Đang tải... (xem toàn văn)
... Innovations, Real Options, Risk and Return: Evidence from the Pharmaceutical and Biotechnology Industries, ” a dissertation prepared by Azizjon Alimov in partial fulfillment of the requirements for the. .. Finance for the degree o f Doctor o f Philosophy to be taken September 2007 Title: INNOVATIONS, REAL OPTIONS, RISK AND RETURN: EVIDENCE FROM THE PHARMACEUTICAL AND BIOTECHNOLOGY INDUSTRIES Approved:... flows from the project at different stages of the project The first row gives the value of the project and the second row gives the risk premium of the project (annualized expected return minus risk- free
INNOVATIONS, REAL OPTIONS, RISK AND RETURN: EVIDENCE FROM THE PHARMACEUTICAL AND BIOTECHNOLOGY INDUSTRIES by AZIZJON ALIMOV A DISSERTATION Presented to the Department of Finance and the Graduate School of the University of Oregon in partial fulfillment of the requirements for the degree of Doctor of Philosophy September 2007 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. UMI Number: 3285588 INFORMATION TO USERS The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleed-through, substandard margins, and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. ® UMI UMI Microform 3285588 Copyright 2007 by ProQuest Information and Learning Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. ProQuest Information and Learning Company 300 North Zeeb Road P.O. Box 1346 Ann Arbor, Ml 48106-1346 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. “Innovations, Real Options, Risk and Return: Evidence from the Pharmaceutical and Biotechnology Industries,” a dissertation prepared by Azizjon Alimov in partial fulfillment of the requirements for the Doctor of Philosophy degree in the Department of Finance. This dissertation has been approved and accepted by: Dr. Wayne Mikkelson, Chair of the Examining Committee Date Committee in Charge: Dr. Dr. Dr. Dr. Wayne Mikkelson, Chair Bruce Blonigen John Chalmers Megan Partch Accepted by: Dean of the Graduate School Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. An Abstract o f the Dissertation of Azizjon Alimov in the Department o f Finance for the degree o f Doctor o f Philosophy to be taken September 2007 Title: INNOVATIONS, REAL OPTIONS, RISK AND RETURN: EVIDENCE FROM THE PHARMACEUTICAL AND BIOTECHNOLOGY INDUSTRIES Approved: ___________________________________________________ Dr. Wayne Mikkelson Based on recently developed real options models, such as in Berk, Green and Naik (1999, 2004), I develop and test hypotheses about the impact o f growth options embedded in innovative investment on the risk and the expected return o f a firm’s equity. I test my hypotheses using a hand-collected dataset on the major innovation activities undertaken by pharmaceutical and biotech firms over the period 1985-2004. The dataset contains detailed information on 112 innovation-targeted acquisitions, 273 highlyinnovative drug introductions and 211 key patent grants underlying these drugs. I find significantly positive abnormal stock returns around the announcements of patent and new drug applications and approvals, and negative abnormal stock returns around acquisition announcements. These results suggest that innovative events provide value-relevant information to the stock market. To examine whether changes in firm value around the events are in part due to predicted revisions in expected return on Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. equity, or cost o f equity capital, I study medium-term post-announcement returns. The results are mixed. I find that patent-related events are followed by positive abnormal returns, which is consistent with the prediction that investors require higher return following the discovery of a relatively risky growth option. However, I find that new drug applications are also followed by positive abnormal returns, which contradicts the prediction that expected returns decline following the conversion o f a growth option into relatively safe assets in place. I then examine the central prediction of the real options models that the new information about growth options affects expected returns by changing a firm’s equity risk. I do not find that the events significantly affect the systematic risk o f innovating firms. Furthermore, changes in the systematic risk do not explain abnormal returns around the event announcements. Taken together, these results do not support the main prediction of real options theory that the systematic risk and hence the expected return o f individual firms change in a predictable manner in response to information about growth options embedded in innovative investment. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V CURRICULUM VITAE NAME OF AUTHOR: Azizjon Alimov GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED: University of Oregon Central Michigan University Banking and Finance Academy, Uzbekistan Samarkand Cooperative Institute, Uzbekistan DEGREES AWARDED: Doctor of Philosophy, Finance, 2007, University of Oregon Master of Business Administration, 2001, Central Michigan University Master of Science, Economics, 1998, Banking and Finance Academy, Uzbekistan Diploma of Economist, Samarkand Cooperative Institute, Uzbekistan AREAS OF SPECIAL INTEREST: Corporate Finance Investments Corporate Investment Decisions and Asset Prices Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ACKNOWLEDGMENTS I gratefully acknowledge the advice and support of my dissertation committee, Bruce Blonigen, John Chalmers, Megan Partch, and especially Wayne Mikkelson (chair). Their guidance and encouragement are ultimately responsible for this dissertation. I also wish to thank Jonathan Berk, Ro Gutierrez, Ilan Guedj, Ron Giammarino, and seminar participants at the University of Oregon, the 2006 FMA Doctoral Seminar in Salt Lake City, and California State University, Fullerton for their helpful comments. I thank the University of Oregon and the Hopewell/Racette Scholarship for financial support. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V ll TABLE OF CONTENTS Chapter Page I. INTRODUCTION ...................................................................................................... 1 II. THEORETICAL PERSPECTIVES ......................................................................... Real Options Models and Innovative Investment ............................................ 12 12 Empirical Implications....................................................................................... 17 The Effect of Competition and Business Cycle................................................ 23 III. THE PHARMACEUTICAL AND BIOTECHNOLOGY INDUSTRIES .......... Background ......................................................................................................... 27 27 Patent Applications and A pprovals................................................................... 32 New Drug Applications...................................................................................... 33 Innovation-Targeted Acquisitions..................................................................... 34 IV. SAMPLE SELECTION AND VARIABLE CONSTRUCTION ........................ Sample Selection................................................................................................ 35 35 Dataset Construction.......................................................................................... 39 Information on New Drug Innovations ....................................................... 39 Information on Patents .................................................................................. 44 Information on Innovation-Targeted Acquisitions...................................... 46 Independent Variables: Equity Risk and R eturn.............................................. 49 Control Variables ............................................................................................... 53 V. EMPIRICAL ANALYSIS........................................................................................ Descriptive Statistics of Innovative Investment E vents................................... 58 58 Abnormal Announcement Returns.................................................................... 64 Medium-Horizon Post-Event Abnormal Returns.............................................. 70 Changes in Risk Around Innovative Investment Events.................................. 78 Returns and Changes in Risk ............................................................................. 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. viii Chapter Page Returns, Risk Changes and Growth O ptions................................................. 93 Discussion ........................... 100 VI. CONCLUSIONS.................................................................................................. 103 APPENDICES............................................................................................................. 106 A. VALUE AND RISK OF THE INNOVATIVE PROJECT OVER TIME 106 B. SAMPLE NEWS REPORTS............................................................................ 107 BIBLIOGRAPHY....................................................................................................... 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ix LIST OF FIGURES Figure Page 1. Timeline of Events and Predicted Changes in Risk and Returns..................... 19 2. Estimates of Time and Cost of Innovative Drug Development ..................... 30 3. Average Raw and Risk-Adjusted Returns for the Three Categories of Firms in Event T im e........................................................................................... 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X LIST OF TABLES Table Page 1. List of Sample F irm s....................................................................................... 36 2. Temporal Distribution of Innovative Investment Events............................. 59 3. Descriptive Statistics....................................................................................... 60 4. Announcement Period Abnormal Returns .................................................... 67 5. Medium-Horizon Post-Event Abnormal Returns.......................................... 73 6. Changes in Systematic and Idiosyncratic Equity R isk .................................. 81 7. Regression of Returns on Changes in Risk Controlling for Firm Characteristics: Patent Applications............................................................... 8. Regression of Returns on Changes in Risk Controlling for Firm Characteristics: Drug Applications ................................................................ 9. Regression of Returns on Changes in Risk Controlling for Firm Characteristics: Innovation-Targeted Acquisitions....................................... 10. Regressions of Returns and Risk Changes on Growth Options and Firm Characteristics: Patent Applications............................................. 11. Regressions of Returns and Risk Changes on Growth Options and Firm Characteristics: Drug Applications............................................... 12. Regressions of Returns and Risk Changes on Growth Options and Firm Characteristics: Innovation-Targeted Acquisitions ..................... Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 89 90 91 97 98 99 1 CHAPTER I INTRODUCTION I investigate whether information about real options embedded in a firm’s innovative investment affects the risk and expected return of a firm’s equity. An increasingly important component of many firms’ portfolio of projects is technological innovations, which can potentially generate new products and markets. Innovations in turn are the result of deliberate multi-stage research and development activities, which are usually characterized by multiple sources of uncertainty about the potential benefits and costs of the investment. Because an innovative project involves a sequence of contingent investment decisions, the project can be viewed as a compound real option on the variables underlying the project value, such as the expected time and costs to completion and the stream of cash flows after completion. These options can have substantial economic value by generating future decision rights for a firm. As reviewed below, real options theory claims that information about the growth options embedded in a firm’s innovative investment plays an important role in determining the required return of a firm’s equity. One area to which my study contributes is a relatively recent but rapidly growing literature that applies real options methods to corporate finance and asset pricing problems. The theory of real options recognizes that firms represent continuously evolving portfolios of ongoing projects and growth options to invest in new projects. Berk, Green and Naik (1999, 2004), Gomes, Kogan, and Zhang (2003), Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2 and Carlson, Fisher, and Giammarino (2004, 2006) theoretically demonstrate that a firm's systematic risk, and hence its expected return, reflect the firm’s exposure to the systematic risks of assets in place and growth options. The relative distribution of assets in place and growth options changes with the external acquisition and internal development of new growth options, the exercise of growth options as well as with other investment decisions. Therefore, a critical implication of the real options models is that important investment events will have a direct impact on the risk and required return for a firm’s equity. Despite the theoretical development of real options theory, little empirical evidence documents the impact of real options on firms’ equity risk and expected return. Existing studies, which have examined some of the implications of real options theory, lack necessary firm-level data and are able to provide only indirect evidence. For example, Anderson and Garcia-Feijoo (2006) and Xing (2006) document an association between the capital investment rate, a proxy for the exercise of growth options, and the information contained in the Fama and French (1993) size and book-tomarket portfolios. Carlson et al. (2005) document an increase in firms’ beta prior to seasoned equity offerings and a decrease thereafter, and argue that this result is generally consistent with a real options explanation of stock underperformance following seasoned equity offerings. In contrast, my paper directly tests the empirical implications of real options theory by identifying discrete instances of the discovery and exercise of growth options and relating the economic characteristics of these options to changes in the measures of firms' equity risk and return around the event. In addition, I Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. focus on investment (not financing) events related to a discovery and exercise of growth options that could be financed from multiple sources. My paper also contributes to the literature that studies the stock return implications of firms’ innovative activities. While real options models argue that movements in the market valuation of such firms reflect the time-varying exposure of firms to risks of growth options, there has been popular support for the notion that the market participants often misreact to innovation news with uncertain economic value. Most of the studies in this area examine the association between accounting data on firms’ research activities and subsequent stock returns. For example, Chan, Lakonishok, and Sougiannis (2001) and Eberhart, Maxwell, and Siddique (2004) report a positive association between the level of and changes in R&D investment and subsequent abnormal stock returns. These studies conclude that the market underreacts to the potential benefits from the firms’ research activity. Daniel and Titman (2006) find that a stock’s future return is unrelated to the firm’s past accounting-based performance and negatively related to the firm’s past non-accounting based performance. They argue that this is evidence that investors overreact to intangible information, such as information about the firm’s growth options. In contrast, my goal is to deepen our understanding of how firms’ innovative activities affect the expected return on their equity in two ways. First, building on real options theory I propose and examine the hypothesis that innovative events are associated with significant changes in firms’ cost of equity capital, and that these changes can explain the pattern of abnormal returns around the event. Real options theory implies that ignoring growth options embedded in most Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. investment projects will tend to understate the risk and value of the projects. As a result, I test whether changes in risk and return around the innovative investment events are related to the properties of growth options embedded in these investments. These results should help distinguish between the risk and investor misreaction stories. Second, I examine a wider class of innovation events than has been examined and measure information about innovation events directly from its source. Therefore, my tests are arguably more direct and should provide meaningful additional evidence. Rather than studying the implications of real options theory in a variety of industries, however, I focus on innovative investments of firms in the pharmaceutical and biotechnology industries. While my focus on firms within the same industry restricts the generality of results, it has four advantages. First, the pharmaceutical and biotechnology industries are one of the most research-intensive industries in the U.S. and real options represent a significant portion of pharmaceutical and biotech firms’ value. The real options of these firms are clearly demarked and come in the form of new patents and drugs created by substantial investment in research and development. Second, the theoretical literature makes specific predictions as to the effects of a pharmaceutical firm’s drug research and development activity on the expected return on its equity. In particular, Cornell (1999) and Berk, Green and Naik (2004) demonstrate how a pharmaceutical firm’s progress towards completion of a drug development project drives the required return for a firm’s equity. Third, due to the Federal Drug and Food Administration regulation, a great deal of data is available on the activities related to the discovery and development of growth options by biotech and pharmaceutical Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. firms. These activities result in highly visible events describing progress towards completion of the project, such as new patent and drug applications. A researcher can also obtain important value-relevant information to compute proxies for the value and risk of real options embedded in innovative investment events. Other critical variables can also be identified and controlled for. Finally, this approach allows me to limit the possibility of contamination of results due to mis-measurement of key variables or the effects of other variables not included in the analysis. For example, it is extraordinarily difficult to obtain comparable project level data across firms. As a result, the properties of real options for a large sample of firms in different industries are likely to be measured with much error. For these reasons, I believe that my sample provides the most promising setting for conducting an investigation of how a firm’s decisions related to the development and exercise of growth options affect the risk and expected return of a firm’s equity. I develop and analyze a unique hand-collected dataset that focuses on two distinct categories of innovative activities that should have particularly important effects on the risk-retum o f characteristics of firms, namely, the discovery and exercise of growth options. Specifically, I explore my research questions using the data on three different innovation events: (i) announcements of a highly-innovative new drug application and approval; (ii) announcements of an application and approval of a patent underlying these drugs; and (iii) announcements of innovation-targeted acquisitions. The patenting and introduction of highly-innovative new drugs are one of the most important innovative investments of biotech and pharmaceutical firms. Before a Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. potentially profitable drug can be marketed, it must be approved by the Federal Drug Administration. A patent grant provides monopoly rights to sell the drug until the patent expires and, therefore, gives the firms exclusive access to a growth option. Innovationtargeted acquisitions are important means through which established pharmaceutical firms obtain new technologies and associated skills as well as promising products. In sum, I argue that these events are important events in the life-cycle of the sample firms and result in significant changes in the composition and risk of firms’ assets. As a result, I expect these events to affect investors’ perception of innovator firms’ riskretum characteristics. I use the real option model of Berk, Green and Naik (2004) to formalize hypotheses about how these innovation events should impact the systematic and idiosyncratic equity risk and thus the expected return. Their model captures different sources of risk embedded in the multi-stage innovative investment project, and the authors analyze their interaction in determining the risk premium and value of the project over its life-cycle. Specifically, there is systematic risk associated with future cash flows the project can potentially produce after completion. There is also idiosyncratic risk pertaining to the uncertainty about the successful completion of the project. The project involves several separate stages of development, and at each stage, the managerial decision of whether to continue the investment depends on the resolution of systematic and idiosyncratic uncertainty. Therefore, an investment project is viewed as a compound real option on systematic uncertainty. The models shows that the systematic risk, and thus the expected return, of the project are likely greatest at the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 7 onset, and decrease as the project approaches completion and idiosyncratic and systematic uncertainty gets resolved. The intuition behind this result is that growth options, because of the implicit leverage they impart, have higher systematic risk and therefore expected returns than the asset on which it is written. As a result, the model implies that important events in the project’s life-cycle that impact its value and risk premium, will also influence a firm’s systematic risk and the expected return for its stock. In the context of my study, I expect both the patent and acquisition events to result in an increase in a firm’s equity risk and thus the expected return because these events signal to the market that a firm expands its portfolio of growth options. Conversely, I expect the drug application and approval events to decrease a firm’s equity risk and expected stock return because a firm is exercising its growth options and converting them into less risky assets in place. Both predictions reflect the assumption that growth options have higher required risk premium and, hence, expected return than the assets in place when the underlying asset of the option closely mimics assets in place. Finally, I hypothesize that the impact of the events on the equity risk and return of firms depends on the intensity of product-market competition and the business cycle. I start my analysis by examining the short and medium window stock price reaction to the event announcements. I conduct tests with stock returns for three reasons. First, such tests permit me to directly examine the pricing effects of news about growth options. Second, by studying post-event patterns of stock returns I can measure the economic compensation for the event-induced change in the systematic risk of the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. firm required by investors. In efficient markets, events that change the equity risk a firm should also change expected stock returns in the same direction. Therefore, I expect that firms with a predicted event-induced increase in their equity risk will exhibit an increase in their expected returns. In contrast, I expect that firms with a predicted decrease in their equity risk will exhibit a decrease in their expected returns. Third, by examining whether return movements following the events are explained by changes in equity risk, I can measure the violations of real options theory in economic terms. The results show that that all three innovative events have significant stock price effects and thus appear to possess substantial information content. I find positive abnormal returns to the announcements of patent grants as well as applications and approvals of new drugs. For example, the average firm announcing a patent grant and a new drug approval experiences the cumulative abnormal returns of 1.45% and 2.2% over the five-day window around the event. In contrast, the average firm announcing an innovation-motivated acquisition realizes a negative abnormal return of 1.9%. In the medium-window tests, I find that patent-related events and applications of new drugs are followed by positive abnormal returns that last for at least nine months after the events. The positive abnormal return pattern after the patent-related events is consistent with the prediction that investors require higher rate of return in response to new information about the discovery of