represent two different constructs (i.e., correlation and fungibility). Insert Tables 1-2 about here

SLIDE 13

13 Econometric Model This study uses a hazard rate model. The most important advantage of hazard rate models is that it allows incorporating in the sample right-censored variables. Right censoring

• ccurs when some observations have not experience a termination (buyout or divestiture) at the

end of the period. For the sample, every equity agreement held at the end of the period (December 1999) is right censored. Hazard rate models incorporate this phenomenon as part of the governance decisions to be explained rather than throw the data away. In addition, survival techniques apply a definition of the dependent variable as a duration or waiting time prior to an event that tends to be intuitively appealing in studies that explore why firms differ in the timing

• f their actions (governance choices here). The main disadvantage of survival analysis lies in its

lack of tools for controlling biases due to left-censoring (Yamaguchi, 1991). Our analysis models the dependent variable as the hazard rate of terminating an equity

• partnership. We observed two different types of termination events, divestiture and acquisition,

and modeled their hazard rates separately. The first set of models defined the dependent variable as the hazard rate of divesting an equity alliance. The second set of models defined the dependent variable as the hazard rate of acquiring a majority stake of the biotechnology partner. In both sets of models, the rate was specified as a Gompertz function of the independent variables, Xt, and a vector of parameters capturing the effects of the variables on the rate of subsequent equity investment. The Gompertz distribution was selected among several possible parameterizations based on the Akaike Information Criterion. The use of maximum likelihood estimation requires the assumption that events are uncorrelated across observations. Given that each pharmaceutical firm has multiple investments, this assumption is highly questionable in our data. To adjust for this problem, the two types of regressions were done using the ‘robust’ and the ‘cluster’ options in STATA. The ‘robust’

• ption specifies that the Huber/White/sandwich estimator of variance is to be used in place of the

traditional calculation. The ‘robust’ option, when combined with the ‘cluster’ option, allows for the presence of observations that are not independent within cluster (i.e., same pharmaceutical firm). However, observations should be independent across clusters (i.e., between pharmaceutical firms), what seems to be a reasonable assumption for the sample. RESULTS The results from the hazard rate models are exhibited in tables 3 and 4. Table 3 presents results pertaining to factors that influence the rate of divestiture. Model 1 is the base model. Model 2 incorporates a proxy for industry uncertainty. A likelihood Ratio test comparing model 2 with model 1 indicates that the addition of this variable provides significant explanatory power (p < 0.001). Examination of the individual coefficient for uncertainty suggests that there is a negative effect (p < 0.10) on the rate of partnership divestiture. This result is consistent with hypothesis 1a, and consistent with models that emphasize the effect of uncertainty on a single transaction in isolation. To test our conjecture about the importance of portfolio effects, model 3 introduces a measure of technological overlap between the alliance and the portfolio. When compared to model 1 using a likelihood ratio test, model 3 provides a significant (p < 0.10) improvement in fit. The individual coefficient is positive (p < 0.05), as expected, and suggests that the larger the technological distance between an alliance partner and the rest of the pharmaceutical firm’s portfolio of alliances, the lower the rate of divestiture. This result is consistent with hypothesis 2. Model 4 adds the variable relating to the technological distance between the pharmaceutical firm and the alliance partner. This variable did not significantly

SLIDE 14

14 contribute to model fit, but no relation was hypothesized. Finally, we can see from model 5 that these relationships hold up in a full model. Table 4 considers an identical set of models as table 3, while examining the rate of partnership acquisition. The small number of buyout events makes the identification of determinants problematic. Industry Volatility and Min Tech. Distance Lab-Portfolio are added in models 2 and 3, respectively, but their effects are not significant. Hypothesis 1b stated an expectation that industry volatility would lower the rate of partner buyout. Model 4 tests our key hypothesis, arguing that pharmaceutical firms with more fungible resources sharable with the partner are more likely to acquire the partner. A Likelihood Ratio tests indicates that model 4 provides a significant (p < 0.05) improvement over model 1. The individual coefficient is positive (p < 0. 05) as expected. Once again, these results hold up on the full model presented in model 5. All the control variables with a significant coefficient show expected relationships. CONCLUSIONS AND DISCUSSION Our main contribution is to better explain and recommend the optimal composition of a portfolio of options for a firm. By highlighting the possibility of sub-additivity and super- additivity between strategic options, this study seeks to provide valuable insights for both scholars and practitioners. From the academic perspective, this study will provide a powerful framework to determine the optimality of the portfolio of R&D strategic alliances or other

