Investigation of optimal valuation methods in the pharmaceutical industry

Master thesis in Finance and International Business Investigation of optimal valuation methods in the pharmaceutical industry Authors: Morten Vester & Andreas Holmgaard Petersen Characters: 185.598 Supervisor: Özlem Dursun-de Neef Aarhus University School of Business and Social Sciences Department of Economics and Business September 1 st 2015

Abstract This thesis has the purpose of determining the optimal valuation method for pharmaceutical drug development projects. A literature review of valuation methods used by practitioners shows that more simple models, such as DCF valuation, are used, regardless of the industry context. The motivation in this thesis is therefore to investigate the potential of using more advanced models in the valuation of drug development projects. More specifically, one of the objectives is to investigate the potential of using Real Option theory. In order to fulfil the purpose of the thesis, it consists of both a theoretical- and practical research approach. The theoretical part is an analysis and discussion of four valuation models; Discounted Cash Flow, Decision Tree Analysis, Binomial-, and Quadranomial Real Option approach. In order to investigate theoretical differences in each of the valuation models, four evaluation criteria are constructed; Concept, Uncertainty, Strategy Flexibility, and Usability. Each of the valuation models is then evaluated upon these criteria. The practical research approach is a case study on a phase I pharmaceutical diabetes development project. The thesis finds that the pharmaceutical industry is characterised by sequential dependent clinical phases, which are uncertain in time-length, success, and cost. An optimal valuation method must thus be able to comprehend and incorporate the uncertainty of each clinical phase. We find that an optimal valuation model must be able to separate market and technological uncertainty to provide strategic flexibility. From the evaluation of each criteria and the practical implementation of the case study, the thesis finds that both the DTA and the Quadranomial Real Options model have high potential. Both models can explicitly incorporate future events i.e. the clinical phase success rates. The thesis further finds that the Real Option approach in general has a good potential in the pharmaceutical industry, but is challenged by the more complex practical implementation. The thesis reaches the conclusion that Decision Tree Analysis is the best tool, when valuating pharmaceutical drug development projects, from an external point of view. Since the findings of this thesis are based on partly a practical research approach, it is likely that we influence the result because of our pre-determined expectations.

Table of Contents 1 Introduction --------------------------------------------------------------------------------------------------------------------- 1 1.1 Methods of science ------------------------------------------------------------------------------------------------------- 3 1.2 Delimitations -------------------------------------------------------------------------------------------------------------- 5 2 Introduction to the pharmaceutical industry ------------------------------------------------------------------------- 7 2.1 Definition ------------------------------------------------------------------------------------------------------------------- 7 2.2 Size and sales ------------------------------------------------------------------------------------------------------------- 8 2.3 Research & development ----------------------------------------------------------------------------------------------- 8 2.4 Patents ---------------------------------------------------------------------------------------------------------------------- 9 2.5 Drug development ----------------------------------------------------------------------------------------------------- 10 2.6 Success rates, cost, and time ---------------------------------------------------------------------------------------- 12 2.7 Sum-up -------------------------------------------------------------------------------------------------------------------- 18 3 Valuation methods in the pharmaceutical industry -------------------------------------------------------------- 19 3.1 The Criteria -------------------------------------------------------------------------------------------------------------- 20 3.1.1 Concept -------------------------------------------------------------------------------------------------------------- 21 3.1.2 Uncertainty -------------------------------------------------------------------------------------------------------- 21 3.1.3 Strategic flexibility ---------------------------------------------------------------------------------------------- 21 3.1.4 Usability ------------------------------------------------------------------------------------------------------------ 22 3.2 Standard DCF model -------------------------------------------------------------------------------------------------- 23 3.3 Decision Tree Analysis ----------------------------------------------------------------------------------------------- 27 3.4 Real Options ------------------------------------------------------------------------------------------------------------- 31 4 Case study - practical implementation -------------------------------------------------------------------------------- 49 4.1 Practical implementation of standard DCF model ----------------------------------------------------------- 52 4.2 Practical implementation of Decision Tree Analysis -------------------------------------------------------- 63 4.3 Practical implementation of the Binomial Lattice approach ---------------------------------------------- 68 4.4 Practical implementation of the Quadranomial Lattice approach -------------------------------------- 73 5 Evaluation & recommendation ------------------------------------------------------------------------------------------ 76 5.1 Evaluation of the DCF model --------------------------------------------------------------------------------------- 76 5.2 Evaluation of Decision Tree Analysis ---------------------------------------------------------------------------- 78 5.3 Evaluation of the Binomial and Quadranomial approach ------------------------------------------------- 79 5.4 DTA vs. ROA ------------------------------------------------------------------------------------------------------------ 81 5.5 Future research --------------------------------------------------------------------------------------------------------- 84 6 Conclusion --------------------------------------------------------------------------------------------------------------------- 85 7 References --------------------------------------------------------------------------------------------------------------------- 88