a risky growth option. However, I do not find a corresponding pattern of increasing stock returns following innovation-driven acquisitions. I also do not find a pattern of stock returns moving downward following the drug applications and approvals, or the exercise of growth options, as would be the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 9 case if the exercise of growth option were to reduce the equity risk and expected returns of innovator firms. In order for these abnormal returns to be consistent with an efficient price formation, the level of equity risk measures in the post-event period must be different from that before the event. To measure changes in the systematic risk and idiosyncratic equity risks of innovator firms, I estimate the Fama and French (1993) three-factor model at the firm-level. The Fama and French model suits the purpose of my study particularly well because, according to Berk, Green and Naik (1999) and Gomes, Kogan and Zhang (2003), the size and book-to-market factors in the model capture the components of a firm’s systematic risk attributable to its growth options and assets in place. Therefore, the analysis in this paper should help us understand whether the real options story is a good description of behavior of asset prices around innovative events. The results show that, in general, firms do not experience changes in their systematic equity risk, measured as both changes in individual loadings on the FamaFrench three factors and change in the overall risk premia. However, I document an increase in idiosyncratic equity risk following the patent events and a decrease in idiosyncratic equity risk after the announcement of drug approval and innovationmotivated acquisitions. While the observed reductions in idiosyncratic risk are economically significant, it is not clear how the changes in non-compensated risk can explain the stock return movements following the event. To test the proposition that seemingly abnormal returns that follow the test events are induced by risk shifts, I estimate multivariate regressions of the post-event abnormal stock returns on the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 10 changes in the systematic and idiosyncratic risk. The results show that, in general, changes in both measures of equity risk do not explain the post-event patterns of abnormal stock returns. My final analysis then directly examines the central prediction of real options theory that the economic properties of growth options embedded in innovative investment are the main determinants of changes in equity risk and expected stock returns around the innovation news. I find that cross-sectional differences in the announcement abnormal stock returns are positively related to the relative value of growth options embedded in these events. However, I do not find that economic properties of growth options are systematically related to changes in the firms’ systematic and idiosyncratic risks that follow the events. Taken together, my results are both interesting and puzzling, in particular, for the current generation of real options models. The results do suggest that announcement of innovation news provide value-relevant information to investors, who rationally revise firms’ stock prices in response to new information about the value of the underlying growth option. However, I do not find support for the central prediction of the real option theory that the new information affects expected stock returns by influencing firms’ cost of equity capital. O f course, the interpretation of the tests is certainly limited by my need to use a proxy for the unobservable model of expected returns. As a result, my tests are joint tests of the changes in the pricing of assets in place and growth options around the event and the Fama and French factors as priced risk factors. Nevertheless, at a minimum, my results cast doubts on arguments in Berk, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Green and Naik (1999) and Gomes, Kogan and Zhang (2003) who argue that changes to the firm’s portfolio of growth options are able to explain the abnormal returns estimated using the Fama and French three-factor model. Perhaps, a new asset pricing model is needed to explain the excess returns following changes in the composition of firms’ portfolios of assets. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 12 CHAPTER II THEORETICAL PERSPECTIVES In this section, I present the theoretical arguments for the impact of innovative investment events on firms’ equity risk, and thus the expected return on their equity. The goal is to make the intuition as clear as possible by synthesizing the main arguments of this literature and my research questions. Section 2.1 presents and discusses a simplified version of the real options model of Berk, Green and Naik (2004) and discuses its implications regarding the impact of growth options embedded in the innovative project on the required risk premium and expected returns of firms. Section 2.2 gives an interpretation of the model in the context of biotech and pharmaceutical innovative activities and develops the empirical implications. In Section 2 .3 ,1 discuss how product-market competition and the business cycle, not considered in the real options models, may affect empirical results. Real Options Models and Innovative Investment The theory of real options provides a model of investment decisions involving upfront irreversible cost commitments in the face of uncertainty and the choice of incurring additional costs in the future. Under these conditions, a firm’s investment policy includes rights or options, but not the obligation, to make further investments or delay such investments. These investment options therefore create value for the firm by generating future decision rights. Once uncertainty is reduced, a firm can then choose to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. exercise only those options that are ‘in the money’ and allow the other options to expire. A good survey of the real options literature can be found in Dixit and Pindyck (1994). Recently, a new strand of research has emerged that uses real options theory to relate stock return dynamics to firms’ optimal investment decisions. This literature, pioneered by Berk, Green, and Naik (1999), includes Carlson, Fisher, and Giammarino (2004, 2006), Gomes, Kogan, and Zhang (2003), Berk, Green, and Naik (2004), Cooper (2006), Kogan (2004), and Zhang (2005). In this paper, I use the Berk, Green, and Naik (2004) model to develop hypotheses on the impact of real options embedded in innovative projects on the firms’ cost of equity capital and expected returns. I summarize here a simplified version of this model and its implications. The model begins when a firm starts a new project to develop a new product at time 0. The new project will generate a stream of stochastic cash flows after the firm successfully completes a number of discrete stages of research. The value and risk premium of the project is determined by its idiosyncratic and systematic risk. These two sources of risk allows the model to embed the capital budgeting problem in an asset pricing framework, which is suitable to address the main question of my paper. The source of systematic risk of the project is uncertainty about the future cash flows after the project is completed. Although the firm does not receive cash flows prior to the completion of the project, it is assumed that the firm and investors knows what the cash flows would be if the project were complete today. These cash flow change as more information is obtained through investment and learning about the demand for the product. The source of idiosyncratic risk is technical uncertainty regarding the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 14 successful completion of the project, as well as the competitive threat. Technical uncertainty can be resolved only through investment. At each stage of the investment process, the decision of whether it should continue to the next stage depends on the resolution of systematic and unsystematic uncertainty. This gives to the project the characteristics of a series of compound options to invest, where the strike price of the option is the expected investment cost to complete the project and the underlying asset is the value of future cash flow. Thus, the value of real options embedded in the innovative project comes from systematic and idiosyncratic uncertainty regarding future cash flows and fixed investment costs. In the model, the market value of a firm at any time is given by VALUE=A+G(X, I) (1) Where VALUE=the firm’s market value A=fhe present value of the expected cash flows generated by the assets-inplace(ongoing projects) G=operator representing an American call option A=the present value of the expected cash flows generated by the project 7=the expected investment cost required to complete the project Consequently, the firm’s overall cost of capital or risk premium at any time reflects the firm’s exposure to risks of the assets in place and growth options, as described by Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 15 AIP MsKim = R is k . Value + Riskao GO Value (2 ) Where Risk =the risk premia of firm assets AIP= assets in place GO=growth options These equations show that events that contain new information about the value and risk of growth options embedded in the project ought to have an important effect on the risk premium and therefore expected returns of the firm. Note that the additive form of the equations is strictly appropriate only when the values of the assets in place and the asset underlying the growth option are described by separate stochastic processes, i.e. there is no interaction between the two assets. To better understand the effect of such information, however, it is necessary to discuss how the value and risk premium of the project varies through its life-cycle. To derive the value and risk premium of the project, the model works backwards in a dynamic fashion. Suppose that the firm completed the project and is ready to commercialize the product. In the parlance of the option literature, this is equivalent to exercising a growth option. At this stage, the value of the project depends only on the cash flows to be generated from the project and the risk premium of the project simply equals the systematic risk of the underlying cash flows. The intuition here is that after completion of the project no further investment decision is needed, and therefore the growth option has the same characteristics as the underlying cash flow. This will not be the case, however, during the period of development of the project. Prior to the completion, the risk premium of the project is higher than that of the underlying asset Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 16 and increases as one moves back from the completion. The intuition underlying this result is that a growth option represents a claim on future cash flows from the projects that a firm can realize by incurring additional fixed R&D expenditures. Those future fixed payments represent the operating leverage, and the Black-Scholes model implies that this implicit leverage, which is reflected in option betas, should be priced. This leverage makes the option riskier than the underlying asset. The risk premium of the project, therefore, is directly related to the expected future payments or the strike price of the option as well as to the value of the option. For example, completion of a stage of the project reduces the future fixed payments and leads to an increase the value of the option. The net result will be a decrease in the risk premium of the project. The option valuation of the project also takes into account an important element of managerial flexibility: the option to suspend or “mothball” the project should the revised cash flow forecasts turn gloomy. Because the innovative investment typically involves multiple stages, the manager makes investment decisions sequentially and can choose to suspend the project in unfavorable situations. Therefore, the option to suspend the project is particularly valuable to the firm in early stages of the project and helps to reduce the overall risk of the project. In general, as future cash flows increase, the value of the option to suspend decreases because it is less likely that the firm will suspend the project. To illustrate the potential magnitude of changes in the value and risk premium of the project over its life-cycle, Appendix A reproduces a table from the Berk et al. paper. They consider an R&D project that requires 20 phases to be completed. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 17 table displays the value and risk premium of the project for three different levels of expected cash flows from the project at different stages of the project. The first row gives the value of the project and the second row gives the risk premium of the project (annualized expected return minus risk-free rate). The numbers in parentheses show the value and risk of the project without the option to suspend or abandon the project. As it is evident from this table, holding the expected cash flows constant, the value of the project increases and its risk premium decreases as stages are completed. In early stages of the project, where real options represent substantial leverage, the risk premium is at least twice than that of the completed project. The option to abandon, predictably, is more valuable in the early stages of the project as the uncertainty and thus probability of abandoning the project is higher. In addition, we can observe that risk premia are generally higher for lower levels of cash flows, which usually corresponds to cases where it is optimal to suspend the project. Empirical Implications This simple real options model allows me to develop empirical predictions about how significant events in the life-cycle of an innovative project should affect both the systematic and idiosyncratic components of the equity risk. Because asset pricing theory claims that investors require compensation for bearing systematic risk, changes in the systematic risk of a stock should lead to changes in its expected return in the same direction. In addition, Merton (1987) argues that in the presence of market frictions and incomplete information, the idiosyncratic risk of a stock may be positively related to its Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. expected return because investors cannot fully diversify away idiosyncratic risk. This motivates my analysis of the systematic and idiosyncratic equity risk and stock return effects of innovative investment events with the purpose of providing a more complete understanding of the relation between stock returns and firm investment decisions. However, before proceeding to the discussion of specific hypotheses it is worth mentioning that, as Berk et al. point out, the complexity of the model precludes closedform solution for the value and risk premium of the firm. I chose to present Equations (1) and (2) in that form in order to provide important insights into the sources and magnitudes of the influence of growth options and thus facilitate hypothesis development. My empirical tests therefore address only the relations suggested by Equations (1) and (2), rather than an exact functional form of the real options theory. In the context of the biotechnology and pharmaceutical industries, the new innovative investment project starts when a firm obtains exclusive rights to the growth option, where the underlying asset is a stream of cash flows from a new drug. An innovator firm can obtain exclusive rights to growth options in two ways: 1) obtaining a patent on its own discovery o f a new chemical or biological compound; or 2) acquiring technologies and know-how of existing firms. The innovative investment project ends after the firm completes the research and development stages and files a new drug application with the FDA. Completion of the project and the subsequent exercise of the growth option is equivalent to deciding to obtain FDA approval for the marketing of the drug. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 19 Figure 1 Timeline of Events and Predicted Changes in Risk and Returns Growth option is discovered or acquired. A firm decides to develop A firm periodically decides whether to continue investing in the project based on the expected costs to completion and projected future cash flows the project based on the revised estimates o f the project variables. Growth option is exercised if a firm decides to bring the product to market based on the new cash flow projections. /A nnoun. R> 0 '’Announ. R> 0 _A_ T=0 A T=1 T=2 Clinical trials I, II and III 20 months N months Change in E(R) >0 Change in Risk >0 -------- ■20 months Change in E(R) < 0 Change in Risk < 0 Events: Patent Applications and Approvals Events: Drug Applications Innovation-Targeted Acquisitions Drug Approvals I argue that the initiation and completion of valuable innovative projects are important events for the biotech and pharmaceutical firms, and should be duly noted as such by investors. Figure 1 illustrates the time-line of the innovative investment events considered in this paper as well as the predicted impact of these events on the equity risk and expected returns of innovator firms. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 20 Prediction 1 All else equal, patent application and approval events increase the systematic and idiosyncratic risk ofpatenting firms and their expected returns. The immediate stock market reaction to the announcement is positive. As discussed below, patents provide biotech and pharmaceutical firms exclusive rights to growth opportunities by securing their claim to commercialize a new chemical or biological entity. Therefore, I argue that patents signal to the market that a firm has discovered a growth option, which will require additional investments before the value of the underlying asset is realized. Because a new growth option now accounts for a certain fraction of firm value, patenting firms will experience an increase in their cost of equity capital, which in turn leads to higher expected returns. This prediction reflects an assumption that the asset underlying the option closely mimics assets-in-place and thus the growth option is riskier than the assets in place. I also expect a positive immediate reaction by the stock market to the patenting event. Equation (1) shows that the price of a firm’s stock increases in the value of growth options. Therefore, an announcement of a discovery of a new growth option should immediately raise stock prices. Prediction 2 All else equal, new drug applications reduce the systematic and idiosyncratic risk o f firms and their expected returns. The immediate stock market reaction to the announcement is positive. A new drug application is submitted to the FDA for review once the development phases have been completed and the firm believes it has sufficient Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 21 evidence for approval. Marketing for the new drug can only begin after the FDA approves the drug. The new drug applications therefore can be considered as a successful completion of the project and the exercise of the growth option. Because exercising the growth option substitutes a riskier asset for relatively less risky assets inplace, the event will lower the firms’ cost of equity capital and, consequently, expected returns. The announcement of a new drug application also raises the value of the project to the firm because the potential cash flows are now more likely. Therefore, the immediate stock market reaction to these announcements is expected to be positive. Prediction 3 All else equal, innovation-targeted acquisitions are associated with an increase in the systematic and idiosyncratic risk o f acquiring firms and their expected returns. The immediate stock market reaction to the announcement is positive. Established pharmaceutical firms often purchase R&D firms in order to gain access to their technologies and knowledge base. Therefore, the acquirer gains the value not only of the target firm’s assets in place but also of the option to develop and market the target firm’s potential research output. If such acquisitions result in a significant expansion in the acquiring firms’ portfolio of growth options, I expect to observe a positive impact of innovation-targeted acquisitions on the cost of equity capital and expected returns of the acquiring firms. This prediction again reflects an assumption that the asset underlying the growth option closely mimics assets in place and thus the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 22 acquired growth option has higher risk premium and therefore expected return than the assets in place. The immediate market reaction to these announcements is expected to be positive because an acquisition of a new growth option increases the overall firm value and thus should immediately raise stock prices. Prediction 4 Changes in the risk and returns o f an innovator firm 's equity following the innovative investment event is positively related to the relative value o f the growth option embedded in the investment. Equations (1) and (2) show that the impact of the event on the firm’s risk and return is a function of the ratio of value of the growth option embedded in innovative investment to overall value of the firm. Thus, relatively more valuable real options will have a greater effect on changes in the equity risk and returns of innovator firms. The implicit null hypothesis is that the new information about growth options is not systematically related to the risk and return of a firm’s equity. This would hold, for example, if investors fail to take into account the full effect of the new information on the required return for a firm’s equity. Because innovative investments, by their very nature, are characterized by high uncertainty, investors face a daunting task in assessing the proper expected return to compensate for changes in the systematic risk. In addition, because the risk of failure and success of the project is largely idiosyncratic in nature, the argument can be made that the overall risk of innovative projects is a diversifiable risk which would not affect the required rate of return. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 23 The Effect of Competition and Business Cycle It is important to note that investment activities of biotech and pharmaceutical firms are more complex than their depiction in the real options models. A natural concern is that the model, while useful for illustrative purposes, is too restricted. I discuss this issue by considering how product-market competition and the business cycle may affect empirical results. The real options models considered here link firm investment decisions and asset return dynamics in the absence of competitive pressures. In the real world, however, the value of firms’ growth options embedded in investment projects are closely related to competitive interaction in an industry. Investors arguably evaluate not only potential cash flows from innovations, but also the firm’s ability to successfully commercialize and protect their innovations from competitors. Standard economic and finance theory predicts that more competition reduces the monopoly rents that reward successful innovators and thus the value of the firm’s growth options. Schumpeter (1934) suggests that an innovative new product initially earns relatively high profits and attracts imitators, which increases the level of competition. Increased competition erodes profits for innovator firms. Schumpeter (1950) suggests, however, that firms with more market power are able to fend off the competition and thus can earn higher and more persistent returns from new products. Grenadier (2002) shows that under perfect competition, the value of growth options is zero due to the possibility of preemption. Garlappi (2004) also shows that competition among R&D ventures erodes Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 24 the option value to wait, increases the risk premia and uncertainty, and accelerates investment. Sundaram, John and John (1996) find that individual firm’s strategic position in an industry has a significant impact on the firms' stock price response to the announcement of R&D spending. They suggest that the R&D investments should be studied in the context of the industry structure, as well as individual firm’s position in the industry. Aguerrevere (2006) introduces product market competition into a real options model and shows how competitive interactions among firms in a given industry affect the risk of growth options and asset in place. He finds that in the more competitive industries the value of growth options is lower because the firms exercise their growth options earlier to avoid losing the investment opportunity to its competitors. In addition, the risk of assets in place for firms in more competitive industries is higher because the firms have less ability to reduce capacity in economic downturns. This implies that change in firms’ equity risk and returns upon either exercise or acquisition of the growth options is predicted to be lower for more competitive industries. This discussion indicates the necessity to control for the intensity o f product market competition in my analysis. Zhang (2005) argues that the magnitude and the direction of the impact of growth options on the risk and expected returns might depend on the aggregate economic conditions. Zhang argues that assets in place are riskier than growth options, especially in bad times. Zhang’s argument relies on two features of his model, costly reversibility and countercyclical price of risk. Costly reversibility means that it is costlier to disinvest physical assets than to install assets and countercyclical price of risk Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 25 means that the discount rate is higher in bad times such as economic downturns. Together, the time-varying price of risk and costly reversibility result in a higher risk premium of assets in place in bad times. Thus, in contrast to Carlson et al. (1999) and Gomes, Zhang and Kogan (2003), Zhang’s model implies that conversion of growth options into assets in place might actually increase the risk and expected returns. This indicates the necessity to control for the business cycle in my analyses. Finally, it is important to state that the real options approach, as in Berk et al. (1999,2004), is not the only economic analysis of this type of problem. The key insight of the real options models is to stress the functional equivalence between real investment decision and financial contracts. This allows a study of the determinants of the risk premia and expected return of the innovative investment project and the firm as a whole, which is what I focus on. Cornell (1999) presents another theoretical attempt to link firms’ decisions related to the development and exercise of growth options to changes in systematic risk of their equity. The argument is that innovative projects, such as R&D projects, are high-duration assets, i.e. cash flows from these projects tend to come further in the future. Cornell argues that this explains why the market betas of Amgen and other biotech and pharmaceutical companies are much too high to be explained by relatively low correlation between R&D projects’ cash flows and the market. I believe that the Cornell’s paper represents an interesting alternative approach, however I do not examine its implication because it is not clear how to model and measure the duration of individual projects. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In a series of papers, Pastor and Veronesi (2003, 2006) relate changes in the stock valuation of innovative firms to time-varying uncertainty about their growth rates. They argue that an increased uncertainty about the average future profitability raises the firm’s current stock price and idiosyncratic stock volatility. Transposed to my setting, Pastor and Veronesi’s model implies that investment in new innovative projects, because o f high uncertainty about projects’ future profitability, is likely to increase the idiosyncratic volatility and current stock prices of investing firms. Their model is, however, silent on the implications of innovative investments for the risk premium and expected returns of the investing firms. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 CHAPTER III THE PHARMACEUTICAL AND BIOTECHNOLOGY INDUSTRIES Background While I believe the analysis in this paper is pertinent across a range of innovative industries, I focus on the pharmaceutical and biotech industries, as a particularly propitious setting to examine the impact of real options on the equity risk and returns of firms. These are dynamic and innovative industries driven by research and development1. The biotech and pharmaceutical industries are of substantial economic importance, representing well over 10% of the U.S. stock market • • • 2 • • capitalization . In addition, the entire process of drug development is regulated and monitored by the Food and Drug Administration (FDA), which provides a rich source of publicly available data to construct necessary variables. While the firms’ financial performance is driven by innovations, there is enough variation across firm and innovation characteristics to study the cross-sectional effects of innovative activity on the risk and return characteristics of firms. In addition, my restriction of the empirical setting to the biotech and pharmaceutical industries allows for a degree of comparability 1 The pharmaceutical industry is one o f the m ost research-intensive industries in the U .S., with an average research and development (R& D) to sales ratio o f 18%, compared to 4% for the U.S. manufacturing industry overall. In 2004 the R&D expenditures for U.S. pharmaceutical companies totaled $38.8 billion (Pharmaceutical Research and Manufacturers o f America (P h R M A ), Industry Profile 2005) 2At the end o f 2005 the total value o f publicly traded U.S biotech companies was $488 billion (Burrill & Company) and the total value o f publicly traded U.S pharmaceutical companies was around $1 trillion (based on CRSP data). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 28 between projects that would be impossible to obtain in a broad sample. The metrics used to measure the value and risks of real options vary widely by industry. The measurement of these variables across industries is likely to introduce substantial error. By focusing on biotech and pharmaceutical firms as my empirical context, I follow a number of recent studies in finance that also examine the investment and financing decisions of pharmaceutical firms. Lemeret al. (2003) examine the role that the availability of equity financing has on the bargaining power of the research intensive firm and the subsequent allocation of property rights in biotechpharmaceutical alliances. Guedj (2005) and Guedj and Scharfstein (2004) study the effects of agency conflicts on the drug development strategies of pharmaceutical and biotechnology firms. Higgins and Rodriguez (2006) and Danson, Epstein and Nicholson (2004) examine the determinants and announcement effects of pharmaceutical and biotech mergers and acquisitions. However, none of these studies look at pharmaceutical and biotech firms’ investment decisions in the real options framework. A natural question to consider is the extent to which the real options framework discussed in the previous section corresponds to the reality of biotech and pharmaceutical firms. R&D investments of pharmaceutical firms have often been characterized in the literature as investment in the creation of growth options (e.g. Dixit and Pindyck, 1994; Berk, Green and Naik, 2004). Unlike in other industries, growth options of pharmaceutical and biotech firms are clearly demarked as they come in the form of new patents and drugs created after costly and lengthy investment in R&D. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 29 Therefore, a patent and a drug development project can be viewed as a compound option on the expected development costs and the estimated cash flows of the drug. The drug discovery and development process is a rigorous and well-documented sequential process. According to Henderson and Cockbum (1996), there are two stages in pharmaceutical research and development: drug discovery and drug development. The goal of the drug discovery stage is to find a chemical or biological entity that has the desirable therapeutic effect, while the goal of the drug development process is to ensure that discovered entities are safe and effective in humans. The process is shown in Figure 1 as well as in Figure 2, which also gives the success rate of drugs and the length of time for each stage. The development phase of an innovative drug begins when the pharmaceutical or biotech firm identifies a chemical or biological structure, sometimes referred to as molecular entity. At that time, the pharmaceutical or biotech firm usually applies for a patent covering discovered chemical or biological entity. If the firm wishes to proceed with the development of the drug, it files for approval an investigational new drug application with the FDA to test the innovative drug on humans. The human clinical tests occurs over three distinct phases, each of which contributes different amounts and types of information on safety and efficacy of the drug candidate. After completing the clinical trials, the company can seek regulatory approval by filing a new drug application or a biologic licensing application with the FDA. A firm can market the new drug only after the FDA approval. While the FDA is reviewing a drug application, the company attempts to develop the manufacturing and sales infrastructure necessary to market the drug. Following approval, drugs enter the market directly. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 30 Figure 2 Estimates of Time and Cost of Innovative Drug Development Yea rs Pre Clinical Testing Phase I Phase II Phase III FDA Approval 3.5 1 -2 2 -4 4 -6 1.5 Total = 12 - 17 20 to 100 Healthy Volunteers 100-300 Patient Volunteers 1,000 to 3,000 Patient Volunteers c _o Laborato — ry and 3 ao Animal Cl. Studies C /3 Revie w •< Proce C z ss I H fa. Post Marketing Safety Monitoring Determine Safety and Dosage Evaluate Effectivene ss Look for Side Effects. Verify Effectiveness , Monitor Adverse Reactions from LongTerm Use 70% of INDs 30% of INDs 27% of INDs fa. N Manufactur ing fa. Distributio n C /3 C /3 g. c « 'S * ■£ x O 3 05 o ec 20% of INDs Source: International Pharmaceutical Aerosol Consortium on Regulation and Science, October 2003 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 31 The long and uncertain investment cycles of pharmaceutical firms imply that the value of these firms is largely comprised of risky but potentially profitable real options. At the end of each development stage, the firm uses the technological and market information revealed up to that point to decide whether to abandon or continue development of the compound. In other words, at the end of each development stage, the firm decides whether to exercise the option on a drug candidate, abandon the drug candidate due to a lack of commercial viability or sell it to another firm. According to PhRMA, an average brand drug spends 10-15 years in the developmental stage and costs $800M to bring it to market. Furthermore, the research and developmental process is highly uncertain. DiMasi et al. (1990, 2003) find that for every 5,000 to 10,000 compounds that are identified, only 250 reach pre-clinical testing, and that only 20% of drugs that begin Phase I trials will be approved by the FDA. Even approved drugs do not guarantee financial success. Grabowski and Vernon (1990) report that only one in five drugs introduced generated total revenues exceeding average R&D costs of $250 million (1990 dollars). This discussion suggests three highly visible innovative investment events that serve as discrete instances of an internal discovery or external acquisition and the ultimate exercise of a growth option by biotech and pharmaceutical firms. These three events are discussed below. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 32 Patent Applications and Approvals As was discussed above, pharmaceutical and biotech firms file a patent when they discover a new chemical or biological entity that has the potential of developing into a new drug. In the U.S., patents grant exclusive production rights for a period of 20 years from the date of filing for the patent. To apply for a patent, firms have to disclose information to the U.S. Patent and Trademark Office showing that their innovation is indeed novel and non-obvious. The filing of a patent application based on a new mechanism of action is, thus, the first concrete indication that a new scientific discovery can potentially lead to new drugs. Patent applications and approvals, especially those for new chemical or biological entities, are important events for all drug-developing firms. Pharmaceutical and biotech firms’ success in securing patents is viewed as a key mechanism that allows a firm to protect the profits generated by a drug once it is approved. Most approved pharmaceutical products have several patents associated with them. There is obvious concern as to how accurately patent applications dates reflect the actual timing of the discovery of a growth option on a potential therapeutic product. Patent applications usually require identification of the proposed therapeutic product, and it is in the interest of the company to file an application without delay in order to protect patent rights. In sum, the heavy reliance on patents in the pharmaceutical and biotech industries allows for a relatively clean measure of patent applications as economically important innovative events. My use of patents as a measure of growth Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 33 options is consistent with other studies that have treated patents as real options (e.g. Pakes, 1986; Schwartz, 2004). New Drug Applications I use information on new drug applications and approvals as the first observable instance of the exercise o f a growth option. Arguably, the most important measure of innovation for a biopharmaceutical firm is new products marketed to consumers. Newly developed drugs align well with the description of a growth option. An approved innovative drug is protected by patents, which exclude other companies from using the ingredient and production methods defined in patents. In addition, the FDA grants a 5year exclusivity protection to all newly approved new chemical entities. Thus, an approved innovative drug confers on the firm the right but not the obligation to make further investments necessary to commercialize the drug. Introduction of new biotech and pharmaceutical innovative products, and radical innovations in particular, always requires sizeable new capital and labor investments, as well as additional advertising and marketing expenditures. Much of this investment most likely to be irreversible once it is undertaken. It is the incremental value that determines a firm’s decision to apply for the approval of the drug and make additional investment towards commercializing the drug. Therefore, firm’s decision to apply for the FDA approval and then commercialize the new drug is analogous to exercising a growth option. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 34 Innovation-Targeted Acquisitions In 1980s and 1990s, the advent of biotechnology based on genetic engineering, genomics, and other novel research resulted in a radically new technological approach with respect to drug discovery and development. Because established pharmaceutical firms could not keep up with all new methods of drug development through internal research and development, acquisitions of biotechnology firms became important means through which pharmaceutical firms could obtain external innovations as well as knowledge in a timely manner (e.g. Pisano, 2002). I suggest that innovation-targeted acquisitions potentially expand the acquirer’s portfolio of growth options and increase its potential to introduce new drugs. I make this assumption because the majority of the acquired firms in my sample are small biotechnology firms focused predominantly on basic research and drug discovery and development. An alternative to these acquired technologies would be the internal development of these technologies. Therefore, an innovation targeted acquisition arguably can be viewed as a discretely observable instance of a firm externally acquiring a growth option instead of internally developing it. At early stages, it may not be clear that an acquired technology will provide the anticipated profitable products. In the future, based on revised forecasts, the acquirer can either proceed with the development of new technologies or sell the acquired assets. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 35 CHAPTER IV SAMPLE SELECTION AND VARIABLE CONSTRUCTION Sample Selection My sample consists of pharmaceutical and therapeutic-oriented biotechnology firms that develop, manufacture, and market prescription and over-the-counter drugs for human use. The initial sample consists of public pharmaceutical and biotechnology firms listed in Standard & Poor’s Industry Surveys for the years 1985, 1990, 1995, and 2000.1 focus on the period 1985-2004 because the amount and quality of the FDA approval data and news articles on the Lexis-Nexus news database deteriorate as one goes back in time. From this list, I exclude firms that primarily produce nutritional supplements or products for animal health, as well as large conglomerate firms with pharmaceutical divisions. For example, I exclude Alpharma, Dupont, BASF, 3M, Dow and Proctor & Gamble. Biotech firms that are subsidiaries of other firms, but have publicly traded shares are included. For example, I include Genentech, which was partially owned by Roche. For each firm, I collect financial data from Compustat and daily stock return data from the Center for Research and Security Prices (CRSP). I require firms to have daily stock return data on CRSP for at least five consecutive years during the sample period. Occasionally, firms change their fiscal year end, and in these situations they issue more than one annual report in a calendar year. In a few cases when Compustat reports missing or negative data on assets, sales, debt or capital expenditures, I use Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 36 information from firms’ annual reports. To exclude firms without significant economic value I eliminate firms that never reached net sales of $20 million in 2000 dollars. Limiting the sample to firms covered by Compustat and CRSP eliminates all but the largest foreign pharmaceutical firms with significant operations in the U.S. The final sample consists of 116 firms listed in Table 1. Next, I use Moody’s Industrial Manual and International Directory o f Company Histories, and Dun and Bradstreet/Gale 2000 Industry Handbook to determine a sample firm’s subsidiaries. I use the parent and subsidiary names to later search the FDA and U.S. Patent Office databases. Table 1 List of Sample Firms The gvkey number and company name are taken from the Compustat. The permanent number (permno) and the IPO Year (start date) are taken from the CRSP database. Gvkey 1078 112178 13308 24302 15708 1359 1478 1602 24843 29643 29252 24661 21685 1902 13365 2222 2226 122257 29955 2403 2616 Permno 20482 86235 11192 76736 75646 64856 15667 14008 77264 80145 79794 77089 76306 15991 11415 18033 11983 87056 80307 19393 20678 Company name ABBOTT LABS ABGENIX INC AGOURON PHARMACEUTICALS INC ALKERMES INC ALLERGAN INC ALZA CORP AMERICAN HOME PRODUCTS CORP AMGEN INC AMYLIN PHARMACEUTICALS INC ANESTA CORP ARRIS PHARMACEUTICAL CORP ATHENA NEUROSCIENCES INC ATRIX LABORATORIES INC AVACARE INC BARR LABORATORIES INC BIO TECHNOLOGY GENERAL CORP BIOGEN N V BIOMARIN PHARMACEUTICAL INC BIOVAIL CORP INTERNATIONAL NEW BRISTOL MYERS CO SCIOS INC IPO year 1937 1998 1987 1991 1989 1983 1926 1983 1992 1994 1993 1991 1990 1983 1987 1983 1983 1999 1994 1933 1983 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13599 27747 64515 2846 23945 3011 3171 14597 62237 24220 65578 63866 66698 12274 24898 12580 4245 4409 4843 61508 5020 22250 12233 24856 5180 30352 29312 65243 20492 24110 24468 24197 64412 24699 5885 5902 29127 133246 24040 6266 6315 112033 64542 25937 6730 12397 13393 6854 11552 78815 84604 22016 76625 23318 24548 12149 83121 76716 85463 84060 85864 10383 77289 10685 31799 33137 45241 82576 38280 76194 10324 77274 75064 80629 79977 85293 76889 76693 76841 76709 84543 77103 43610 43757 79906 87785 76661 22111 46950 86176 84620 78081 50876 10513 11346 50359 CELGENE CORP CELL GENESYS INC CELL THERAPEUTICS INC CENTOCOR INC CEPHALON INC CHIRON CORP COLLABORATIVE RESEARCH INC COLUMBIA LABORATORIES INC CONNECTIVE THERAPEUTIC INC COR THERAPEUTICS INC CORIXA CORP CUBIST PHARMACEUTICALS INC CURAGEN CORP CYTOGEN CORP DEPRENYL U S A INC DURAMED PHARMACEUTICALS INC ELAN PLC ENZON INC FOREST LABS INC GELTEX PHARMACEUTICALS INC GENENTECH INC GENSIA PHARMACEUTICALS INC GENZYME CORP GILEAD SCIENCES INC GLAXO HOLDINGS LTD GUILFORD PHARMACEUTICALS INC HUMAN GENOME SCIENCES INC HYSEQ INC I A F BIOCHEM INTERNATIONAL INC IC O S CORP I D E C PHARMACEUTICALS CORP I D E X X LABORATORIES INC ILEX ONCOLOGY INC IMCLONE SYSTEMS INC IMMUNEX CORP IMMUNOMEDICS INC INCYTE PHARMACEUTICALS INC INTERMUNE PHARMACEUTICALS ISIS PHARMACEUTICALS INC JOHNSON & JOHNSON K V PHARMACEUTICAL CO KING PHARMACEUTICALS INC KOS PHARMACEUTICALS INC LIGAND PHARMACEUTICALS INC LILLY ELI & CO LIPOSOME COMPANY INC LIPOSOME TECHNOLOGY INC LYPHOMED INC 1987 1993 1997 1982 1991 1983 1982 1988 1996 1991 1997 1996 1998 1986 1992 1986 1984 1984 1967 1995 1980 1990 1986 1992 1972 1994 1993 1997 1991 1991 1991 1991 1997 1991 1983 1984 1993 2000 1991 1944 1972 1998 1997 1992 1970 1986 1987 1983 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 38 7040 14836 24191 21186 24008 7257 62784 7504 7637 30255 29650 7936 61655 8020 12180 30775 127234 61601 8530 11040 24878 15803 23812 12181 20192 9236 62977 29390 9459 24473 212340 14261 10243 133731 30554 14538 29992 63941 65553 17764 121440 11040 24344 30007 27845 12252 28272 47837 11895 76706 76391 76656 22752 83531 54798 69550 80560 80146 57913 82779 63263 10143 81002 86806 82604 21936 82647 77292 78688 76614 10200 75929 39570 88811 80030 25013 76845 85888 75648 37102 88199 78447 75652 81066 84279 85511 75849 87006 26681 76744 80485 78916 10333 79363 MARION LABS INC MARROW TECH INC MEDAREX INC MEDICIS PHARMACEUTICAL CORP MEDIMMUNE INC MERCK & CO INC MILLENNIUM PHARMACEUTICALS INC MOLECULAR GENETICS INC M YLAN LABS INC N P S PHARMACEUTICALS INC NEXAGEN INC NORTH AMERICAN BIOLOGICALS INC N O VAV AX INC NOVO INDUSTRIAL A S ONCOGENE SCIENCE INC ORPHAN MEDICAL INC APPLERA CORP CELERA GENOMICS PATHOGENESIS CORP PFIZER CHAS & CO INC PHARMACIA & UPJOHN INC PROTEIN DESIGN LABS INC QUADRA LOGIC TECHNOLOGIES INC REGENERON PHARMACEUTICALS INC REPLIGEN CORP ROBERTS PHARMACEUTICAL CORP RORER WILLIAM H INC SALIX PHARMACEUTICALS LTD SANGSTAT MEDICAL CORP SCHERING CORP SEPRACOR INC SHIRE PHARMACEUTICALS GROUP PLC SMITHKLINE BEECHAM PLC SYNTEX CORP TANOX INC TARO LTD TEVA PHARMACEUTICAL INDS LTD TEXAS BIOTECHNOLOGY CORP TRIANGLE PHARMACEUTICALS INC TRIMERIS INC U S BIOSCIENCE INC UNITED THERAPEUTICS CORP UPJOHN CO VERTEX PHARMACEUTICALS INC VIVUS INC WATSON PHARMACEUTICALS INC XOMA CORP ASTRAZENECA 1969 1988 1991 1990 1991 1946 1996 1982 1973 1994 1994 1973 1995 1981 1986 1994 1999 1995 1944 1995 1992 1988 1991 1986 1990 1963 2000 1993 1953 1991 1998 1989 1962 2000 1982 1982 1994 1996 1997 1989 1999 1959 1991 1994 1993 1986 1993 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 39 Dataset Construction Information on New Drug Innovations To construct the drug innovations database I use information from the U.S. Food and Drug Administration’s (FDA) Reports on new drug application (NDA) approvals from 1985 to 2004. These reports provide comprehensive information on all approved drugs, including trade name, generic name, firm name, FDA approval date, and withdrawal date. The total number of new drugs introduced by the sample firms from 1985 to 2003 is more than 3,000. However, not all of the approved drugs are considered important in terms of their innovativeness as well as their ability to generate cash flows for the developing firm. Some drugs represent radical new breakthroughs in treatment and some represent new formulation of an existing drug. In addition, bringing a drug to market does not guarantee that the drug will generate large cash flows. Grabowski et al. (2002) report that the distribution of returns to approved drugs is highly skewed and only 34% of drugs introduced between 1990 and 1994 generate returns in excess of the average R&D costs. Because data retrieval is time-consuming process, I chose to retain only a subset of potentially important drugs that represent radical new breakthroughs in treatment and are potentially financially important for the innovator firm. The subset of potentially important drugs is determined using FDA’s classification system and includes new molecular entities (NME) and priority review drugs. Firms in the industry derive a large Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 40 portion of their revenues from the NMEs and priority review drugs because these are the drugs that have patent and FDA exclusivity protection {The National Institute fo r Health Care Management Research and Educational Foundation, 2004). In the late 1970s, the FDA developed a system to accelerate the approval of what it considered to be important innovative drugs. To do this the FDA classifies all the new drug applications according to the novelty of their chemical type and their therapeutical novelty. On the basis of chemical type, the FDA designates drugs as either new molecular entities (NME) or incrementally modified drugs. NME are the most innovative drugs that use active ingredients that have never been approved for the U.S. market. Until recently, NME approvals did not include drugs of biotechnology firms, which are developed using biological ingredients. However, because biological drugs use novel ingredients, I include biological products in the NME category. On the basis of their therapeutical novelty, the FDA classifies drugs into two groups: priority review drugs, which represent a significant therapeutical improvement over available drugs, and standard review drugs, which have no significant therapeutical improvement over available drugs. The FDA gives priority reviews to NMEs that are radical innovations: that is, the first medicines using a new mechanism of action to treat a certain disease. The FDA gives priority reviews to non-NMEs, which provide significant improvement in the treatment of prevalent diseases. This classification also provides a rough measure of the expected value of growth options to the sample firms. In addition to classifying pharmaceutical product innovations using the FDA definitions, I create an alternative measure of potential Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 