• ptions. From practitioners’ perspective, this study will provide the basis for analyzing how

efficiently their companies are creating and building their portfolios of strategic alliances in R&D and, therefore, how well they are building technological capabilities. Further, beyond the immediate context of the empirical analysis, the model has implications for other instances where real options perspectives have been deemed appropriate in previous research. These include entry into uncertain product and international markets. In our analysis here, we have found some interesting results. We postulated and then confirmed that the correlation between the outcomes of strategic alliances diminishes the individual option value and, therefore, favors termination through divestiture. Secondly, we proposed and found evidence that fungibility between the resources of a focal firm and the alliances diminishes the cost of the alliance and, therefore, favors termination through acquisition. Of late, the appropriateness of the real options lens for examining exploratory activities has come under question. Adner and Levinthal (2001) argue that the real options formulations are neither accurate nor adequately rich to capture the path-dependent, endogenous and complex

• rganizational processes that resolve uncertainty and determine firm-specific outcomes. Our

empirical analysis of biotech/pharmaceutical R&D alliances clearly suffers from such over- simplifying assumptions. However, our basic model does take into account the firm specific potential for sharing inputs across projects as well as other insights from the diversification

• literature. In this sense, we hope that future models will build further on this attempt to enrich the

literature on exploratory activities, even more in light of Chi’s (2000) assertion about the need to explicitly modeling the option structure. Our empirical study has several obvious limitations. For one, the sample is restricted to a particular industry and, further, to those firms that appear in BioScan, possibly reducing the generalizability of results. We would need to show evidence of applicability to broader samples in order to counter this limitation. A second limitation is that the study does not consider non- equity strategic alliances such as research agreements. It can be argued that such alliances also

SLIDE 15

15 represent firm options, and should be considered a legitimate part of a firm's portfolio. Our measure of technological distance is based on BioScan classification of technological domains, and suffers from its inherent limitations. Also, target cash constraints and target past performance (e.g., discoveries, patents) may also significantly explain some governance choices. Lastly, observations for different governance choices are not balanced. There are relatively fewer observations for buyout than divestiture. We hope that all these limitations do not significantly affect the support for our model and propositions. It is also interesting to compare the implications and assumption of the real options and the network approaches. Recent studies have used network theory for generating interesting hypotheses and explaining ambiguous empirical findings regarding alliances. Although the network literature substantially contributes to the understanding of the configuration of the portfolio of alliances, it has several limitations relative to the line of inquiry of this study. In general, network theory suggests that a distinct cooperative element in the behavior of firms

• exists. In such a framework, strategic alliances are a means of enhancing knowledge generation

(Powell et al., 1996). Viewed in this light, the boundaries of the firm become much more fuzzily defined and the formation of strategic alliances can no longer be viewed as a clear-cut

• transaction. In sharp contrast, the real option lens uses transactions as its main element of
• analysis. Another critical limitation of network theory is that it assumes the existence of a social

structure, but has only begun to explain its formation. These characteristics discourage the use

• f network theory for the analysis of the trade-off between flexibility and commitment in

exploration activities. Besides overcoming the above limitations, future research needs to consider some other related theoretical issues. One, we need to explicitly consider the effect of the presence of

• competitors. Second, we should develop our analysis to permit a clear comparison with results
• btained using network theory. We also hope to better understand how this theory relates to the

diversification literature The real options lens can be quite useful in understanding how firms can cope with exogenous events in their technological and market domains. However, previous real options models have generally made somewhat unrealistic assumptions. This study enhances this literature by integrating the conclusions from the diversification literature in a real options formulation.

SLIDE 16

16 REFERENCES Adner R, Levinthal D. 2001. What is not a real option: finding boundaries for the application of real options to business strategy. Working Paper, Wharton School. Bishop M. 1994. Corporate Governance. The Economist, Jan 29. Black F, Scholes M. 1973. The pricing of options and corporate liabilities. Journal of Political Economy 81: 637-659. Burgers WP, Hill CWL, Kim WC. 1993. A Theory of Global Strategic Alliances: the Case of the Global Auto Industry. Strategic Management Journal 14: 419-432. Capron L, Dussuage P, Mitchell W. 1998. Resource redeployment following horizontal mergers and acquisitions in Europe and North America, 1988-1992. Strategic Management Journal 19(7): 631-661. Chang SJ. 1995. International expansion strategy of Japanese firms: capability building through sequential entry. Academy of Management Journal 38(2): 383-407. Chi TL. 2000. Option to acquire or divest a joint venture. Strategic Management Journal 21(6): 665-687. Cohen WM, Levinthal DA. 1990. Absorptive capacity: a new perspective on learning and

• innovation. Administrative Science Quarterly 35: 128-152.

Deeds DL, Hill CWL. 1998. An examination of opportunistic action within research alliances: evidence from the biotechnology industry. Journal of Business Venturing 14: 141-163. Dixit AK, Pindyck RS. 1994. Investment Under Uncertainty. Princeton University Press: Princeton NJ. Folta TB. 1994. Innovation Through Quasi-Integration: an Application of Option Theory to Governance Decisions in the Biotechnology Industry. Unpublished Doctoral Dissertation. Purdue University, West Lafayette, IN. Folta TB. 1998. Governance and uncertainty: the tradeoff between administrative control and

• commitment. Strategic Management Journal 19(11): 1007-1028.

Folta TB, Johnson DR, O'Brien J. 2001. Uncertainty and the likelihood of entry: an empirical assessment of the moderating role of irreversibility. Working Paper #1143. Purdue University. Folta TB, Miller KD. 2001. Buyout options in equity partnerships: the effects of uncertainty and

• rivalry. Strategic Management Journal - Forthcoming.

Galambos L, Sturchio JL. 1998. Pharmaceutical firms and the transition to biotechnology: a study in strategic innovation. Business History Review, 72(Summer), 250-278. Gulati R. 1995a. Social structure and alliance formation patterns: a longitudinal analysis. Administrative Science Quarterly 40: 619-652. Henderson RM, Clark KB. 1990. Architectural Innovation: the Reconfiguration of Existing Product Technologies and the Failure of Established Firms. Administrative Science Quarterly 35(9): 9-30. Johnson H. 1987. Options on the maximum or the minimum of several assets. Journal of Financial and Quantitative Analysis 22(3), 277-283. Khanna T, Gulati R, Nohria N. 1998. The Dynamics of Learning Alliances: Competition, Cooperation, and Relative Scope. Strategic Management Journal 19: 193-210. Kogut B. 1991. Joint ventures and the option to expand and acquire. Management Science 37(1): 19-33. Kogut B, Kulatilaka N. 1994a. Operating flexibility, global manufacturing, and the option value

• f a multinational network. Management Science 40(1): 123-139.

SLIDE 17

17 Kogut B, Zander U. 1992. Knowledge of the firm, combinative capabilities, and the replication

• f technology. Organizational Science 3(3): 383-397.

Krimsky S. 1991. Biotechnology and Society: The Rise of the Industrial Genetics. Praeger: New York. Kulatilaka N, Perotti EC. 1998. Strategic Growth Options. Managment Science 44(8): 1021- 1031. Kumar R, Nti KO. 1998. Differential learning and interaction in alliance dynamics: a process and outcome discrepancy model. Organizational Science 9: 356-367. Levinthal DA, March JG. 1993. The myopia of learning. Strategic Management Journal 14(Winter): 95-112. Li J. 1995. Foreign entry and survival: effects of strategic choices on performance in international markets. Strategic Management Journal 16: 333-351. Luehrman TA. 1998b. Strategy as a portfolio of real options. Harvard Business Review September-October: 89-99. Madhok A. 1997. Cost, value and foreign market entry mode: the transaction and the firm.” Strategic Management Journal 18(1): 39-61. Mang PL. 1998. Exploiting innovation options: an empirical analysis of R&D intensive firms. Journal of Economic Behavior and Organization 35: 229-242. March JG. 1991. Exploration and exploitation in organizational learning. Organizational Science 2: 71-87. McGrath RG. 1997. A real options logic for initiating technology positioning investments. Academy of Management Review 22: 974-996. McGrath RG. 1999. Falling forward: real options reasoning and entrepreneurial failure. Academy

• f Management Review 24(1): 13-30.