41 financial importance of a growth option on a new drug to a firm. Because not all approved innovative products are worth the same to the firm, I assess a drug’s potential financial importance using ex-ante equity analysts’ projections of the drug’s annual sales relative to the firm’s current period sales. The analyst projections are typically published in the news reports when a drug advances to the next development stage, as well as when an NDA is filled and approved. Analysts attempt to forecast the future sales of drugs by taking into account the novelty of the drug, the incidence of the disease that the drug targets, potentially additional therapeutic uses of the drug, and the competition in the therapeutic market. In rare cases when the analysts’ estimates were not available, I use either the firm’s own sales estimates or the actual product sales using information in the firm’s reports two years after the drug approval. Appendix B provides an example of the news reports that I used to construct variables. By using a sales-based measure of the value, I expect to create a rather precise measure of the relative value of the growth option to the firm. I expect the impact of the investment event on risk and return to be stronger for firms developing a drug that is expected to generate higher future cash flows relative to firms’ current cash flows. The total number of the new molecular entities and priority review drugs introduced by the sample firms is 340, or about 9 percent of the total number of approved drugs in the sample period. I read the news reports and underlying patents to verify that a firm that filled an NDA (a drug-sponsoring firm) was indeed an original inventor of the drug’s active compound or it is indeed a firm that conducted clinical trials. I dropped drugs which were licensed from another company in late stages of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 42 development or while a drug was already under the FDA review. It would be incorrect to consider a drug licensed in late stages of development a part of the firm’s internal research activity. If a drug was developed in collaboration with more than one company in my sample, the drug application was recorded for each company. I split the expected sales of the jointly developed drugs between the parties as indicated in the collaboration agreement. In cases when a firm applies for more than one NDA within the three-month period, I use only the earlier drug applications. I also eliminated 20 observations from the analysis because of confounding effects of major merger announcements made within the six-month window around the application. Therefore, any post-news patterns are likely due to market reaction to single innovative investment events, not the accumulated reaction to multiple events. Finally, I dropped eight observations where the new drugs are expected to add less than one percent to the innovator firm’s existing sales. These filters lead to a final sample of 273 drugs developed by 91 firms. The number of potentially important drugs developed by each sample firm appears to reflect sample firms’ underlying research focus and activity. For example, Merck is known for a strong focus on in-house research rather than external collaborations for innovation development {The Economist, 1997). This is reflected in my sample where six firms introduced more than 10 highly innovative drugs during the sample period, with Merck and its subsidiaries leading the count with 19 new drug introductions. In order to test whether firms experience a change in their equity risk and return around the event announcements, it is necessary to use only public information known in real time. In the empirical tests, I use both the announcement date of a new drug Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 43 application and the announcement date of a drug approval by the FDA as events dates. I use the application date because it is a better indication of the time of actual decision to exercise the growth option. If investors anticipate the FDA approval of the drug, then the actual approval announcement should provide only marginal new information. The FDA’s approvals statistics indicate that, on average, about 80% of NME applications submitted between 1993 and 2003 were approved, with the approval rate ranging from 46% in 2001 to 100% in 1997. If the market incorporates this fact into its reaction to the announcement of a filing of a NDA, then all else equal, the return response to an approval will be lower than to an application. I obtain the announcements of a filing of a new drug application and the drug approval by the FDA from the news reports in the LEXIS-NEXIS and Wall Street Journal database. There is significant variance in review times across drugs in my sample. By law, the FDA is allowed 6 to 10 months to review an NDA. The actual review time for new drug applications for priority review drugs ranged from 6 to 15 months, for standard drugs ranged from 14 to 27 months and for biologic products ranged from 12 to 32 months. The most serious problem of correctly identifying an event date arise when a firm and investors may have expected to receive its NDA at date t and encountered unexpected FDA delays. In several instances, a company resubmitted an application or used a “rolling NDA” review in which portions of the NDA were submitted to prior to the actual filing of the NDA. In these cases, I used the original filing date. In some cases, the drug was already approved in foreign markets. I Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 44 exclude the NDAs that were approved more than one year before their approval in the U.S. Information on Patents The focus of my analysis is on patents associated with new drugs. To collect data on patents associated with new drugs, I first identify which patents protect a new drug, using the FDA’s publication known as the Orange Book. Unfortunately, the Orange Book version available on the FDA website does not contain information about patents for most of the biological drugs as well as inactive or expired patents. To find missing patents, I manually conducted a keyword search of patents on the U.S. Patent Office website (http://www.uspto.gov/patft/index.htmn using the firm’s (or its subsidiary’s) name and drug’s key chemical or biological ingredient as keywords. The search produced the initial patents database consisting of 779 patents linked to approved therapeutic products. I next read the text of all patents to obtain the name of inventor, application and approval dates, and scope of claims. I then refine the sample as follows. Firstly, I exclude patents not owned by the company filing a new drug application. Secondly, I retain only patents that cover the pharmaceutical compound, or product patents, and drop patents that cover intermediate manufacturing processes, such as formulation or dosage of the drug. I drop patents on intermediate processes because these patents are likely to provide little new information about the value of underlying growth option and, therefore, should have little pricing effect. Thirdly, when a single patent is attached to more than one innovative drug, I link the patent to the first product Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 45 that was developed from the patent. Finally, I eliminated patents that were applied or approved within the six month window around major merger announcements. Imposing all these filters result in a sample of 212 patents. Similar to the drugs sub-sample, I use the patent application date and the patent grant date as event dates. I use the patent application date because it is a better indication of when the market learns about the discovery of a growth option. Given the fact that about 70% of patent applications are eventually granted by the U.S. Patent Office, the patent grants should be at least partially anticipated by the market. I obtained the actual application dates from the patent body. However, when I tried to verify the application dates searching the news reports, I found that firms almost never announce the patent application. Therefore, the patent approval announcement may have the same or even higher information content as the patent application announcement. As was discussed, because patents are obtained in early stages of uncertain drug development, it is infeasible to accurately estimate the monetary value of individual patents in my dataset. To estimate the value of individual patents I draw on the prior literature that has used patent-based statistics as economic indicators (Griliches, 1990; Hall, Jaffe and Trajtenberg, 2001). Specifically, I use the number of citations each patent receives in the subsequent years (plus one) as a proxy for the relative value of a patent. Intuitively, a patent which is cited many times is more likely to be more technologically productive and thus more valuable than a patent which is rarely cited. Hall et al. (2005) establish that patent citations are positively related to market value of firms. I obtain forward citations for sample patents from the NBER Patent Data File Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 46 (Hall, Jaffe & Trajtenberg, 2001), supplemented by the U.S. Patent Office’s patent database. One complication with using citations as a measure of importance is that patents towards the end of the sample tend to have fewer citations, which creates a truncation bias. In order to compare patents across years, I follow Hall et al. (2005) and compute an adjusted count of patent citations as follows: , ^ Patent Citations. Adj_Patent Citations,, = ------ .nil--_-------X ,.„ ... (3) f i * Where the numerator is the unadjusted number of total number of patent citations and the denominator is the average of the number of patents citations in the patent’s cohort (application year and USPTO technical subclass). I construct the patent cohort variables using information from the NBER patent dataset. Information on Innovation-Targeted Acquisitions For each sample firm, I determine its acquisition activity for the years 19852004 by searching press accounts in the LEXIS-NEXUS and Wall Street Journal database. I study each acquisition description in detail to ensure that the acquisition in question was indeed targeted towards acquiring innovations or growth options of other firms. I use two criteria to distinguish innovation-targeted acquisitions from all other acquisitions. First, I read news stories associated with each acquisition announcement to establish whether the acquiring firm reported target firm’s research, technology, or promising pipeline of drugs in development as a the primary motivating factor for the acquisition. Second, I require that the target firm or its leading scientists have obtained Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 47 at least one U.S. patent in the 5 years preceding the acquisition. I retain only those acquisitions for which I was able to find both corroborating news stories and patenting activity. Such innovation-targeted acquisitions arguably serve as substitutes for internal research and development and offer potential for future growth. Therefore, such acquisitions should enhance acquiring firms’ portfolio of real options and have immediate impact on firms’ risk and return characteristics. The acquisition activity in the pharmaceutical industry is complex and ranges from outright acquisition of the firms to acquisition of the branded drugs. Often firms purchase an equity stake in other companies, form drug development or marketing alliances. This makes the definition of a merger and acquisition somewhat arbitrary. I focus on acquisitions in which the acquirer buys at least 50 percent of the target’s equity or an entire division of the target and intends to integrate the acquired assets into the company. Such acquisitions usually require reorganization of firm assets and affect a firm’s portfolio of growth options. To concentrate on investments that are more likely to have a significant pricing effect, I require that the value of the acquisition be greater than $3 million and be at least one percent of the firms’ market value. From news reports I obtain data on the announcement and effective dates of the completed acquisition, the value and method of payment, and percent of shares acquired, if the target is a public firm, and whether merging companies have related therapeutic indications or prior alliances. The financial details for some of the smaller acquisitions are not disclosed and I do not include these acquisitions in the sample. The lack of information on these acquisitions suggests that these acquisitions were relatively Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 48 unimportant. The final acquisitions dataset contains 112 acquisitions made by 59 firms. Genzyme Corp. made 8 innovation-targeted acquisitions in the sample period. In most cases, pharmaceutical companies purchase other firms for specific technologies such as gene therapy and genomics. For example, in discussing the April 1996 acquisition of Corvita, the vice-president of Pfizer said, “The acquisition of Corvita brings new products and biomaterial technologies highly complementary to one of Pfizer’s strategic emphases, interventional cardiology and radiology.” (Pfizer news release, 1996). Less frequently, firms purchase other companies for specific products or product pipelines. For example, in discussing the December 2002 acquisition of Triangle Pharmaceuticals, a Gilead Sciences’ spokesperson said, “This acquisition brings to Gilead not only a late-stage product that could launch next year, but a pipeline of other drugs in development” (Smartmoney.com, 2002). Measuring the value of growths options acquired in these transactions is the key operational issue in testing the hypotheses. Specifically, I need to identify empirical measures that capture the relative value of acquired growth options. Using information from the news accounts, firm annual reports and U.S. Patent Office, I create the following two proxies for the value of growth options embedded in these transactions. Product Potential variable measures the number of products in the target’s pipeline as the sum of launched products and advanced to the Phase 2 and Phase 3 trial stage. Products advanced to the Phase 2 and Phase 3 trial stage are adjusted using the probabilities a potential drug has of receiving FDA approval estimated by Krieger and Ruback (2001). Innovation Potential is the citations weighed patents assigned to the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 49 target firm in the five-year period preceding acquisition. I collect information on patents and their citations for public firms from the NBER patent database and for private firms from the U.S. Patent Office’s website. To account for the citation truncation bias, I adjust simple count of patent citations using Equation (3). Independent Variables: Equity Risk and Return The real option theory suggests that the required risk premium for all equityfinanced firms is the market-value-weighted sum of the risks of firms’ ongoing project (assets in place) and growth options. However, finance theory suggests that for firms financed with both equity and debt, the required rate of return will also depend on financial leverage. Specifically, for a given level of asset risk, greater leverage will increase the required rate of return for an equity holder. Therefore, throughout the paper I focus on changes in the equity risk and stock returns equity after controlling for the changes in financial leverage. I chose this design instead of trying to use unlevered measures of risk and return for two reasons. First, firms in my sample have fairly low leverage levels and removing the impact of financing from the estimated beta had little impact on the estimates. Second, any ad hoc separation of asset risk and financing risk adds potential noise to my research design. I estimate firms’ systematic risk exposures and residual idiosyncratic risk from the Fama and French (1993) three-factor regression model: Rn ~ Rft = a i+ P mkt (Rmi ~ R/,) + PsmbSMB, + Phm, HML, + where Ru is the daily raw return on stock i at time /; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. (4) 50 Rml is the return on the market (the value-weighted CRSP market index); Rft is the return on the one-month Treasury Bill rate; SMB, is the zero investment portfolio representing the return difference between a portfolio of small and large stocks; HML, is the zero investment portfolio representing the return difference between a portfolio of high book-to-market and low book-to-market stocks. My measures of systematic risk are the firm-specific exposure to the market risk factor (Market Beta), to the HML risk factor (HML Beta), and to the SMB risk factor (SMB Beta). A firm’s idiosyncratic risk is the standard deviation of residuals from the regression. The Newey-West procedure is used to correct for autocorrelation and heteroscedasticity. An important point needs to be discussed here, which also applies to the estimates of abnormal returns. As Fama (1970) points out, testing market efficiency also requires a model of market equilibrium. In other words, to link new information about growth options to changes in the systematic and idiosyncratic risk, I must assume that the Fama and French factors capture priced risk in the economy. The obvious concern is that the Fama and French factors are not adequate proxies for the risk factors within the equilibrium multifactor model. This possibility is a concern particularly because the HML and SMB factors are empirically motivated and constructed to explain the deviations from the CAPM. However, there are theoretical and empirical reasons to believe that the Fama and French model is appropriate for testing the real options theory. Fama and French (1993) argue that the HML and SMB are state variables in the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. spirit of Merton’s (1973) Intertemporal Capital Asset Pricing Model (ICAPM). In the ICAPM, the market return does not completely capture all risks relevant to investors, and thus additional factors are required to explain expected returns. More importantly for this study, the real options models of Berk, Green and Naik (1999) and Gomes, Kogan and Zhang (2003) establish the role of size (SMB factor) and book-to-market (HML factor) as determinants of firm’s true systematic risk (market beta). Gomes, Kogan and Zhang (2003) construct a multiple-firm general equilibrium model that links expected returns to size and book-to-market characteristics. They note that a firm’s size captures the components of systematic risk attributable to its growth options, and its book-to-market ratio captures the systematic risk of assets in place. The HML and SMB factors therefore provide good proxies to account for the covariation of each component of systematic risk and the market risk premium. Empirically, Petkova (2006) finds that the SMB and HML factors are indeed correlated with innovations in various variables that describe investment opportunity set. Following the high-frequency or “realized beta” approach used by Lewellen and Nagel (2006), and Andersen, Bollerslev, Diebold, and Wu (2005), I estimate the factor loadings for each stock using underlying daily returns. There are three main advantages of the use of short-windows and high-frequency data to produce estimates of individual firms’ loadings on risk factors. First, the short-window estimate is appealing for the purpose of my study and shows whether infrequent corporate decisions impact risk. The key assumption of this methodology is that the betas are fairly stable during the estimation period, so each short-window regression can produce an unbiased estimate of Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 52 the true conditional alpha and betas. Second, as Lewellen and Nagel (2006) point out, this methodology does not require an assumed model of conditional betas as a function of observed macroeconomic variables, as in Shanken (1990) or Ferson and Schadt (1996). Cochrane (2001) notes that such assumed models are valid only when a researcher knows the full set of state variables available to market participants. Finally, the use of daily returns allows me to use fairly high sampling frequency relative to the horizon of interest, which increases the efficiency of estimates. Merton (1980), and French, Schwert, and Stambaugh (1987) document that equity beta and volatility can be better estimated from data sampled at higher frequency. However, the use of daily returns is subject to nonsynchronous trading problem caused by thin trading. To mitigate this problem, in the spirit of Dimson (1979), I include both current and up to ten lags of risk factor in the regressions, estimating beta as the sum of the slopes on all lags. I impose the constraint that lags 2-10 have the same slope in order to reduce the number of parameters. I should also note that the estimates of changes in risk measures, which are later used in the tests, are computed using an average of 200 days of returns data prior to and after the event. It is necessary to specify that my analysis takes an individual firm as an observational unit. Because I want to study the risk and returns of individual firms and relate changes in these variables to firm-specific measures of growth options, it is not possible to use a portfolio methodology as it produces average values of risk and returns for a set of firms. In addition, Litzenberger and Ramaswamy (1979) and Avramov and Chordia (2006) show that the use of single securities, rather than portfolios, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 53 significantly increases the number of observations and avoids the loss of cross-sectional information that results when stocks are sorted into portfolios. Control Variables In the cross-sectional multivariate tests, I regress changes in innovator firms’ risk and return on the value of growth options embedded in the events and control for a variety of factors that real options theory and prior empirical studies predict will marginally affect changes in risk and expected returns. Competition. Grenadier (2002), among others, predict that the level of competition in a product market has a negative impact on the value of a growth option. Although the use of patents enhances the ability of a pharmaceutical firm to appropriate rents from its innovation, a patent does preclude other firms from producing new similar products under new patents. Therefore, a firm usually faces competition of other patented drugs within the same therapeutic class. I use a drug’s projected market share in its product market as a proxy for the intensity of competition the firm faces in a particular therapeutic market. I either use the analysts’ forecast of the drug’s market share or construct the drug’s market share as the ratio of the analysts’ projections of the drug’s annual cash flows to the overall sales in the drug’s therapeutic market. I obtain the annual sales in the therapeutic market from the news accounts or analyst reports. The definition of the therapeutic market requires some thought. The existing studies define therapeutic markets in several ways, usually at a more aggregate level. If the market is defined narrowly, then small firms could have large market shares if they Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 54 have a dominant position in a particular disease. If, on the other hand, the market is defined too broad then large firms could have small market share in very large therapeutic markets with many other competitors. I take a broader view and define therapeutic markets based on the FDA classification with some adjustments. The FDA classifies all drugs into the 20 major categories. These categories are based on a combination of therapeutic area and chemical mechanism of action. I then subdivide some of these categories according to the specific diseases that the drugs are used to treat. For example, I subdivide the oncology category into breast cancer, prostate cancer, lung cancer, leukemia and other oncology related diseases. In sum, my analysis refers to competition between firms as competition between drugs in the same drug class (e.g. competition between cholesterol-lowering drugs). Use of a firm’s market share is attractive because it reflects the results of competition in the therapeutic market as well as the therapeutic novelty of the drug. A drug will have an initial market share of 100% when no other drugs have similar therapeutic effects to be considered competitors. In contrast, a drug’s initial market share is projected to be negligible when there are already a number of established products with sufficiently similar therapeutic effects. In addition, the projected market share also takes into account the analysts’ expectations with regard to the resources that the firm will deploy to promote, protect and further develop the product. Pipeline. Intuitively, the impact of the event on the firm’s risk and return characteristics will depend not only on the properties of the underlying growth option but also on the properties of other projects being developed by the firm. For example, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 55 the impact of the exercise of the growth option will be smaller when a firm has more projects in its pipeline. I measure the number of other projects in the firm’s pipeline as the sum of citation-adjusted patents the firm received in the five year period prior to the event and the ratio of R&D expenditures to book value of assets measured at the beginning of the period. Another alternative to be investigated in the future is the use of Pharmaprojects, a comprehensive private database on pharmaceutical firms’ pipeline of drugs. Business Cycle. Zhang (2005) predicts that interest rates and general macroeconomic conditions will affect the relative value and risk of growth options versus assets in place. Hence, it is reasonable to believe that changes in risk and returns of firms’ equity following the event can depend on the prevailing business conditions. Following Avramov and Chordia (2006), I use the yield spread between the ten-year Treasury bond and the three-month Treasury bill (TERM) and the one-month Treasury bill rate (RF) as proxies for business conditions. The data on Treasury bond and bill rates are from the St. Louis Federal Reserve Bank. Firm characteristics and other variables. While my goal is to test for the effect of growth options on risk and return, I recognize that the variation in individual firms’ risk and returns may be impacted by firm-specific characteristics. Therefore, in all regressions I also control for other variables that may be correlated with changes in firms’ risk and returns. Campbell (1991) shows that unexpected stock returns must be explained by changes in expected cash flows and changes in expected required returns. Therefore, I need to control for changes in expected cash flows. In addition, I use other Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 56 firm-specific characteristics as controls suggested by the literature. Specifically, Hamada (1972) points out that changes in the capital structure could affect the systematic risk of firm equity; Amihud and Mendelson (1986) argue that investors require higher expected returns to hold stocks with higher transaction costs and lower stock liquidity. O’Hara (2003) suggests that because stock liquidity is related to the price discovery it affects the systematic and idiosyncratic volatility of stock returns. I also include firm size (the log of total market capitalization) among the firm-specific variables because smaller firms tend to be less profitable and less liquid (e.g. Fama and French, 1995). Finally, I also include cumulative abnormal return from the Fama and French regression estimated over the previous twelve months as an additional control for the momentum effect in stock returns (e.g. Jegadeesh and Titman, 1993). In sum, the included firm characteristics are as follows: (i) ROA, the ratio of operating income (Compustat item 13) to book assets (item 6 ); (ii) A ROA, the average of change in operating income from year t - 1 to year t scaled by book assets in year t-1 and change in operating income from year t to year t+1 scaled by book assets in year t; (iii) Size, the natural logarithm of the market value of equity measured two months prior to the event; (iv) Illiquidity, Following Amihud (2002), measured as the daily ratio of absolute stock return to its dollar volume, averaged over the 3 month Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 57 period prior to the announcement. This ratio provides the daily price -3 impact of the order flow ; (v) A Illiquidity, change in Amihud’s measure of illiquidity from the average of the 3 months prior to the event to the average of the 3 months after the event; (vi) Leverage and A Leverage, the ratio of total debt to book value of assets and change in this ratio from year t - 1 to t; (vii) Cash and A Cash, the ratio of cash (item 1) to book value of assets (item 6) and change in this ratio from year t - 1 to t; (viii) Alpha_12: Cumulative abnormal return over the last twelve months. All firm-level ratios are winsorized at the 1% and 99% levels in order to minimize the effect of outliers. 