McGrath RG, MacMillan I. 2000. The Entrapreneurial Mindset. Harvard Business School Press: Boston MA. Mitchell W, Singh K. 1992. Incumbents' use of pre-entry alliances before expansion into new technical subfields of an industry. Journal of Economic Behavior and Organization 18: 347- 372. Mowery DC, Oxley JE, Silverman BS. 1998. Technological overlap and interfirm cooperation: implications for the resource-based view of the firm. Research Policy 27: 507-523. Penrose ET. 1959. The Theory of the Growth of the Firm. Basil Blackwell: London. Powell WW, Koput KW, Smith-Doerr L. 1996. Interorganizational collaboration and the locus of innovation: networks of learning in biotechnology. Administrative Science Quarterly 41(1): 116-145. Reuer JJ, Leiblein MJ. 2000. Downside risk implications of multinationality and international joint ventures. The Academy of Management Journal 43(2): 203-214. Rothaermel FT. 2001. Incumbent's advantage through exploiting complementary assets via interfirm cooperation. Strategic Management Journal 22(June-July): 687-699. Sarkar MB, Echambadi RAJ, Harrison JS. 2001. Alliance entrepreneurship and firm market

• performance. Strategic Management Journal 22(June-July): 701-711.

Seth A. 1990. Value creation in acquisitions: a re-examination of performance. Strategic Management Journal 11(2): 99-116. Stuart TE, Hoang H, Hybels RC. 1999. Interorganizatinal endorsements and the performance of entrepreneurial ventures. Administrative Science Quarterly 44: 315-349.

SLIDE 18

18 Stuart TE, Podolny JM. 1996. Local search and the evolution of technological capabilities. Strategic Management Journal 17: 21-38. Stulz RM. 1982. Options on the minimum or the maximum of two risky assets. Journal of Financial Economics, 10: 161-185. Teece DJ. 1992. Competition, cooperation, and innovation: organizational arrangements for regimes of rapid technological progress. Journal of Economic Behavior and Organization 18: 1-25. Teece DJ, Pisano G, Shuen A. 1997. Dynamic capabilities and strategic management. Strategic Management Journal 18: 509-533. Teitelman R. 1989. Gene Dreams: Wall Street, Academia, and the Rise of Biotechnology. Basic Books: New York. Trigeorgis L, Mason SP. 1987. Valuing managerial flexibility. Midland Corporate Finance Journal 5(1): 14-21. Williamson OE. 1985. The Economic Institutions of Capitalism. Free Press: New York. Williamson OE. 1991. Comparative economic organization: the analysis of discrete structural

• alternatives. Administrative Science Quarterly 36: 269-296.

Zucker LG, Darby MR. 1997. Present at the biotechnological revolution: transformation of technological identity for a large incumbent pharmaceutical firm. Research Policy 26: 429-446.

SLIDE 19

19 FIGURE 1 Technological Overlapping The matrix that results from this pattern is: Firms A B A 1 B 0.5 Firm A Firm B T1 T2

SLIDE 20

20 TABLE 1 Descriptive Statistics for Termination Sample Variable Mean S.D. Min. Max. Divest 0.22 0.41 1 Buyout 0.03 0.16 1 Industry Volatility 0.00038 0.00067 0.00009 0.00562

• Tech. Distance Pharm.-Lab

0.11 0.08 0.41 Min Tech. Dist. Lab-Portfolio 0.20 0.14 1 Interest 0.05 0.01 0.03 0.09 License 0.56 0.50 1 Option 0.33 0.47 1 Foreign Transaction 0.49 0.50 1 Log Pharmaceutical Sales 8.88 0.74 6.62 10.59 # of Pharma. Technologies 23.48 10.86 3 42 # of lab technologies 6.94 5.43 1 30 Industry Returns 0.01 0.03