3 Hasbrouck (2003) finds that the Amihud (2002) measure is highly correlated with trade-based measures o f liquidity (transaction costs). Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 58 CHAPTER V EMPIRICAL ANALYSIS This section contains my main empirical tests of the impact of innovative investment events on the risk and returns of firms. I start my analysis by describing the main financial characteristics of the sample. In Section 5.2 and 5.3 ,1 examine the short and medium horizon price reaction to the event announcement. Section 5.4 presents univariate comparisons of risk changes following the event. Section 5.5 studies whether changes in risk can explain abnormal returns around the event. Section 5.6 directly examines whether the properties of growth options embedded in innovation investment are related to changes in returns and risk of my sample firms. In Section 5.7,1 end with a discussion of whether the results are consistent with the real options predictions. Descriptive Statistics of Innovative Investment Events Table 2 tabulates the temporal distribution of innovative investment events. There is a noticeable rise in the number of innovative events in the second half of the sample period, in part reflecting growth in the number of sample firms over time. As was discussed, concentration of innovation-targeted acquisitions in the late 1990s and the early 2000s is a result of the advent of biotechnology based on genetic engineering, genomics, and other novel research. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 59 Table 2 Temporal Distribution of Innovative Investment Events N ew Drug Approvals for each sample firm are identified from the FDA reports and from news reports in the LEXIS-NEXIS and Wall Street Journal Database. Patents are collected using the FDA Orange Book, and U.S. Patent and Trademark Office. Innovation-Targeted Acquisitions are identified from the press accounts in the LEXIS-NEXIS and Wall Street Journal Database. Year 1978 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 Total Patents A cquisitions Drugs 2 3 1 3 3 4 7 5 5 7 16 7 12 10 7 6 16 21 15 16 17 12 7 5 2 2 211 1 5 5 9 2 10 9 16 10 11 14 16 28 22 20 19 10 13 12 23 18 273 3 0 1 0 5 3 2 4 8 5 11 9 15 14 18 5 6 3 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 60 Table 3 Descriptive Statistics This table reports the summary statistics o f the key variables used in the analysis. The definitions o f firm characteristics are as follows (Compustat codes in parentheses). Book Assets is Compustat Item 6. Sales is Compustat Item 12. Market value o f equity is computed as Close Price (199) times Shares (25). Book value o f equity is computed as book value o f equity (60) + deferred taxes (74). Market assets to book assets ratio is the ratio o f market value o f equity plus book value o f debt to book value o f assets (6). R&D is research and development expenditures (46). Capex is capital expenditures (128). Advertisement is advertising expenditures (45). Cash is cash and short-term investment (1). Equity issue is new equity issuance (108-115) and Debt issue is new debt issuance (111-114). Book Debt is long-term debt (9) plus current liabilities (34). Operating Income is Operating Income before Depreciation (13). All these accounting variables are winsorized at the 1st and 99th percentile. Amihud’s Illiquidity, as in Amihud (2002) measured as the daily ratio o f absolute stock return to its dollar volume, averaged over the 3 month period prior to the announcement. Annual stock volatility is the standard deviation o f daily stock returns in the fiscal year. Annual return is the cumulative stock return in the fiscal year. Total number o f patents is the simple count o f patents obtained by the firm over the previous 5 years. Total number o f citations is the simple count o f forward citations received by the firm’s patents. Adjusted number o f patent citations is the simple citations count corrected for truncation in citations. I collect information on patents and their citations for public firms from the NBER patent database and for private firms from the U.S. Patent Office. Mean Variable Name N Panel A: Patent Applications and Approvals Median Std Dev 5978 5108 21960 2922 4.454 0.158 0.06 0.062 0.335 0.081 0.02 0.051 0.16 0.097 0.028 40.8 36.3 283 20.65 3.16 3110 2734 6718 1603 3.68 0.108 0.056 0.033 0.223 0.005 0 0.038 0.13 0.202 0.001 33 27 88 9 1.88 7539 6636 33064 3662 2.85 0.128 0.036 0.077 0.271 0.23 0.083 0.06 0.16 0.226 0.098 20 62 376 29.02 3.9 Book Assets ($m il) Sales ($m il) Market Value o f Equity ($mil) Book Value o f Equity ($mil) Market Assets/Book Assets R&D/Assets Capex/Book Assets Advertisement/Book Assets Cash/Book Assets Equity Issue/Book Assets Debt Issue/Book Assets Book Debt/Market Assets Book debt/Book Assets Operating Income/Book Assets Amihud's Illiquidity Measure Annual stock volatility (%) Annual return (%) Total # o f patents Total # o f patent citations Total # o f adjusted patent citations 211 211 211 211 211 203 211 110 211 180 175 211 211 211 211 211 211 211 211 211 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. N Mean Variable Panel B; Drug Applications and Approvals Median Std Dev 8124 6745 29586 4194 4.423 0.152 0.057 0.053 0.286 0.061 0.036 0.056 0.198 0.103 0.018 38.5 39.4 267 502.45 0.172 5005 3644 11071 2502 3.708 0.108 0.054 0.035 0.196 0.004 0 0.044 0.166 0.194 0 34.6 28.7 68 300 0.1 9221 8657 37365 6682 2.435 0.154 0.03 0.054 0.236 0.21 0.106 0.053 0.182 0.236 0.137 17 83.3 168.74 134.79 0.143 4952 3800.27 17720 3106 3.791 0.224 0.046 0.036 0.375 0.066 0.053 0.082 0.225 0.008 0.049 52 29.8 131 129.44 37.59 351.02 47.1 2.317 964. 266 2711 524 3.497 0.115 0.033 0.021 0.337 0.014 0 0.053 0.183 0.079 0.001 52.7 15.3 22 21.66 13 63 11.174 1.5 8844 8664 34083 8471 2.3 0.263 0.044 0.044 0.276 0.183 0.129 0.086 0.197 0.249 0.166 20.8 77.3 176.54 124.86 96.23 107.4 103.24 3.334 Book Assets ($m il) Sales ($m il) Market Value o f Equity ($mil) Book Value o f Equity ($m il) Market Assets/Book Assets R&D/Assets Capex/Book Assets Advertisement/Book Assets Cash/Book Assets Equity Issue/Book Assets Debt Issue/Book Assets Book Debt/Market Assets Book debt/Book Assets Operating Income/Book Assets Amihud's Illiquidity Measure Annual stock volatility (%) Annual return (%) Total # o f patents Projected Drug Annual Sales ($mil) Drug Sales/Firm Annual Sales 273 273 273 273 273 261 273 132 273 261 232 264 264 273 273 273 273 273 273 273 Panel C: Innovation-Targeted Acquisitions Book Assets ($m il) Sales ($m il) Market Value o f Equity ($mil) Book Value o f Equity ($m il) Market Assets/Book Assets R&D/Assets Capex/Book Assets Advertisement/Book Assets Cash/Book Assets Equity Issue/Book Assets Debt Issue/Book Assets Book Debt/Market Assets Book debt/Book Assets Operating Income/Book Assets Amihud's Illiquidity Measure Annual stock volatility (%) Annual return (%) Total # o f patents Total # o f adjusted patent citations Target's # o f total patents Target # o f patent citations Target # o f adj .patent citations Target's Product Capital 112 112 112 112 112 112 112 30 112 110 98 112 112 112 112 112 112 112 110 112 112 112 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3 describes some of the important characteristics of the sample firms and the events. Firm profitability, financial policy, and investment policy variables are normalized by the beginning-of-year book value of assets. There is considerable variation within the sample for all of the variables of interest. The financial measures are quite skewed to the right, with a few firms that are much larger and more established than the rest. The descriptive statistics indicate that the patenting and drugsponsoring subgroups have similar levels for all of the variables. In contrast, the patenting and drug-sponsoring firms differ from firms acquiring external growth options along some of the dimensions. Acquiring firms are smaller than firms introducing new drugs in terms of total book and market value of assets as well as sales. The median acquirer has about $524 million in book value of equity and $2.7 billion in market value of equity while the median drug-developing firm has about $2.5 billion in book value of equity and $11 billion in market value of equity. The difference is significant at the 5 percent level. Much larger market value of equity compared to book value is a natural consequence of growth opportunities: a firm with many growth opportunities has a higher market value of equity than a firm with few growth opportunities. The median market-book asset ratio of the sample firms is 3.5, much larger than the overall median ratio of 1.1 for all Compustat firms. All firms in the sample spend much more on intangible investments than on tangibles. Take the drug applications subsample as an example. The median ratio of R&D investments to last year’s book assets is 11%, the median ratio of capital investments to last year’s book assets is about 5%, and the median ratio of advertisement expenses to last year’s book assets is 3.5%. The research Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 63 efforts of drug developing firms resulted in a median number of 68 patents over the previous five years, whereas the median firm making innovation-targeted acquisition has a total of 22 patents in the same time period, with the difference being significant at the 5% level. This suggests that both groups of firms have similar research intensity, but realize different outcomes from their research projects. Consistent with the notion that high growth firms have lower leverage, the median patent and drug sponsoring firm has a book debt to market assets ratio of about 4%, a book debt to book assets ratio of about 16%, and it holds 20% of its book value of assets in cash. The acquiring firms have slightly higher debt and cash ratios. Specifically, the median acquirer has a book debt to market assets ratio of about 5%, a book debt to book assets ratio of about 18%, and it holds 34% of its book value of assets in cash. Finally, the median drug developing firm experiences stronger one-year buy-and-hold returns of 29% and lower stock volatility of 34.6% compared to one-year return of 15% and stock volatility of 52.7% for acquiring firms. The differences are significant at the 5% level. The summary statistics also reveal several noteworthy points about the events. Panel A shows that the average patent on underlying drug receives 20 forward raw citations and 3.161 cohort-adjusted citations. The corresponding median numbers are 9 and 1.9, respectively. For information, the average number of raw citations per patent in the Hall et al.’s (2005) study is 7. Panel B shows that new drugs are expected to be an important source of sales for the innovator firm. The average drug is expected to generate $500 million in annual sales, which corresponds to 17% of the average firm’s last year sales. The corresponding median numbers are $300 million and 10%, Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 64 respectively. Finally, Panel C shows that innovation-targeted acquisitions are important external innovative investment events: the median target firm has received 13 patents in the last five years, which generated 63 raw and 11.7 adjusted forward citations. The median target firm has 1.5 products in its pipeline. I also find that the median acquirer pays $190 million or 8% of the firm's market value of equity to acquire the target. Because the relationship between any two variables is likely to be complex, I refrain from presenting and discussing simple correlations between any two variables and postpone inferences until the regression analysis. Abnormal Announcement Returns In order to provide a straightforward test of whether my sample events contain value-relevant information, I start my analysis by studying stock price reaction to the event announcements. As pointed out by Fama (1991), the announcement returns are the most reliable indicator of shareholder value creation or destruction in corporate events. Campbell (1991) shows that unexpected, or abnormal, returns must reflect either changes in expected cash flows or changes in expected return. In efficient markets, expected stock returns are the same as the cost of equity capital. Thus, the implied hypothesis is that at least a part of the price reaction is attributed to changes in investors’ beliefs about the cost of equity capital of innovator firms. For each announcement I compute the announcement effects as the sum of actual stock return less the benchmark or expected return over various event windows. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 65 For example, the cumulative abnormal returns for the five-day window surrounding the announcement date (-2, +2) are estimated as follows: CAR- = 2 K - m „ ) (5) t = T —2 where R„ is the return on firm / and E(R„) is the expected return and T is the date on which firm i announces the event in question. The expected return is computed using the Fama and French three-factor model given in equation (4), where the model parameters are estimated using daily returns from -200 trading days to -15 trading days before the announcement. Significance tests are based on a standardized test statistics and Wilcoxon rank-sum statistics constructed to determine whether the mean and median abnormal returns are significantly different from zero. For robustness, I also calculate the benchmark returns using the more traditional market model as in Brown and Warner (1985). My results remain qualitatively the same regardless of the definition of the benchmark return. I chose to discuss only CARs estimated with the Fama and French model to maintain consistency with the subsequent tests on the changes in the firms’ systematic and idiosyncratic risk. I present the cumulative abnormal returns for the 2-day event window (from day 0 to day 1), 5-day event window (from day -2 to day +2), 11-day event window (from day -5 to day +5), and the 3-day window prior to the event (from day -5 to day -2). I compute the CARs for the longer-windows to guard against mis-measurement of the actual day when the information about the event in question becomes available to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 66 investors, as well as to examine existence of any short-term pre and post-announcement movements in stock prices. Table 4 reports the results for firms grouped by the type of event. The table provides the average and median CARs across firms for the respective event windows, as well as the associated p -values. For the patent and drug sub-samples, I present CARs for both the application and approval dates. Panel A reports results for the patent sub sample, Panel B reports results for the drug sub-sample and Panel C reports results for the acquisition sub-sample. The results in Panel A show that patent application events have statistically significant announcement effects only when measured over the ten-day period (day -5 to day +5). The average CAR of patent applications over the 10-day event window is 1.7% and the median is 1.34%. Both numbers are significantly greater than zero at the 5% level. Note that the average and median abnormal returns for the 2-day (from day 0 to day 1) and the 5-day (from day -2 to day -2) event windows are not significantly different from zero. Because the event dates for patent applications come from the U.S. Patent Office and may not necessarily correspond to actual announcement dates, it is perhaps not surprising to find that most of the price adjustment in my sample occurs in the 10-day period surrounding the announcement. Interestingly, the average abnormal return for the 3-day pre-announcement window (from day -5 to day -2) is positive 0.5%, and is significant at the 10% level. It implies that investors may learn that a firm filed a new patent application before the U.S. patent office officially recognizes it and react accordingly. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 67 Table 4 Announcement Period Abnormal Returns This table reports cumulative abnormal returns, expressed in percent, computed using the Famaand French (1993) three factor model. The CARs are calculated for various event windows for the sample of firms announcing patent applications and approvals, new drug applications and approvals, and innovation-motivated acquisitions between 1985 and 2004.1report the mean and median CARs across all firms. Corresponding p-values from the t-test and the Wilcoxon signedrank test are reported below each mean and median CAR. I also report p-values for tests of the Event Period (days relative to announcement date, t=0) "-5 to 5" "-2 to 2" "Oto 1" "-5 to -3" Panel A: Patents Patent Application (N=211) Mean P-Value Median P-Value 0.494 0.096 0.419 0.017 0.143 0.584 0.258 0.401 -0.236 0.569 0.331 0.560 1.704 0.012 1.34 0.010 Patent Approval (N=212) Mean P-Value Median P-Value -0.023 0.953 0.277 0.993 0.644 0.079 0.402 0.058 1.45 0.010 0.849 0.004 1.546 0.075 1.569 0.006 Difference Mean P-Value Median P-Value 0.473 0.379 -0.037 0.128 -0.393 0.349 0.036 0.503 -1.546 0.021 -0.168 0.075 0.18 0.869 -0.411 0.724 Mean P-Value Median P-Value 0.999 0.003 0.742 0.002 0.77 0.012 0.324 0.046 1.502 0.000 0.578 0.001 2.937 0.000 1.618 0.000 Mean P-Value Median P-Value 0.742 0.010 0.273 0.039 1.875 0.000 0.631 0.000 2.228 0.000 1.108 0.000 3.243 0.000 2.786 0.000 Mean 0.256 -0.726 P-Value 0.540 0.263 Median 0.002 -0.186 P-Value 0.233 0.149 Panel C: Panel C: Innovation-Targeted Acquisitions Acquisitions Mean 0.888 -1.531 (N=l 11) P-Value 0.112 0.050 Median 0.651 -0.146 P-Value 0.098 0.132 -1.105 0.043 -0.116 0.013 -0.306 0.701 0 0.387 -1.925 0.025 -1.498 0.012 -1.587 0.138 -2.266 0.109 Panel B: Drugs Drug Application (N=273) Drug Approval (N=273) Difference Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The average 5-day CAR (from day -2 to day 2) for patent grants is 1.45% and the median is 0.85%. Both CARs are significant at the 5% level. The average and median 11-day CARs (from day -5 to day 5) for patent grants are only slightly greater at 1.5%. In contrast to patent applications, the average 2-day CAR (from day 0 to day 1) for patent grants is 0.64%, and is significant at the 10% level, while the average 3-day pre-announcement CARs (from day -5 to day -2) is indistinguishable from zero. This suggests that most of the reaction occurs immediately around the announcement and there is no leakage of information prior to the event. Also note that the 11-day CARs for patent grants are not statistically different from those of patent applications. This is somewhat surprising because, given that the U.S. Patent Office approves about 70% of patent applications, all else equal, one could expect that stock price reaction to patent applications should be twice as large as that of patent grants. Because sample firms differ vastly in their sizes, I also compute the dollar abnormal return made by innovator firm shareholders (not reported but available on request). To estimate abnormal dollar returns, I multiply the abnormal return earned by the firm over the event window by its market value of equity at the beginning of the event window. For example, the average dollar abnormal return for patent grants over the 5-day event window (from day -2 to day +2) is $100 million and the median is $11 million. If one to assume that the market incorporates the new information into stock prices fully and immediately, the above finding represents a substantial increase in shareholder wealth as a result o f biotech and pharmaceutical firms’ innovation activities. The results confirm that the innovation event that provides new information Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 69 about a firm’s growth options is an economically important event for a typical firm in my sample. Panel B reports results for the drug applications and approvals sub-sample. Among five different event categories, the application and approval of a drug result in the greatest positive average and median CARs. The average abnormal return of a new drug application for the 5-day event window (from day -2 to day +2) is 1.5% and the median is 0.58%. Both numbers are significant at the 5% level. The average 5-day abnormal return of the new drug approval is 2.3% and the median is 1.1%. Also note that both announcements are preceded by positive abnormal returns. Over the 3-day period from day -5 to day +2, the average firm with drug applications and approvals earns abnormal returns of 1% and 0.7%, respectively. Both numbers are significant at the 5% level. This finding implies that investors are expecting the announcement. In unreported results, I find that the average dollar abnormal returns to the new drug application and approval news are $205 million and $260 million, respectively. This result represents a substantial increase in shareholder wealth due to the biotech and pharmaceutical firms’ innovation activities and is consistent with Hypothesis 2 that new information about drug development process causes an immediate jump in the market value of firms. Finally, Panel C shows that the average abnormal return earned by the acquiring firms over the 5-day event window (from day -2 to day +2) is negative 1.9%, and is significant at the 5% level. However, I find that the average abnormal dollar return of acquisitions is positive $77 million, albeit insignificant. Negative stock market Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 70 reaction to innovation-targeted acquisitions does not support the prediction that the new information about an increase in acquiring firms’ portfolio of valuable growth options immediately raises the market value of the firm. However, I find that in the 3-day period before the announcement (from day -5 to day -2), the acquiring firms experience a positive pre-announcement CAR of 0.9%, which is significant at the 10% percent level. This again seems to suggest that investors are expecting the announcement and may have already