• 0.08

0.12

SLIDE 21

TABLE 2 Pearson Correlation Coefficients for Termination

Variable 1 2 3 4 5 6 7 8 9 10 11 12 1Divestiture 1.00 2Buyout

• 0.08

(0.10) 1.00 3Industry Volatility 0.01 (>0.1) 0.02 (>0.1) 1.00

• 4Tech. Distance

Pharm.-Lab 0.13 (0.01) 0.03 (>0.1) 0.02 (>0.1) 1.00 5Min Tech. Distance Lab-Portfolio 0.12 (0.05)

• 0.03

(>0.1) 0.00 (>0.1) 0.02 (>0.1) 1.00 6Interest 0.09 (0.10)

• 0.06

(>0.1)

• 0.16

(0.01) 0.07 (>0.1)

• 0.16

(0.01) 1.00 7License

• 0.03

(>0.1)

• 0.03

(>0.1)

• 0.00

(>0.1) 0.09 (0.10)

• 0.01

(>0.1) 0.02 (>0.1) 1.00 8Option

• 0.07

(>0.1)

• 0.11

(0.05) 0.05 (>0.1)

• 0.02

(>0.1)

• 0.04

(>0.1) 0.02 (>0.1) 0.30 (0.01) 1.00 9Foreign Transaction

• 0.03

(>0.1)

• 0.01

(>0.1) 0.08 (>0.1)

• 0.00

(>0.1) 0.06 (>0.1)

• 0.00

(>0.1)

• 0.16

(0.01)

• 0.05

(>0.1) 1.00 10Log Pharmaceutical Sales 0.05 (>0.1) 0.06 (>0.1) 0.08 (>0.1) 0.10 (0.05) 0.18 (0.01)

• 0.17

(0.01)

• 0.05

(>0.1) 0.00 (>0.1) 0.01 (>0.1) 1.00 11# of Pharmaceutical Technologies 0.01 (>0.1)

• 0.02

(>0.1) 0.09 (0.10) 0.08 (>0.1) 0.15 (0.01)

• 0.23

(0.01) 0.11 (0.05)

• 0.03

(>0.1)

• 0.01

(>0.1) 0.21 (0.01) 1.00 12# of lab Technologies 0.10 (0.05) 0.01 (>0.1) 0.04 (>0.1) 0.59 (0.01)

• 0.05

(>0.1) 0.07 (>0.1) 0.06 (>0.1)

• 0.02

(>0.1) 0.03 (>0.1) 0.08 (0.10)

• 0.01

(>0.1) 1.00 13Industry Returns 0.05 (>0.1)

• 0.01

(>0.1) 0.14 (0.01)

• 0.03

(>0.1) 0.01 (>0.1) 0.28 (0.01)

• 0.01

(>0.1) 0.07 (>0.1) 0.06 (>0.1) 0.14 (0.01) 0.02 (>0.1) 0.00 (>0.1)

SLIDE 22

22 TABLE 3 The Hazard Rate Maximum-Likelihood Regression Dependent Variable: Alliance Divestiture - N = 435, Failures = 61

Variable Name 1 2 3 4 5 Interest 6.53e+43* (2.82e+45) 1.41e+49* (7.64e+50) 9.28e+43* (4.10e+45) 1.32e+44* (5.68e+45) 5.41e+48* (2.94e+50) License 0.83 (0.25) 0.78 (0.23) 0.78 (0.23) 0.84 (0.26) 0.75 (0.23) Option 0.53† (0.18) 0.53* (0.17) 0.55† (0.18) 0.53* (0.17) 0.55† (0.18) Foreign Transaction 0.70 (0.19) 0.85 (0.23) 0.71 (0.98) 0.70 (0.19) 0.85 (0.23) Log Pharmaceutical Sales 0.54** (0.11) 0.52*** (0.10) 0.52*** (0.10) 0.53*** (0.11) 0.51*** (0.10) # of Pharmaceutical Technologies 0.99 (0.012) 0.99 (0.01) 0.99 (0.01) 0.99 (0.01) 0.99 (0.01) # of Lab Technologies 0.98 (0.02) 0.98 (0.02) 0.98 (0.02) 0.99 (0.03) 0.99 (0.03) Industry Returns 1.37e-08** (8.33e-08) 5.28e-07** (2.95e-06) 6.70e-09** (4.23e-08) 9.86e-09** (5.97e-08) 2.57e-07** (1.54e-06) Industry Volatility

• 5.08e-221†

(1.42e-218)

• 2.2e-204†

(6.0e-202) Min Tech. Distance Lab-Portfolio 4.75* (3.53) 3.78† (3.07)

• Tech. Distance

Pharmaceutical-Lab 0.25 (0.58) 0.46 (0.93) Log-Likelihood

• 51.64
• 46.99
• 49.84
• 51.43
• 45.74

Log-Likelihood Ratio Test 9.31** 3.61† 0.43 11.79** Standard errors appear in parentheses

† p < .10

* p < .05 ** p < .01 *** p < .001

SLIDE 23

23 TABLE 4 The Hazard Rate Maximum-Likelihood Regression Dependent Variable: Alliance Buyout - N = 435, Failures = 14

Variable Name 1 2 3 4 5 Interest 1998.12 (108414.6) 215.38 (12197.89) 1874.00 (9714901) 31.81 (1712.00) 1.91 (104.95) License 1.05 (0.62) 1.05 (0.62) 1.11 (0.63) 0.96 (0.53) 0.97 (0.56) Option 0.12† (0.14) 0.12† (0.14) 0.11* (0.12) 0.13† (0.15) 0.12* (0.13) Foreign Transaction 0.91 (0.48) 0.93 (0.49) 0.87 (0.47) 0.97 (0.56) 0.99 (0.57) Log Pharmaceutical Sales 0.61 (0.30) 0.61 (0.30) 0.64 (0.27) 0.61 (0.31) 0.61 (0.28) # of Pharmaceutical Technologies 0.99 (0.03) 0.99 (0.03) 0.99 (0.03) 0.99 (0.03) 0.99 (0.03) # of Lab Technologies 1.06 (0.06) 1.06 (0.06) 1.07 (0.06) 1.01 (0.06) 1.02 (0.06) Industry Returns 1.80e-07 (2.09e-06) 3.22e-07 (3.88e-06) 2.04e-07 (2.33e-06) 4.59e-07 (5.35e-06) 1.68e-06 (0.000) Industry Volatility 1.58e-35 (4.70e-33) 1.4e-115 (4.5e-113) Min Tech. Distance Lab-Portfolio 0.01 (0.02) 0.01 (0.02)

• Tech. Distance

Pharmaceutical-Lab 1734.35* (6222.75) 2735.55* (10030.70) Log-Likelihood

• 47.39
• 47.36
• 45.40
• 45.35
• 42.93

Log-Likelihood Ratio Test 0.06 3.97* 4.08* 8.92* Standard errors appear in parentheses

† p < .10

* p < .05 ** p < .01 *** p < .001

SLIDE 24

24 APPENDIX Technological Distance Measure For a hypothetical technological Dyad

m t

A between firms i and j, the representation of the technological distance αd, generated form the technological overlapping α, will be the following matrix: Firms i j i

m

diit

α

m

dijt

α j

m

djit

α

m

djjt

α where αdij = 1-(proportion of i’s technologies that is occupied by j), and αdji = 1-(proportion of j’s technologies that is occupied by i). Notice that the elements of the diagonal are conceptually irrelevant, since they measure the technological distance of a firm with itself. Therefore, they are ignored. Considering the non-diagonal part of the matrix, each element is:

1 1 1 1

∑ ∑ ∑ ∑

= = = =

= =

P v djvt P v djvt divt jvt P v divt P v djvt divt dijt

m m m m m m m m

α α α α α α α α where αdijtm = ijth dyad of the matrix Atm at time tm, i ≠ j, αdivtm = captures if the technological domain v is also covered by firm i at time tm, v = technological domain, and P = total number of distinct domains for the sampled firms at time tm. Each of the α’s is a 0 or 1 value such that      =      =

• .w.

at time j firm in available is domain v if 1

• .w.

at time i firm in available is domain v if 1

m m

t t

m jvt m ivt

α α