incorporated the positive information into stock prices. The average CAR over the entire the 11-day period from day -5 to day 5 remains negative, however, suggesting that the positive pre-announcement return movements do not offset the negative announcement return. Medium-Horizon Post-Event Abnormal Returns The short-window tests above document that the stock market views the announcement of the innovative investment event as informative in valuing my sample firms. However, the short-window tests present no clear evidence on whether these events signal an increase or decrease in the equity risk of firms because stock price reaction at announcement may reflect changes in investors’ expectations of future cash flows, the cost of equity capital, or both. The post-event stock returns, however, might shed light on whether these events affect the risk and thus the expected return of firms’ equity. This is because in efficient markets, events that alter the equity risk of the firm should also change stock return expectations in the same direction. In my sample, this argument requires that for stocks with an event-induced increase in their risk, such as Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. stocks of patenting firms, the subsequent return response should be positive. On the other hand, for stocks with a decrease in their risk, such as stocks of drug-sponsoring firms, the subsequent return response should be negative. Furthermore, if risk changes that follow the innovation events contain permanent component, I can make an even stronger prediction. Specifically, the abnormal medium-horizon post-event returns that are based on asset pricing model parameters estimated during the pre-event period should be significantly positive after risk-increasing events and significantly negative after risk-decreasing events. Fama (1998) and Fama and French (1997) also argue that post-event abnormal returns will reflect systematic mis-estimation of expected returns when a study fails to adjust for the event-induced changes in firms’ cost of equity capital. To preserve comparability with the announcement period returns, I sum abnormal returns over various time frames to obtain cumulative abnormal returns (CARs). The stock’s abnormal return for each trading day is computed as the return on the stock minus the expected return predicted by the Fama and French three-factor model estimated using daily returns from -200 trading days to -15 trading days before the announcement. I report the cumulative abnormal returns computed for various event windows following the announcement day 0. To facilitate this endeavor, I consider the average cumulative returns over the following separate time frames: +2 through +30 trading days, +2 through +60 trading days, +2 through +120 trading days, and +2 through +180 trading days. These windows roughly correspond to one, three, six and nine month Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 72 periods after the event and should include enough time to allow the results of the changes in expected returns to manifest themselves in data. I do not examine the longerperiod returns because a typical biotech and pharmaceutical firm makes innovation announcements on a frequent basis. This makes it hard to justify long-run movements in stock returns as a response to publicly available innovation news than it is to explain shorter-term movements. In addition, as Fama (1998) points out, the interpretation of long-term abnormal returns is highly controversial. To check whether the results are sensitive the procedure employed, I also use a procedure suggested by Barber and Lyon (1997) and compute abnormal returns as returns in excess of return of size-and-book-tomarket matched control firms. The results do not change materially and therefore are not reported. I also check whether the results are due to outliers. After dropping outliers, the results remain qualitatively the same. Finally, note that the cross-sectional correlation problem is not very severe in my study because I exclude the clustered news announcements for the same firm and use relatively short time frames to compute abnormal returns. Table 5 reports the average and median cumulative abnormal returns for various event windows separately for the three event categories. The average cumulative raw and risk-adjusted returns are also graphed in Figure 3. The data in Panel A of the table and the illustration in Figure 3 show that following the announcement the patenting firms earn higher than expected returns. The positive abnormal returns can be observed for various event windows. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 73 Table 5 Medium-Horizon Post-Event Abnormal Stock Returns This table reports cumulative abnormal returns, expressed in percent, computed using the Famaand French (1993) three factor model. The CARs are calculated for various event windows for the sample of firms announcing patent applications and approvals, new drug applications and approvals, and innovation-motivated acquisitions between 1985 and 2004.1report the mean and median CARs across all firms. Corresponding p-values from the t-test and the Wilcoxon signedrank test are reported below each mean and median CAR. I also report p-values for tests of the null hypothesis of equality of means and medians across events. Event Period (days relative to announcement date, t=0) "2 to 180" "2 to 30" "2 to 60" "2 to 120" Panel A: Patents Patent Application (N=211) Mean P-Value Median P-Value 3.56 0.03 2.40 0.019 6.08 0.001 4.72 0.000 9.98 0.001 8.44 0.001 18.32 0.001 13.59 0.001 Patent Approval (N=212) Mean P-Value Median P-Value 2.48 0.03 1.92 0.002 2.11 0.16 2.11 0.007 10.80 0.001 7.67 0.001 16.84 0.001 13.91 0.001 Difference Mean P-Value Median P-Value 0.87 0.57 0.75 0.485 3.61 0.1 1.72 0.308 -0.26 0.93 0.57 0.861 1.81 0.61 2.08 0.800 Mean P-Value Median P-Value 2.75 0.02 3.53 0.001 4.17 0.002 4.68 0.001 10.47 0.001 10.17 0.001 13.89 0.001 10.28 0.001 Mean P-Value Median P-Value 1.06 0.21 1.47 0.084 3.28 0.009 3.44 0.003 2.96 0.09 3.08 0.033 3.52 0.11 5.73 0.014 Mean 1.69 0.90 0.12 P-Value 0.58 Median 0.00 0.00 P-Value 0.129 0.313 Panel C: Panel C: Innovation-Targeted Acquisitions Acquisitions Mean 0.82 1.32 (N=l 11) P-Value 0.64 0.59 Median -0.02 1.22 P-Value 0.524 0.706 7.51 0.01 2.00 0.001 10.38 0.003 0.43 0.001 3.06 0.4 5.59 0.182 7.59 0.08 8.15 0.037 Panel B: Drugs Drug Application (N=273) Drug Approval (N=273) Difference Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 74 Figure 3 Average Raw and Risk-Adjusted Returns for the Three Categories of Firms in Event Time The figure plots cross-sectional average raw and risk-adjusted returns surrounding announcements of new patent applications and approvals, new drug applications and approvals, and innovation-targeted acquisitions. Announcement date corresponds to t=0. Risk-adjusted returns are intercepts or alphas from the Fama and French (1993) three factor model estimated at the firm level. The intercepts are then averaged across firms in each day for each event sub sample and then cumulated over the 210 days surrounding the event announcements. Fig 3A. Patents: Raw and Risk-Adjusted Mean Returns ,---- -30 -15 j--------- ,-------- ,--------: 0 15 30 1------ ,--------- ,-------- ,--------- ,--------- 1--------- 1--------- r-------- 1---------r~ 45 60 75 90 105 120 135 150 165 180 Days relative to event ----------- Application-Raw ----------- Approval-Raw ----------- Application-CAR ----------- Approval-CAR Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 75 Fig 3B. Drugs: Raw and Risk-Adjusted Mean Returns Returns in % oCO _ o . o -30-20-10 0 10 20 30 40 50 60 70 80 90100110120130140150160170180 Days relative to event ----------- Application-Raw ----------- Approval-Raw ----------- Application-CAR ----------- Approval-CAR Fig 3C. Acquisitions: Raw and Risk-Adjusted Mean Returns _ Returns in % o in - o -30 -15 0 15 30 45 60 75 90 105 120 135 150 165 180 Days relative to event Raw R e t u r n -------------CAR Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 76 Figure 3 shows that the average patenting firm’s stock price begin rising before the announcement, and the firm outperforms the benchmark on a risk-adjusted basis by slightly more than 6% over the next 60 trading days or roughly 3 months. Average and median CARs grow in a steady pattern over the event window. The average cumulative 120 trading day abnormal return is 10% and the average 180 trading day return is 18%. The corresponding median numbers are 8.4% and 13.6%, respectively. All CARs are significantly different from zero at better than the 5% level. The results are similar for the patent approvals sub-sample. In fact, similar to short-window abnormal returns, the expected return difference at each window between the two patent sub-samples is not significant. The positive abnormal return movements after both patent applications and approvals announcements are consistent with the prediction that the new information about the discovery of a risky growth option increases the cost of capital and expected returns for innovator firms’ equity. To attribute these abnormal ex-post realized returns to changes in return expectations, one must of course believe that investors ex-ante increase the required return on firms’ equity around the patent related announcements. Panel B of Table 5 reports results for the drug applications and approvals sub samples; the same data are displayed in Panel B of Figure 3. The results show that both sub-samples earn positive post-event abnormal returns and that the post-event drift in stock returns is significantly stronger after the announcement of a new drug application. For the drug application sub-sample the average post-announcement abnormal returns for the 120 and 180 trading day periods are 10.5% and 14%, respectively. The Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 77 corresponding median numbers are 10.2% and 10.3%, respectively. For the drug approvals sub-sample the average cumulative 120 and 180 trading day abnormal returns are 3% and 3.5%, respectively. The corresponding median numbers are 3% and 5.7%, respectively. All these returns are significantly different from zero at conventional levels. The direction of the post-event return pattern is, however, opposite to the predicted negative change in expected stock returns. One possible explanation for positive and not negative returns following new drug-related announcements is that these events could lead to unexpected positive changes in cash flows, which then swamp the effect of any changes in the cost of equity capital. In multivariate tests, I attempt to isolate these effects by including proxies for changes in expected cash flows in the regressions. My hope is to measure the marginal impact of changes in risk after controlling for the effect of changes in cash flow expectations. Finally, the results in Panel C of the table show that acquiring firms do not earn abnormal returns in the next 30, 60 or 120 trading days following the innovationmotivated acquisitions. The absence of any post-event abnormal returns is not consistent with the prediction that the acquisition of a risky growth option leads to an increase in the cost of equity capital and, hence, expected stock returns. The data in the table and its depiction in Panel C of Figure 3, however, shows that the cumulative abnormal stock response becomes significantly positive after 120 trading days following the event and remains that way for at least the next 60 trading days. The overall result is that the average cumulative 180-day abnormal return is positive 7.6%, and is significantly different from zero at the 10% level. However, I am reluctant to Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 78 draw inferences from this result because it is harder to attribute longer-term returns to a particular news item than shorter-term returns, especially when most of the post-event period returns show no particular pattern. In sum, the medium-window tests of abnormal stock price response following the innovation events show that, consistent with the real options prediction, stock returns move upward after firms receive patents, or relatively risky growth options. However, I do not find support for a predicted similar increase in stock returns following innovation-targeted acquisitions. I also do not find a pattern of stock returns moving downward following the drug applications and approvals, or the exercise of growth options, as would be the case if the exercise of a risky growth option were to reduce the equity risk and, hence, expected returns of innovator firms. However, positive stock price movements following the patent and drug events could also result from unexpected positive changes in cash flows. Because the behavior of firms’ equity risk is more important for my purposes, I now explore the changes in the systematic and idiosyncratic risk of innovator firms around the events. Changes in Risk Around Innovative Investment Events In this section, I examine the prediction on the impact of innovative investment events on the equity risk of innovator firms. The purpose of this exercise is to examine the extent to which changes in stock returns following these events are due to investors’ revisions of estimates of the systematic and idiosyncratic equity risk in response to the new information about firms’ growth options. For the above cumulative abnormal Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 79 returns to be consistent with an efficient stock price formation and thus with real options theory, the level of equity risk in the post-event period must be different from that measured during the pre-event period. Further, the direction of risk changes should be the same as the direction of return movements following the innovation events. I measure changes in the systematic risk of equity using the Fama and French (1993) three-factor regression equation (4) and idiosyncratic risk using the residuals from this model. In particular, I look for changes in factor loadings that occur after the innovative event. My tests are similar to those used by Grullon, Michaely and Swaminathan (2002), Boehme and Sorescu (2002), and Chen, Shevlin and Tong (2007), who study the effect of dividend changes on changes in the market’s estimate of systematic risk. To estimate shifts in risk factor loadings, I estimate the modified version of equation (4) for each firm, using 400-day estimation interval consisting of 200 trading days prior to the announcement date and 200 trading days after the announcement date. More precisely, I estimate the following equation: R-it ~ Rfl = a i « A ,A + P mKT (R-mt ~ + P m MKT^I i^m t ~ + Psmb^MB' + fiA/smDtSMBt + PHM, HMLt + PAiHM[DtHMLt + /iit + where R jt is ^ the daily raw return on stock i at time t, R jt is the daily return on the one-month Treasury Bill rate, Rmtis the daily return on the value-weighted market portfolio (CRSP market index), SMBt is the daily return on the stock zero investment portfolio representing the return difference between a portfolio of small and large stocks, HMLt is the daily return on the zero investment portfolio representing the return Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 80 difference between a portfolio of high book-to-market and low book-to-market stocks. Dt is a dummy variable that is equal one when the time period relates to the 200 trading days subsequent to the end of firm i’s event month, and zero otherwise. I omit 15 days before and 15 days after the announcement date. P mki ^ P sm ; a n d P hmi. denote the factor loadings for firm i with respect to market beta, size and book-to-market factors prior to the event. My focus is on the coefficient estimates MKTA , /?A(SMBDt,and/?A,HM, Dt , which represent the change in systematic risk of the firm after the event. a ' represents the risk-adjusted abnormal • return (alpha) of firm i before the investment spike and oc D A/ ' is the change in abnormal return after the investment spike. The annualized standard deviation of the residual from this regression is my measure of a firm’s idiosyncratic risk. For each sample firm with an innovative event I estimate the model over the event window. In testing for statistical significance, I report t-statistics that use the average factor loading based on all firm-month observations. To assess the economic impact of risk changes on the innovator firms’ cost of capital, I multiply the changes in the factor loadings with the risk premium for each factor measured by its average historical return. The average annual returns on the market, SMB and HML factors over the period 1926-2004 are 7.8%, 2.92%, and 4.87%, respectively. These average annual returns on the risk factors imply positive compensation for bearing factor risk. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 81 Table 6 Changes in Systematic and Idiosyncratic Equity Risk This table summarizes and compares the changes in risk factor loadings for around the event dates for all three innovation events. The actor loadings are estimated using the following model over a 400-days, comprising 200 days prior to and 200 days after the event: ^it ~ Rft = a , + a &,D, + PMKJ (K , ~ Rfi) + /Amat A i^ml ~ Rft) + P s m S M B , + P a,smbD ,S M B , + P hmi.H M L i + P mhm iA H M L , + At + A At (3) where Rjt is the daily raw return on stock i at time t, RJt is the daily return on the one-month Treasury Bill rate, Rmtis the daily return on the value-weighted market portfolio (CRSP market index), SMBt is the daily return on the stock zero investment portfolio representing the return difference between a portfolio of small and large stocks, HMLt is the daily return on the zero investment portfolio representing the return difference between a portfolio of high book-tomarket and low book-to-market stocks. D, is a dummy variable that is equal one when the time period relates to the 200 trading days subsequent to the end of firm i’s event month, and zero otherwise. I omit 15 days before and 15 days after the announcement date. Idiosyncratic Risk is the annualized variance of the residuals from this regression. The overall risk change is computed by multiplying the changes in the factor loadings with the average annul factor returns over the 1926-2004 period. I report pooled p-values next to the risk measure. N Panel A: Alpha Beta p-value Patent Applications Beta p-value Patent Approvals 211 0.063 0.000 0.086 0.000 211 0.009 0.629 -0.009 0.639 211 0.984 0.001 1.010 0.000 211 0.060 0.499 -0.073 0.375 211 0.335 0.008 0.326 0.014 211 0.053 0.674 0.027 0.841 211 -0.679 0.000 -0.561 0.000 P mhmi.D 211 0.289 0.036 0.108 0.444 Overall risk change 211 0.019 0.022 -0.119 0.912 211 0.352 0.001 0.380 0.001 211 0.013 0.074 0.017 0.043 « a/A Market Factor P mkt /W rA Size Factor P smb Pm SMBA Book-to-market factor P hml Idiosyncratic risk VarC^1) Var( A A t ) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 82 N Panel B: Alpha Beta p-value Drug Applications Beta p-value Drug Approvals «,■ 273 0.080 0.000 0.112 0.000 « a,A Market Factor 273 -0.013 0.366 -0.088 0.000 P mkt 273 1.037 0.000 0.989 0.000 P mmkt ^ i 273 -0.117 0.123 -0.011 0.889 PsMB 273 0.222 0.034 0.234 0.015 P m SMb Q 273 -0.031 0.774 -0.124 0.232 P hml 273 -0.471 0.000 -0.534 0.000 P mhmiJ-* 273 -0.003 0.984 0.125 0.306 Overall risk change 273 -0.009 0.370 0.0058 0.556 273 0.379 0.000 0.361 0.000 273 -0.017 0.075 -0.014 0.119 Panel C: Innovation-Targeted Acquisitions Alpha 112 0.12 «,■ 0.000 Size Factor Book-to-market factor Idiosyncratic risk VaK^*) Var(D'^ " ) 112 -0.061 0.0624 112 1.137 0.000 112 -0.07 0.6087 PsMB 112 0.642 0.0003 P&i SMBD, 112 -0.003 0.9872 P hml 112 -0.646 0.0001 P muM, D 112 0.2 0.3292 Overall risk change 112 0.0033 0.0084 112 0.502 0.000 112 -0.024 0.1237 Market Factor P mkt P aiMKj D, Size Factor Book-to-market factor Idiosyncratic risk VarC^") Var( D>^" ) Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 Table 6 summarizes the results on risk characteristics separately for the patenting, drug approval and acquisition sub-samples. I present average alphas, factor loadings, and idiosyncratic risk prior to and following the innovation event announcement. The results in the table indicate that the patenting firms have an average market beta that is slightly below one, and the drug sponsoring and acquiring firms have an average market beta that is slightly above one. Thus, the sample firms are as risky as the market portfolio. The average SMB beta shows more dispersion, ranging from 0.22 for the drug-sponsoring firms to 0.64 for the acquiring firms. The average HML beta is between -0.47 for drug-sponsoring firms and -0.68 for patenting firms. The negative loadings on the HML factor suggest that the sample firms behave like growth stocks. All pre-announcement factor loadings are significant at the 5% level. The Fama and French three factors, however, cannot completely explain variation in stock returns prior to the event. Under the hypothesis that the three-factor model explains average returns, the intercepts or alphas in the regressions should be zero. The evidence in the tables, however, shows that all event firms earned significant positive alphas during the 200-day period prior to the event announcement. In terms of changes in the risk characteristics of the sample firms, I focus on P mmktA >PrusMBA >and HM[ Dt , which provide estimates of the post-event changes in firm-specific factor loadings. Contrary to the prediction, the table shows very few changes in the factor loadings across the event firms. Further, the changes in factor loadings are mixed with some loadings increasing and some decreasing. As could be seen in Panel A of Table 6 , 1 find significant changes only for the sub-sample of patent Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 84 applications. The average firms announcing a new patent application experiences a large increase in overall risk premium equal to approximately 1.9 % per year, which is significantly greater than zero at the 5% level. The increase in risk is driven by a change in the HML beta, which increased by 0.28, or by more than 30% from its pre-event level. From the examination of risk changes for the other groups of events, the only consistent result that emerges is that, there is no evidence that any of the other innovation events are associated with a significant increase or decrease in firms’ cost of equity capital. The data in the table provides some evidence that the events change the idiosyncratic risk of innovator firms. The average firm announcing a new patent application experiences an approximately 1.3% increase in its idiosyncratic volatility, with significance below 10%. As a percentage of the average patenting firm’s previous risk level, this increase represents a 3.5% increase in firm-specific risk. In contrast, firms filing a new drug application and firms undertaking innovation-motivated acquisitions experience a significant decrease in their idiosyncratic risk. The average drug sponsoring firm realizes a 1.7% decline in its idiosyncratic risk, which represents a 4.4% decrease in firm-specific risk from its pre-event level. Finally, the average acquirer realizes a 2.4% decline in its idiosyncratic risk. As a percentage of the acquiring firm’s pre-event idiosyncratic risk level, this reduction represents a 4.8% decrease in firm-specific risk. All these numbers are significantly different from zero at conventional levels. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The observed economically and statistically significant changes in idiosyncratic risk around the events merit some discussion. Asset pricing theory puts forth that risky assets provide investors with expected returns that are positively related to the systematic risks of the assets. In this theory, the idiosyncratic risk is diversified away and, therefore, not priced. Merton (1987), however, suggests in the presence of market frictions and incomplete information, the idiosyncratic volatility of a stock may be positively related to its expected return because investors cannot fully diversify away idiosyncratic risk. This proposition therefore requires that events that change the idiosyncratic risk of a firm should also change expected returns in the same direction. However, the results for the sub-sample of drug applications are not consistent with this proposition. Specifically, return movements following the new drug applications are positive while changes in idiosyncratic risk are negative. Therefore, while my results may suggest that changes in idiosyncratic risk are related to changes in stock returns, their interpretation is not clear and debatable. Further, whether idiosyncratic risk should be positively priced in firm-level stock returns has been the subject of considerable debate in recent years. In contrast to Merton (1987), Ang, Hodrick, Xing, and Zhang (2006) find that stocks with high idiosyncratic volatility, measured using the Fama and French model, tend to have lower expected returns than stocks with low idiosyncratic volatility. Spiegel and Wang (2005) dispute this result and show that stocks with high expected idiosyncratic risk earn high returns. Nevertheless, in the next section I formally test whether abnormal return patterns that follow the innovation events are caused by shifts in the firms’ equity risk. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 86 Returns and Changes in Risk In this section, I examine the importance of changes in systematic and idiosyncratic risk as well as changes in cash flow expectations and other firm fundamentals in explaining the abnormal returns earned by sample firms following the events Specifically, I estimate the following equation: CAR = f (ASystematic _ Risk , AIdiosyncratic _ Risk, Firm Characteristics) (7) Where CAR = cumulative abnormal returns computed over the 60,120 and 180 day time frames following the event; /^Systematic _ Risk = the change in risk premium of the firm after the announcement computed over the time frame corresponding to the respective CAR. This is computed by multiplying the change in factor loadings with the corresponding risk premium; AIdiosyncratic _R isk = the change in annualized idiosyncratic risk of the firm after the announcement estimated over the time frame corresponding to the respective CAR. This is computed by as difference between the post and pre-announcement levels of the standard deviation of the residual from the Fama and French model; Firm Characteristics include the following variables: ROA, the ratio of operating income (Compustat item 13) to book assets (item 6); A ROA, the average of changes in operating income from year t - 1 to year t scaled by book assets in year t-1 and from year t to year t+1 scaled by book assets in year t; Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 87 Size, the natural logarithm of the market value of equity measured two months prior to the event; Illiquidity, following Amihud (2002), measured as the daily ratio of absolute stock return to its dollar volume, averaged over the 3 month period prior to the announcement; A Illiquidity, change in Amihud’s measure of illiquidity from the average of the 3 months prior to the event to the average of the 3 months after the event; Leverage and A Leverage, the ratio of total debt to book value of assets and change in leverage from year t - 1 to t; Cash and A Cash, the ratio of cash (item 1) to book value of assets (item 6) and change in this ratio from year t - 1 to t; Alpha_12: Cumulative abnormal return over the last twelve months. As was discussed, unexpected changes in cash flows, financial leverage and stock illiquidity can complicate the retum-risk association. My hope in this test is to measure the marginal impact of a risk change on change in expected returns and thus to provide evidence on the links among changes in expected returns, changes in equity risks, and news about growth options. If abnormal returns are caused by changes in the systematic or idiosyncratic equity risk, the CARs should be related to the magnitude of the risk change. More precisely, I expect the coefficient on the change in the systematic and idiosyncratic risk to be positive because firms with positive changes in risk should have higher post-announcement abnormal returns. I use clustered standard errors to account for correlation across observations for a given firm and White’s (1980) correction to account for heteroscedasticity. I would like to note that the cross-sectional correlation problem should not be severe in my Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 88 study because I have excluded the clustered news announcements and the examined time frames are much shorter in my study. Tables 7, 8 and 9 present regression results separately for the sub-samples of new patent applications, new drug applications and innovation-targeted acquisition. Column 1 shows results for the 60-day post-event period CAR (from day -2 through day 60), Column 2 shows results for the 120-day post-event period CAR (from day -2 through day 120), and Column 3 shows results for the 180-day post-event period CAR (from day -2 through day 180). Consider the Table 7 regressions for the sub-sample of patenting firms. The main result is easy to summarize. Contrary to the prediction of the real options models, the results indicate that there is no relation between changes in systematic risk premium and any of the different post-event CARs. More specifically, the estimated coefficient on changes in systematic risk is not significantly different from zero in any of the regression specifications, suggesting that changes in systematic risk following the innovative events have no impact on post-event stock price movements. Similarly, the estimated coefficient on changes in idiosyncratic risk is also not indistinguishable from zero in any of the specifications, indicating that changes in idiosyncratic risk do not explain any variation on the stock returns following the event. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 7 Regression of Returns on Changes in Risk Controlling for Firm Characteristics: Patent Applications This table summarizes the coefficients estimated from the following regression equation: C A R —f (ASystematic _ Risk, Aldiosyncratic _ Risk, Firm Characteristics) Equation and variables are described on pages 86 and 87. A,Systematic _ Risk Aldiosyncratic _ Risk Ln (Market equity) ROA _1 Cashl Leveragel Illiquidity Alpha_12 Change in ROA Change in Cash Change in Leverage Change in Illiquidity Constant Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Observations R-squared Event Window (relative to event day t=0) "+2 to 180" "+2 to +60" "+2 to 120" -0.213 0.006 -0.025 -0.97 -0.67 -0.04 -0.382 -0.378 -0.168 -1.34 -0.96 -1.45 -0.059 -0.037 -0.011 -0.86 (2.06)** (2.60)** 0.039 0.071 0.148 -0.2 -0.84 -0.92 -0.038 0.038 0.096 -0.22 -0.49 -0.81 0.033 0.006 0.102 -0.16 -0.7 -0.07 -0.069 -0.089 0.031 -0.7 -1.34 (2.08)** 0.21 0.411 0.081 (1.77)* (2.24)** -1.06 0.296 0.754 0.323 -0.67 (1.7)* (2.47)** 0.497 0.325 -1.02 -1.31 0.249 0.168 -0.82 -0.61 -0.235 -0.275 -1.41 -0.8 0.625 0.102 0.313 -0.87 (1.99)** (3.23)*** 211 211 211 0.29 0.21 0.13 Robust t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T able 8 Regression o f Returns on Changes in R isk C ontrolling for Firm Characteristics: D rug Applications This table summarizes the coefficients estimated from the following regression equation: C A R = f (ASystematic _ Risk, Mdiosjncratic _ Risk, Firm Characteristics) Equation and variables are described on pages 86 and 87. ASystematic _ Risk Mdiosjncratic _ Risk Ln (Market equity) ROA _1 Cashl Leveragel Illiquidity Al phal 2 Change in ROA Change in Cash Change in Leverage Change in Illiquidity Constant Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Event Window (relative to event day t=0) "+2 to +60" "+2 to 120" "+2 to 180" 0.062 0.248 0.001 -0.02 -1.6 -0.58 -0.458 -0.083 -0.439 (2.38)** -0.71 (2.19)** -0.052 -0.009 -0.035 -0.7 (2.19)** (2.49)** 0.119 -0.006 -0.055 -1.21 -0.04 -0.29 -0.042 -0.152 0.063 -0.94 -0.8 -0.3 0.286 0.36 0.06 (2.12)** (2.24)** -0.61 -0.018 -0.041 -0.103 -0.99 -0.87 -1.61 0.47 0.21 0.331 (2.92)*** (3.29)*** (2.56)** 0.959 0.073 0.523 1.3 (1.74)* (2.45)** 0.04 -0.072 -0.21 -0.26 0.114 0.23 -0.55 -0.72 0.002 -0.065 -0.05 -1.21 0.531 0.07 0.349 (2.14)** (2.68)*** -0.59 273 273 273 0.2 0.24 0.06 Observations R-squared Robust t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 91 T able 9 R egression o f Returns on Changes in R isk Controlling for Firm Characteristics: Innovation-Targeted Acquisitions This table summarizes the coefficients estimated from the following regression equation: C A R = / (ASystematic _ Risk, Mdiosjncratic _ Risk, Firm Characteristics) Equation and variables are described on pages 86 and 87. A.Systematic _ Risk Mdiosjncratic _ Risk Ln (Market equity) ROA _1 Cashl Leveragel Illiquidity Alpha_12 Change in ROA Change in Cash Change in Leverage Change in Illiquidity Constant Observations R-squared Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Event Window (relative to event day t=0) "+2 to 120" "+2 to 180" "+2 to +60" 0.54 1.13 0.072 1.48 0.76 1.3 -0.144 0.06 -0.012 -0.04 -0.23 -0.33 -0.029 -0.01 -0.032 -1.23 -0.93 -0.53 -0.169 0.27 0.01 -0.67 (1.76)* -0.05 -0.208 -0.043 -0.228 -1.04 -0.39 -1.3 -0.014 -0.232 0.038 -0.07 -0.97 -0.23 0.144 -0.037 0.1 (3.17)*** -0.68 -0.12 0.078 -0.278 -0.283 -1.19 -1.04 -0.52 0.817 0.45 0.971 (1.71)* 1.28 (2.29)** -0.188 -0.455 -0.62 -1.37 -0.024 -0.221 -0.11 -0.86 0.024 -0.065 -0.09 -0.16 0.074 0.397 0.478 -0.43 -1.55 -1.51 112 112 112 0.15 0.31 0.39 Robust t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 Turning to the relation between cumulative abnormal returns and firm-specific variables, we can see that the cash flow variables seem to matter. Specifically, changes in future return on book assets (ROA) are positively related to the 60 and 120 day post event CARs. Thus, with controls for risk changes, firm size and other firm fundamentals, higher profitability growth is associated with higher post-announcement stock returns. Note that in none of the regressions the level and changes in book leverage and the Amihud’s illiquidity measure have significant coefficients. Firm size and the past 12 month abnormal returns are both significant in the 120 and 180 day post-event CAR regressions, indicating that smaller stocks and stocks with higher past twelve month abnormal returns have higher longer-term risk-adjusted returns. None of the other firm characteristics are significantly related to stock returns in any of the regression specifications. The data in Tables 8 and 9 present similar evidence for the drug application and innovation-targeted acquisition sub-samples. Specifically, in none of the regressions, changes in the systematic risk premium explain abnormal stock returns that follow the events. As for the idiosyncratic risk-retum relation, only in the 120 and 180 day post event CAR regressions for the drug application sub-sample we can observe a negative and statistically significant coefficient on the idiosyncratic risk change. However, the negative slope coefficient corresponding to idiosyncratic risk change is inconsistent with a predicted positive relation between changes in equity risk and stock price movements following the events. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 Similar to the sub-sample of patent applications, smaller stocks and stocks with higher past twelve month abnormal returns have higher longer-term risk-adjusted returns following the announcement of new drug applications. Finally, an increase in profitability (ROA) over the next two years is positively associated with the 120 and 180 day post-event CARs for both the acquiring and the drug sponsoring firms. In sum, my findings show no evidence that changes in the systematic risk of innovator firms’ equity following the event announcement can explain the mediumwindow post-event abnormal returns across all three categories of the events. While I find some evidence that idiosyncratic risk changes are associated with the post-event abnormal returns for the subsample of drug applications, the sign of the coefficient on idiosyncratic risk changes is inconsistent with the predicted one. This obviously does not allow one to make specific inference about the relation between change in idiosyncratic risk and the post-event return movements. Finally, in order to insure that the results reported here are robust to measurement of independent variables, I reestimate the regression using measures of equity risk computed over different time frames, ranging from 6 month around the event to 30 months around the event. The results (not reported) remain qualitatively the same. Returns. Risk Changes and Growth Options The regressions in the previous section give an overall picture of the economic significance of changes in risk characteristics and firm fundamentals in explaining abnormal returns that follow the innovation events. The final task is to examine the Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 94 prediction of the real options theory that the magnitude of event-induced changes in firms’ equity risk and stock returns is directly related to the economic properties of growth options embedded in the underlying events as well as the effects of competition and business cycle conditions. All else constant, the magnitude of changes in risk and returns are predicted to be positively related to the value of growth options and negatively related to the intensity of competition and the prevailing business conditions. To test these predictions, I estimate the following models for each category of events: C A R ^ = f (Forward Adjusted Patent Gtations, Market Share, Research Pipeline, Firm Characteristics, Term Premia, RF) M{iskMriil = f (Forward Adjusted Patent Gtations, Market Share, Research Pipeline, Firm Characteristics, Term Premia, RF) C AR^ —f (Radical Innovation or Financial Value, Market Share, Research Pipeline, Firm Characteristics, Term Premia, RF) ^ AR i s k ^ = / (Radical Innovation or Financial Value, Market Share, Research Pipeline, Firm Characteristics, Term Premia, RF) = J (Target Firm Product and Research Potential, Research Pipeline, Firm Characteristics, Term Premia, RF, Deal Characteristics) ARisk^ ,;^ —f T ( arget Firm Product and Research Potential, Research Pipeline, Firm Characteristics, Term Premia, RF, Deal Characteristics) Where CAR is the cumulative abnormal returns for the eleven day window surrounding the announcement date (-5, +5) and for 180 trading days (+2 through +180 days) following the announcement. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 95 ARisk is the overall change in the systematic and idiosyncratic equity risk. The change in the systematic risk is computed by multiplying the changes in the factor loadings with the average daily factor returns over my sample period. I assess a drug’s potential financial importance {Drug Value) using ex-ante equity analysts’ projections of the drug’s annual sales relative to the firm’s current period sales. Patent Citations variable measures the number of forward citations that a patent received. Product Potential variable measures the number of products in the target’s pipeline as the sum of launched products and advanced to the Phase 3 and Phase 2 trial stage. Products advanced to the Phase 3 and Phase 2 trial stage are adjusted using the probabilities that reflect the chance a potential drug has of receiving FDA approval. The relative value of a patent is estimated using the number of citations each patent receives in the subsequent years. Innovation Potential variable measures the citations weighed patents assigned to the target firm in the period preceding acquisition. I collect information on patents and their citations for public firms from the NBER patent database and for private firms from the U.S. Patent office website. The construction of all other explanatory variables is explained in Section 4.4. Note that competition, firm research pipeline and business cycle variables enter the equation as interactions with the value of the underlying growth option. By allowing the slope on the value of growth option to vary, the coefficients on interactive variables provide tests of whether the investors’ response to the innovation news depends on the intensity o f competition, the business cycle and the presence of other projects in the firms’ pipeline. I use clustered standard errors to account for correlation across Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. observations for a given firm and W hite’s (1980) correction to account for heteroscedasticity. Table 10 presents results for the patent application announcements, Table 11 presents results for the drug application announcements, and Table 12 presents results for the innovation-targeted acquisition announcements. In all three tables, Columns 1 and 2 present the regression results of the 11 day and 180 day cumulative abnormal returns on the value o f embedded growth options and other variables. As predicted, the results in all three tables show that the value of embedded growth options in all three innovative events is positively related to the magnitude of 11-day abnormal returns surrounding the announcement date (-5, +5), after controlling for all other factors. This finding confirms the hypothesis that investors put a positive price premium on the value of growth options across all three innovative events. In addition, consistent with the competition hypothesis, I find that the projected market share of the drug positively affects the pricing effect of the underlying growth option. Examining other variables, we can see that very few other variables impact the announcement returns and none systematically impacts the announcement returns across all three events. The presence of other projects in a firm’s pipeline, measured as the R&D intensity and the five-year stock of patents, also does not appear to significantly affect the slope of the relation between the value of embedded growth options and the announcement returns. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 97 T able 10 Regressions o f Returns and R isk Changes on Growth Options and Firm Characteristics: Patent Applications This table presents results of Equation (8). Variables are described on pages 94 and 95. Patent Citations R&D*Patent Citations Patentstock* Patent Citations Risk-free rate*Patent Citations Term Premia* Patent Citations Ln (Market equity) ROA _1 Cashl Leveragel Illiquidity Change in ROA Change in Cash Change in Leverage Change in Illiquidity Constant Beta t-stat Beta t-stat CAR______________Risk Changes "-5 to 5" "2 to 180" Systematic Idiosyncratic -0.006 0.009 -0.012 -0.001 (1.86)* -0.07 -1.15 -0.56 0.002 0.002 0 0.004 -0.1 -1.09 (1.73)* -1.19 Beta t-stat 0.001 -0.95 0 -0.01 0.002 -0.81 0.002 -1.29 Beta t-stat 0.002 -1.61 -0.001 -0.18 -0.002 -0.62 0 -0.2 Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat Beta t-stat 0 -0.16 0.004 -0.65 -0.027 -0.63 0.002 -0.07 -0.009 -0.18 -0.005 -1.53 0.141 (1.96)* 0 -0.06 -0.061 (2.12)** 0.012 -0.05 -0.097 -0.57 0.157 -0.7 -0.082 -0.83 0.019 (1.73)* 0.471 -1.33 0.146 -0.48 -0.275 -0.93 0.696 (2.99)*** 211 0.22 -0.001 -0.9 0.013 -0.9 0.074 -0.78 0.091 -0.94 -0.251 (2.15)** 0.115 (4.20)*** -0.414 (1.75)* 0.17 -1.12 0.047 -0.37 0.187 (2.66)*** -0.084 -0.61 211 0.23 -0.002 -1.55 0.011 -1.34 0.009 -0.15 0.098 (1.76)* -0.008 -0.1 0.026 -1.45 0.182 (1.68)* 0.007 -0.06 -0.022 -0.23 0.056 -1.16 -0.128 (1.68)* 211 0.13 -0.021 -0.36 211 0.09 Observations R-squared Robust t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 98 Table 11 Regressions o f Returns and R isk Changes on Growth Options and Firm Characteristics: Drug Applications This table presents results of Equation (8). Variables are described on pages 94 and 95. __________________________________ CAR_______________Risk Changes "-5 to 5” "2 to 180" Systematic Idiosyncratic Drug Value 0.01 0.00 Beta 0.01 0.02 -0.07 t-stat (3.26)*** (1.68)* -1.36 Market Share*Drug 0.00 Value Beta 0.01 0.01 -0.01 t-stat (3.30)*** -1.19 (2.46)** -0.68 R&D*Drug Value Beta 0.00 0.00 0.00 0.00 t-stat -1.27 (3.11)*** -0.61 -0.76 Patentstock* Drug Value Beta 0.00 0.00 0.00 0.00 (2 92)*** t-stat (2.11)** -0.62 -0.57 Risk-free rate*Drug Value Beta 0.00 0.00 0.00 0.00 t-stat -1.41 (2.64)*** -0.87 -0.46 Term Premia* Drug Value Beta 0.00 0.00 0.00 0.00 t-stat (3.24)*** -1.37 -1.10 -0.46 Ln (Market equity) -0.02 -0.01 Beta 0.00 -0.06 t-stat -0.05 (2.93)*** (2.22)** -0.72 ROA _1 Beta 0.02 -0.01 0.09 0.06 t-stat -0.34 -0.10 -0.44 -0.58 Cashl Beta 0.09 -0.01 0.01 -0.13 t-stat (1.98)** -0.09 (1.67)* -0.16 Leveragel Beta -0.04 -0.08 0.39 -0.03 t-stat (2.06)** (2.18)** -0.27 -0.30 Illiquidity Beta 0.00 -0.09 -0.07 0.14 t-stat -0.18 -1.12 -1.11 (8.63)*** Change in ROA Beta -0.14 1.07 0.03 0.22 t-stat (1.78)* (3.06)*** -0.20 (1.90)* Change in Cash Beta 0.12 0.19 0.10 t-stat -0.35 -0.98 -1.03 Change in Leverage Beta 0.11 0.08 0.10 t-stat -0.28 -0.55 -0.73 Change in Illiquidity Beta -0.16 0.04 0.17 t-stat (1.66)* (7.21)*** -0.51 Constant Beta 0.02 0.53 0.24 0.05 t-stat -0.43 (2.84)*** (2.37)** -0.70 Observations 273 273 273 273 R-squared 0.17 0.25 0.20 0.14 Robust t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 99 Table 12 Regressions o f Returns and R isk Changes on Growth Options and Firm Characteristics: Innovation-Targeted Acquisitions This table presents results of Equation (8). Variables are described on pages 94 and 95. Risk Changes CAR "-5 to 5" "2 to 180" Systematic Idiosyncratic 0.001 -0.002 0.001 Target_ InnovationPotential Beta 0.001 -0.81 -0.86 -1.13 t-stat (1.89)** 0.004 0 Product Potential Beta 0.007 0.002 -1.24 -0.02 t-stat (2.69)*** -0.15 0 0 0 R&D* Innovation Potential Beta 0 -0.92 -0.62 -1.25 t-stat -0.73 Patentstock* 0 0 InnovationPotential Beta 0 0 -1.41 -0.32 -1.25 t-stat -1.11 Risk-free rate* 0 0 InnovationPotential Beta 0 0 -1.6 -0.23 t-stat -0.08 -0.16 Term Premia* InnovationPotential 0 0 Beta 0 0.001 t-stat -1.21 -0.14 -0.88 -1.18 Ln (Market equity) Beta -0.011 -0.045 -0.013 0.03 -0.64 t-stat -1.02 -1.09 (2.23)** ROA _1 0.121 -0.016 -0.306 Beta -0.156 (2.97)*** t-stat (1.94)* -0.53 -0.13 Cashl 0.024 -0.347 0.155 Beta -0.63 (1.94)* t-stat -0.49 (2.47)** (2.76)*** 0.002 Leveragel Beta -0.11 0.128 0.06 -1.02 -0.94 -0.02 t-stat -0.33 Illiquidity -0.116 Beta -0.01 0.481 0.222 -0.79 (1.67)* -1.23 -0.72 t-stat Change in ROA Beta 1.01 0.191 0.01 0.413 -0.04 t-stat (4.12)*** -1.57 -0.47 Change in Cash 0.208 Beta -0.658 -0.158 t-stat -1.4 -0.66 -1.5 Change in Leverage Beta -0.186 -0.045 -0.035 -0.27 t-stat -0.56 -0.36 Change in Illiquidity Beta 0.542 0.174 -0.126 t-stat -1.47 -0.73 -0.6 Constant Beta 0.662 0.179 -0.267 0.06 (2.11)** t-stat -0.58 -1.56 -0.86 Observations 104 104 104 104 R-squared 0.22 0.14 0.2 0.15 Robust t statistics in parentheses * significant at 10%; ** significant at 5%; *** significant at 1% Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 Finally, only in the regression of the 5-day abnormal returns around new drug application announcements, I find an incrementally positive relation between changes in returns and the value of the growth option conditioning on level of the risk free rate and term premia, a proxy for business conditions. In all three tables, Column 3 presents the regressions of changes in the systematic risk on the value of embedded growth options and other explanatory variables, and Column 4 presents the same regressions using changes in the idiosyncratic risk as the dependent variable. The principal result here is that, contrary to the prediction, the value of embedded growth options does not appear to have any systematic relation to the changes in the systematic and idiosyncratic risk. This directly contradicts the real options prediction that firms with relatively more valuable growth options will experience a greater change in their equity risk measures. Finally, I reestimate the regressions using risk measures computed over different time frames, ranging from 6 month around the event to 30 months around the event. The results (not reported) remain qualitatively the same. Discussion My empirical results run counter to the predictions that were developed based on the real options models of Berk, Green and Naik (1999, 2004), Carlson, Fisher, and Giammarino (2004, 2006), and Gomes, Kogan and Zhang (2003). These studies argue that important investment events change the importance of growth options relative to existing assets and therefore raise or reduce overall equity risk and expected stock Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 101 returns of investing firms. I do find significant medium-horizon increase in stock returns following the announcements of an internal discovery of growth options (patents) as well as the exercise of these options (new drug applications). I also find that the value of embedded growth options is positively related to the stock market reaction to the event announcement. In contrast, I find that the market seems to penalize firms for an external acquisition of growth options, at least in the short-run. Taken together, these results suggest that the new information about growth options provides value relevant information to the stock market. However, the real options models considered in this paper suggest the changes in the firm loadings on the three Fama and French (1993) factors should capture much of the changes in firm risk and expected returns following the events. Therefore, I feel that the tests of equity risk changes around the event may provide more direct evidence about the validity of the real options arguments. However, I do not find evidence that the new information significantly affects firms’ equity risk or that the properties of growth options are systematically related to changes in the systematic or idiosyncratic equity risk of the innovator firms. O f course, the interpretation of my results, like the results for any asset-pricing test, is limited by my need to use a proxy for the unobservable model of expected returns. An econometrician will never know the true dynamics of firm-level risk premia because o f lack of information about the investors’ time varying information set. Therefore, I leave it to the reader to ultimately judge whether the results should be viewed as a rejection of the real options models or simply evidence that they do not Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. apply in my sample. At a minimum, my results suggest that recent real options model, while providing useful guidance about the firm-level asset price formation, are incomplete because they may not have enough explanatory power to alone address the deficiencies of basic models. In this regards, the results in this paper thus add to the large literature in finance that documents the empirical limitations of the existing asset pricing models, such as the CAPM and Fama and French (1993) model. As Lewellen and Nagel (2006), point out there is not enough time-variation in betas to explain anomalies, such as value premium. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 103 CHAPTER VI CONCLUSIONS This paper investigates how innovative investment events affect the dynamics of the equity risk and thus the expected stock returns of pharmaceutical and biotechnology firms. The pharmaceutical and biotechnology industries are an attractive setting because of the strong theoretical and empirical rationale for the news about innovative investment of firms in these industries to affect the properties of stock prices, as well as because of the large amount of value-relevant information about innovative investment. I directly test the predictions of real options theory that the economic properties of growth options embedded in innovative investment are main determinants of the asset pricing effects of innovative investment events. I find that the innovation events cause a significant change in stock prices at and following the event announcement. Additional tests indicate that the magnitude of the stock price reaction to the innovation news is positively related to the value of embedded growth options. I then examine the prediction derived from the real options models that the force driving these return movements is the event-induced change in the equity risk of innovator firms. Contrary to the prediction, however, I find that the events do not change the systematic equity of innovator firms. While the events appear to significantly affect the idiosyncratic risk, the results of the multivariate tests demonstrate that changes in neither systematic nor idiosyncratic risk can explain the stock return movements that follow the innovation events. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 104 Overall, my results indicate that investors revise firms’ stock prices in response to information about the value of firms’ growth options. However, the results cast doubt on the real options models (e.g. Berk, Green and Naik, 1999, 2004) that link predictability in the firm-level systematic risk and therefore expected returns to firm decisions related to the development and exercise of growth options. My results suggest that the real options theory while providing useful guidance about the empirical modeling of firm-level systematic risk may not have enough explanatory power to alone address the deficiencies of basic asset pricing models. The analysis raises several questions, some of which may provide the spark for future research in this area. First, to what extent are the findings of this paper representative of a larger universe of firms? Because the biotech and pharmaceutical industries are unique in some ways, my findings should be extrapolated to other industries with caution. Additional research, using data from other industries and/or contexts, will be valuable in exploring the generalizability of my results, especially since the importance of and the types of real options can vary across industries. Second, to what extent do limitations of real options models impact the results? I assume that the main source of the impact of innovative investment events on the risk and expected returns of firms is related to the nature of growth options embedded in these events. However, a large agency literature posits that as a result of conflicts of incentives among stakeholders firms undertake sub-optimal projects that change firm risk. For example, Jensen and Meckling (1976) argue that value-maximizing shareholders can expropriate wealth from existing debtholders by investing in new Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. projects that are riskier than those presently held in the firm’s portfolio. However, due to the high growth and dynamic nature of the pharmaceutical and biotechnology industries and given that most firms in my sample have very low debt ratios and mostly issue convertible debt, such alternative explanations do not seem to be relevant. However, in recent paper Guedj and Scharfstein (2005) argue that due to agency problems between managers and shareholders, some pharmaceutical and biotech companies are willing to develop marginally uneconomic drug projects. Incorporation of the effect of agency conflict and managerial compensation issues can potentially help refine real options theories to account for deviations from predictions of the current models. In the future, I intend to extend my study using Phamaprojects, a comprehensive database on pharmaceutical drug research and development project. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 106 APPENDIX A VALUE AND RISK OF THE INNOVATIVE PROJECT OVER TIME x(t) Value J 1U 15 Risk Prem Value n ■, Risk Prem Value Risk Prem 0 0 .7 2 ( 0. 03) 20.00 ( 2 0. 0 0 ) 4.13 ( 0. 16 ) 14.19 (20.00) 9.22 ( 0. 45 ) 11.91 (20.00) J umber of Complete* j Stages ( y(t) = £) 4 16 20 12 8 88.74 5.34 0.93 20.91 47.5 (88.74) ( 0. 68) ( 5 .3 0) (21.56) (48.3) 11.37 8.87 8. 20.00 17.07 (12.04) (20.00) ( 20 . 0 0 ) (9.08) (8.) 177.5 5.34 52.44 21.48 101.3 ( 3 .8 7) (22.14) (54.18) (177.5) (103.1) 8.41 15.88 11.98 9.43 8. ( 20 . 0 0 ) (9.65) (8.51) (8.) (12.98) 12.72 38.74 155.2 266.2 84.03 (266.2) (10.46) (40.24) (86.85) (157.9) 8.27 13.29 10.4 8.9 8. (10.84) (18.03) (9.03) (8.33) (8.) The cells in the table show the value and risk premium (in percent) of a venture that requires 20 stages to complete for a given X(t), a cash flow per period and the time spent in development is half the number of completed stages so that _(n; y) = 2:0. The numbers in parentheses give the value and risk premium when the probability of success is known. The risk premium is the instantaneous expected return minus the risk free rate and is annualized. Source: Berk, Green and Naik (2004), page 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 107 APPENDIX B SAMPLE NEWS REPORTS Source: PR Newswire (Available on Lexis-Nexus) July 24, 1998 Friday DAILY EDITIONS HEADLINE: Investors optimistic on Forest's antidepressant It's no wonder the shares of Forest Laboratories Inc. have been soaring. Celexa, its new blockbuster competitor to the anti-depressant drug Prozac, will likely grab a significant chunk of a runaway market. Since the start of the year Forest shares (FRX/AMEX) have climbed 55%, largely on expectation that the New York-based company could successfully bring Celexa to market. On Monday, the shares jumped to a 52-week high of US$ 39 3/8 after the U.S. Food and Drug Administration gave Forest the green light to start selling Celexa. Yesterday, they fell 11/16 to close at US$ 37 3/8. Depression is one of the most common mental disorders, yet the market is dominated by only a handful of drugs, including Eli Lilly & Co.'s Prozac, the world's top selling antidepressant, Pfizer Inc.'s Zoloft and SmithKline Beecham PLC's Paxil. James Flynn, an analyst at ING Baring Furman Selz LLC, says each of those products has sales in excess of US$ 1 billion. 'This is an enormous category and it's growing at a double-digit pace,' he adds. Analysts predict U.S. sales of antidepressants will rise from US$ 5 billion today to US$ 7 billion annually three years from now. One of the problems with many of the drugs now on the market is that they interact with other commonly prescribed therapies, says Flynn. When taken in combination with drugs for cancer, asthma and heart disease, many antidepressants produce side effects. In fact, since many people suffering from depression are frequently afflicted with those ailments as well, drug interaction is a serious issue. Forest Labs says Celexa has no known interactions with other common prescription drugs. Observers say that gives the drug a good shot at grabbing market share from the incumbents. 'We expect Celexa will gain as much as a 10% share of this lucrative market over the next five years,' wrote Steven Gerber, an analyst at CIBC Oppenheimer, in a report released the same day Celexa received FDA approval. Based on that estimate, Gerber predicts that sales of Celexa will hit US$ 500 million within the next three to four years, effectively doubling Forest's revenues. With the FDA approval for Celexa safely under its belt, Forest hopes to launch its marketing program for the drug by the end of September. 'There's nothing to worry about now,' Flynn says. 'They still have to get the FDA's approval on their marketing plan, but that shouldn't be an issue.' CIBC Oppenheimer's Steven Gerber expects Celexa to grab as much as 10% of the U.S. market over the next five years. He predicts that the new drug could have sales Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 108 approaching US$ 500 million by the middle of 2002. If that comes to pass, Celexa will become Forest's top selling product, accounting for half its revenue. Its largest revenue generators have been asthma medicine Aerobid, with 1997 sales of US$ 57 million; pain-reliever Lorcet, with sales of US$ 27 million; and hypertension drug Tiazac, at US$ 25 million. Most analysts' earnings estimates take a conservative view of the sales potential of other drugs in Forest's development pipeline. Although none is likely to knock Celexa from its perch, they could add significantly to Forest's revenue. They include the infant respiratory medication Infasurf, expected to go on sale in 1999, and Alzheimer's drug Synapton. July 01, 2001 HEADLINE: Merck Buys Gene Research Company Merck & Co. Inc. took another step away from its reliance on in-house research in May when it bought biotech research company Rosetta Inpharmatics Inc. in a stock swap worth about $620 million. "The acquisition of Rosetta Inpharmatics fits Merck's objective of expanding our lead in cutting-edge science by continuing to enhance our internal research capabilities," said Anthony Ford-Hutchinson, EVP for worldwide basic research at Merck Research Laboratories. Rosetta, based in Kirkland, Wash., specializes in informational genomics, a discipline that combines informatics and genomics to speed up drug discovery. Through informatics, companies use software to help them decipher data recently made available by the Human Genome Project and other sources of raw data about genes. Rosetta uses microarray technology that analyzes thousands of genes in an attempt to pinpoint the ones that cause disease. The Merck-Rosetta pact comes shortly after the acquisition of Aurora Biosciences Corp. by Vertex Pharmaceuticals Inc. Aurora makes technology that screens chemical compounds to discover potential drugs. But these deals are recent highlights of what remains a small, fragmented industry. The 75 U.S. publicly traded genomics companies have a combined value of about $75 billion. Gene Logic Inc., a Gaithersburg, Md.-based seller of information about genes, is one of the larger genomics companies that could be on the sale block. "If Rosetta is worth $620 million to Merck, Gene Logic could be worth just as much," says Winton Gibbons, a pharmaceuticals analyst at William Blair & Co. Rosetta is expected to help Merck discover new drugs and get them to market more quickly. "Rosetta will add value for Merck's early-stage development process. Merck has shown more willingness recently to look outside of its own laboratories to enhance its drug discovery pipeline," says Bob Kirby, a pharmaceuticals analyst at Edward Jones, a St. Louis-based investment bank. He cited Merck's acquisition of Medco in the late 1990s as one of the company's first efforts at growth through m&a. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 109 BIBLIOGRAPHY Abel, A. B., A. K. Dixit, J. C. Eberly, and R. S. Pindyck, 1996, Options, the value of capital, and investment, Quarterly Journal o f Economics 111, 753-777. Aguerrevere, Felipe, 2006, Real options, product market competition and asset returns, Working Paper, University of Calgary. Amihud, Y., 2002, Illiquidity and stock returns: Cross-section and time-series effects, Journal o f Financial Markets 5, 31-56. Anderson, C. W. and L. Garcia-Feijoo, 2006, Empirical evidence on capital investment, growth options, and security returns, Journal o f Finance 61, 171-194. Andersen, T., T. Bollerslev, F. Diebold, and J. Wu, 2005, Realized beta: Persistence and predictability, in T. Fomby and D. Terrell, ed: Advances in Econometrics: Econometric Analysis o f Economic and Financial Time Series in Honor o f R.F. Engle and C. W.J. Granger^Elsevier, New York). Austin, D., 1993, An event-study approach to measuring innovative output: The case of biotechnology, American Economic Review 83, 253-258. Berk, J. B., R. C. Green, and V. Naik, 1999, Optimal investment, growth options and security returns, Journal o f Finance 54, 1153-1607. Berk, J. B., R. C. Green, and V. Naik, 2004, Valuation and return dynamics of new ventures, Review o f Financial Studies 17, 1-35. Boehme, R., and S. M. Sorescu, 2002, The long-run stock performance following dividend initiations and resumptions: underreaction or product of chance?, Journal o f Finance 57, 871-900. Brown, S. J. and J. B. Warner, 1985, Using daily stock returns: The case of event studies, Journal o f Financial Economics 14, 3-31. Campbell, J., 1991, A variance decomposition for stock returns, Economic Journal 101, 157-179. Campbell, J. Y., M. Lettau, B. G. Malkiel, and Y. Xu, 2001, Have individual stocks become more volatile? An empirical exploration of idiosyncratic risk, Journal o f Finance 56, 1-43. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 110 Carlson, M., A. Fisher, and R. Giammarino, 2004, Corporate investment and asset price dynamics: Implications for the cross-section of returns, Journal o f Finance 59, 2577-2603. Carlson, M., A. Fisher, and R. Giammarino, 2006, Corporate investment and asset price dynamics: Implications for SEO event studies and long-run performance, Journal o f Finance 61, 1009-1034. Carlson, M., A. Fisher, and R. Giammarino, 2005, SEO's, real options, and risk dynamics: empirical evidence, Working Paper, University of British Columbia. Chambers, D., R. Jennings, and R. Thompson II, 2002, Excess returns to R&Dintensive firms, Review o f Accounting Studies 7, 133-158. Chan, L. K.C., J. Lakonishok, and T. Sougiannis, 2001, The stock market valuation of research and development expenditures, Journal o f Finance 56, 2431-2456. Chen, S., T.J. Shevlin, Y.H. Tong, 2007, Does the pricing of financial reporting quality change around dividend changes?, Journal o f Accounting Research 45, 1-40. Clayton, M. C., J. C. Hartzell, and J. Rosenberg, 2005, The Impact of CEO turnover on equity volatility, Journal o f Business 78, 1779-1808. Cockbum, I., Henderson, R., 1996, Scale, scope, and spillovers: The determinants of research productivity in drug discovery, Rand Journal o f Economics 27, 32-60. Cochrane, J. H., 1991, Production-based asset pricing and the link between stock returns and economic fluctuations, Journal o f Finance 46, 1, 209-237. Cochrane, J., 2001. Asset Pricing (Princeton University Press, Princeton, NJ). Cooper, I., 2006, Asset pricing implications of non-convex costs of adjustment, Journal o f Finance 61, 139-170. Cornell, B., 1999, Risk, duration, and capital budgeting: new evidence on some old questions, Journal o f Business 72, 183-200. Daniel, K. and S. Titman, 2006, Market reaction to tangible and intangible information, Journal o f Finance 61, 1605-1643. Danzon, P., A. Epstein, and S. Nicholson, 2004, Mergers and acquisitions in the pharmaceutical and biotech industries, Working Paper, University of Pennsylvania. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Ill DiMasi, J.A. 2001. New drug development in U.S. 1963-1999, Clinical Pharmacology & Therapeutics 69, 286-96. DiMasi, J.A., H.G. Grabowski and J. Vernon, 1995, R&D costs, innovative output and firm size in the pharmaceutical industry, International Journal o f the Economics o f Business 22, 201-219. Dimson, E., 1979, Risk measurement when shares are subject to infrequent trading, Journal o f Financial Economics 7, 197-226. Dixit, A. K. and R. S. Pindyck, 1994. Investment under uncertainty (Princeton University Press, Princeton, NJ). Eberhart, A., W. Maxwell, and A. Siddique, 2004, An examination of long-term abnormal stock returns and operating performance following R&D increases, Journal o f Finance 59, 623-650. Fama, E. F., 1998, Market efficiency, long-term returns and behavioral finance, Journal o f Financial Economics 49, 283-306. Fama, E. F., and K. R. French, 1992, The cross-section of expected stock returns, Journal o f Finance 47, 427-465. Fama, E. F., and K. R. French, 1993, Common risk factors in the returns on stocks and bonds, Journal o f Financial Economics 33, 3-56. Fama E., and K. French, 1997, Industry costs of equity, Journal o f Financial Economics 43, 153-93. French, K., Schwert, G.W., Stambaugh, R., 1987, Expected stock returns and volatility, Journal o f Financial Economics 19, 3-29. Garlappi, L., 2004, Preemption risk and the valuation of R&D ventures, Journal o f Financial and Quantitative Analysis 39, 843-872. Gomes, J., Kogan, L., and L. Zhang, 2003, Equilibrium cross-section of returns, Journal o f Political Economy 111, 693-732. Grenadier, S., 2002, Option exercise games: An application to the equilibrium Investment strategies of firms, Review o f Financial Studies 15, 691 - 721. Griliches, Zvi, 1981, Market value, R&D, and patents, Economic Letter 7, 83-187. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 112 Guedj, I., 2005, Ownership vs. contract: how vertical integration affects investment decisions in pharmaceutical R&D, Working Paper, University of Texas. Guedj, I., and D. Scharfstein, 2005, Organizational scope and investment: Evidence from the drug development strategies and performance of biopharmaceutical firms, Working Paper, NBER and Harvard University. Hall, B., Jaffe, A. and Trajtenberg, M., 2001, The NBER Patent citations data file: lessons, insights and methodological tools, NBER Working Paper # 8498. Hall, B., Jaffe, A. and Trajtenberg, M., 2005, Market value and patent citations, RAND Journal o f Economics 36, 16-38. Hamada, Robert S., 1972, The effect of the firm's capital structure on the systematic risk of common stocks, Journal o f Finance 27, 435-452. Higgins, M.J., and D. Rodriguez, 2006, The outsourcing of R&D through acquisitions in the pharmaceutical industry, Journal o f Financial Economics 80, 351-383. Huberman, G., and T. Regev, 2001, Contagious speculation and a cure for cancer: a nonevent that made stock prices soar, Journal o f Finance 61, 387-396. Kogan, L., 2004, Asset prices and real investment, Journal o f Financial Economics 73, 411-431. Krieger, D., Ruback, R.S., 2001, Merck & company: evaluating a drug licensing opportunity, Harvard Business School Case #9-201-023. Lamont, O., 2000, Investment plans and stock returns, Journal o f Finance 55, 27192745. Lemer, J., Shane, H., Tsai, A., 2003, Do equity financing cycles matter? Evidence from biotechnology alliances, Journal o f Financial Economics 67, 411-446. Lewellen, J., and S. N agel, 2006, The Conditional CAPM does not explain assetpricing anomalies, Journal o f Financial Economics 82, 289-314. Litzenberger, R., and K. Ramaswamy, 1979, The effect of personal taxes and dividends on capital asset prices: theory and empirical evidence, Journal o f Financial Economics 7, 163-196. Mayers, D., 1998, Why firms issue convertible bonds: the matching of financial and real investment options, Journal o f Financial Economics 47, 83- 102. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 113 Merton, R.C., 1973, An Intertemporal Capital Asset Pricing Model, Econometrica 41, 867-887. Merton, R. C., 1987, A Simple model of capital market equilibrium with incomplete information, Journal o f Finance 42, 483-510. O’Hara, M., 2003, Liquidity and price discovery, Journal o f Finance 58,1335-1354. Myers, S., and C. Howe, 1997, A Life-cycle model of pharmaceutical R&D, Technical Report, Program on the Pharmaceutical Industry, Massachusetts Institute of Technology. Nicholson, S., P. M. Dranzon, and J. Mccullough, 2005, Biotech-pharmaceutical alliances as a signal of asset and firm quality, Journal o f Business 78, 14331464. Pakes, Ariel, 1985, On Patents, R&D, and the stock market rate of return, Journal o f Political Economy 93, 390-409. Pastor, L., and P. Veronesi, 2003, Stock valuation and learning about profitability, Journal o f Finance 58,1749-1789. Pastor, L., and P. Veronesi, 2006, Was there aNasdaq bubble in the late 1990s?, Journal o f Financial Economics 81, 61 -100. Pastor, L., and R. F. Stambaugh, 2003, Liquidity risk and expected stock returns, Journal o f Political Economy 111, 642-685. Petkova, R., 2006, Do the Fama-French factors proxy for innovations in predictive variables?, Journal o f Finance 61, 581-612 Pharmaceutical Research and Manufacturers of America, 2005, Industry Profile 2005, http://www.pharma.org. Pindyck, R. S., 1984, Risk, inflation, and the stock market, American Economic Review 74,334-351. Pisano, G. P., 2002, Pharmaceutical Biotechnology, in Benn Steil, David G. Victor and Richard R. Nelson, ed: Technological Innovation & Economic Performance (Princeton University Press, Princeton, NJ). Schwartz, E., 2004, Patents and R&D as real options, Working paper, UCLA. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 114 Schwert, G. W., 1989, Why does stock market volatility change over time?, Journal o f Finance 44, 1115-1153. Shanken, J., 1990, Intertemporal asset pricing: An empirical investigation, Journal o f Econometrics 45, 99-120. Spiegel, M., and X. Wang, 2005, Cross-sectional variation in stock returns: Liquidity and idiosyncratic risk, Working Paper, Yale University. Titman, S., K.C. Wei and F. Xie, 2004, Capital investments and stock returns, Journal o f Financial and Quantitative Analysis 39, 677-700. Xing, Y., 2005, Interpreting the value effect through the Q-theory: An empirical investigation, Working paper, Rice University. Zhang, L., 2005, The value premium, Journal o f Finance 60, 67-103. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. [...]... gives the value of the project and the second row gives the risk premium of the project (annualized expected return minus risk- free rate) The numbers in parentheses show the value and risk of the project without the option to suspend or abandon the project As it is evident from this table, holding the expected cash flows constant, the value of the project increases and its risk premium decreases as stages... leverage makes the option riskier than the underlying asset The risk premium of the project, therefore, is directly related to the expected future payments or the strike price of the option as well as to the value of the option For example, completion of a stage of the project reduces the future fixed payments and leads to an increase the value of the option The net result will be a decrease in the risk premium... implications of real options theory in a variety of industries, however, I focus on innovative investments of firms in the pharmaceutical and biotechnology industries While my focus on firms within the same industry restricts the generality of results, it has four advantages First, the pharmaceutical and biotechnology industries are one of the most research-intensive industries in the U.S and real options represent... fashion Suppose that the firm completed the project and is ready to commercialize the product In the parlance of the option literature, this is equivalent to exercising a growth option At this stage, the value of the project depends only on the cash flows to be generated from the project and the risk premium of the project simply equals the systematic risk of the underlying cash flows The intuition here... Therefore, the option to suspend the project is particularly valuable to the firm in early stages of the project and helps to reduce the overall risk of the project In general, as future cash flows increase, the value of the option to suspend decreases because it is less likely that the firm will suspend the project To illustrate the potential magnitude of changes in the value and risk premium of the. .. different from that before the event To measure changes in the systematic risk and idiosyncratic equity risks of innovator firms, I estimate the Fama and French (1993) three-factor model at the firm-level The Fama and French model suits the purpose of my study particularly well because, according to Berk, Green and Naik (1999) and Gomes, Kogan and Zhang (2003), the size and book-to-market factors in the. .. completion of the project no further investment decision is needed, and therefore the growth option has the same characteristics as the underlying cash flow This will not be the case, however, during the period of development of the project Prior to the completion, the risk premium of the project is higher than that of the underlying asset Reproduced with permission of the copyright owner Further reproduction... firm’s equity Despite the theoretical development of real options theory, little empirical evidence documents the impact of real options on firms’ equity risk and expected return Existing studies, which have examined some of the implications of real options theory, lack necessary firm-level data and are able to provide only indirect evidence For example, Anderson and Garcia-Feijoo (2006) and Xing (2006)... Where Risk =the risk premia of firm assets AIP= assets in place GO=growth options These equations show that events that contain new information about the value and risk of growth options embedded in the project ought to have an important effect on the risk premium and therefore expected returns of the firm Note that the additive form of the equations is strictly appropriate only when the values of the. .. of the project The project involves several separate stages of development, and at each stage, the managerial decision of whether to continue the investment depends on the resolution of systematic and idiosyncratic uncertainty Therefore, an investment project is viewed as a compound real option on systematic uncertainty The models shows that the systematic risk, and thus the expected return, of the