List of figures Figure 1.1 Chapter structure of thesis Figure 2.1 The drug development process Figure 3.1 Criteria for evaluation of valuation methods Figure 3.2 Standard discounted cash flow model Figure 3.3 Cost of equity Figure 3.4 Summary of DCF Figure 3.5 Example of a decision tree Figure 3.6 Summary of DTA Figure 3.7 The intrinsic value of options Figure 3.8 Payoff positions Figure 3.9 Recombining binomial tree Figure 3.10 Binomial risk-neutral formula Figure 3.11 Option value formula in a binomial tree Figure 3.12 Option value tree for a two period call option Figure 3.13 Brownian Motion Figure 3.14 Possibilities in quadranomial tree Figure 3.15 Quadranomial formula Figure 3.16 Direct and indirect volatility estimation methods Figure 3.17 Logarithmic present value approach Figure 3.18 Market and technological uncertainty Figure 3.19 Summary of Real Option Figure 4.1 Illustration of the development and commercialisation phase Figure 4.2 Practical steps in DCF valuation Figure 4.3 Cost of equity for case project Figure 4.4 Sensitivity analysis of input in cost of equity Figure 4.5 Sensitivity analysis of various input Figure 4.6 Overview of steps in DTA Figure 4.7 DTA calculations Figure 4.8 Sensitivity analysis of the components in the discount rate Figure 4.9 Sensitivity analysis of the technological success rates Figure 4.10 Sensitivity analysis of the research and development cost Figure 4.11 Practical implementation of the binomial lattice approach Figure 4.12 Asset tree in the binomial model Figure 4.13 Binomial option value tree Figure 4.14 Sensitivity analysis of volatility and risk free rate Figure 4.15 Steps in the quadranomial approach Figure 4.16 Quadranomial option value Figure 5.1 DTA versus ROA

List of tables Table 2.1 Clinical success rates Table 2.2 Cost of drug development Table 2.3 Clinical time length Table 3.1 Difference between financial and real options Table 3.2 Overview of selected types of real options Table 4.1 Development cost assumed for NN9927 Table 4.2 Cost assumptions in the commercialisation phase Table 4.3 Free Cash Flow of project NN9927 Table 4.4 Overview of Beta estimation Table 4.5 Calculation of DCF value Table 4.6 Probabilities used for in each of the decision nodes Table 5.1 Evaluation of the investigated models

1 Introduction Pharmaceutical drug development is a rather cumbersome and costly affair. Cost estimates on the development of a new approved drug in the United States go as high as exceeding US$ 2 billion (DiMasi, Grabowski, & Hansen, 2014). This combined with the substantial time effort required to develop a new successful drug and the relatively low probability of making a successful blockbuster-drug are well describing characteristics of the highly uncertain business environment pharmaceutical companies operate within. The drug development process is composed of several stages, in which the drug company gathers evidence to satisfy government regulations that it can consistently manufacture a safe and efficacious form of the compound for the medical condition it is intended to treat. At the end of each development stage, the company uses the technological and market information revealed up to that point to decide whether to abandon or continue development of the compound (Kellogg & Charnes, 2000). This high uncertainty calls for considering financial decision tools that can expand the notion of a static environment and enable the possibility of evaluating the decisions management faces as time progresses. To address these concerns, real option valuation models have been suggested as a suitable way to include uncertainty in investment decisions (Myers, 1984), and this is where this thesis has its starting point. Both recent research and previous studies have shown that certain valuation methods are more commonly used by practitioners than others. Particularly Discounted Cash Flow methods and Relative Valuation are widely and intensively used by practitioners in valuation, while Real Option valuation is hardly ever used (Bancel & Mittoo, 2014; Block, 2007; Hartmann & Hassan, 2006; C. V. Petersen & Plenborg, 2012). In their articles (Bancel & Mittoo, 2014; Demirakos, Strong, & Walker, 2004) more investigate what kind of theoretical financial models financial analysts use in terms of valuation in different industries. Their findings show that the only industry, in which a Real Option valuation approach is applied, is the pharmaceutical industry. About 10 per cent of the participating analysts answered that they use or have used a Real Option valuation approach to pharmaceutical project valuation. The study showed that often not a single method is used but several methods are combined to reach conclusions and recommendations on valuation. Given the characteristics of the pharmaceutical business environment and the fact that practitioners have preference towards simpler and more static valuation methods, the Page 1 of 92

objective of this thesis is to investigate the differences between fundamental cash flow valuation- and Real Options Valuation (ROV) models both in theory and in practical use. Another objective with this thesis is to evaluate the potential of using real option theory in valuation of pharmaceutical development projects, when having an external point of view. The focus of the thesis is to evaluate the relative differences between the valuation models, and not to analyse any mispricing. The focus is on the theory behind the different valuation models which assumption they are based on, and how they can be implemented in practice, when having the characteristics of the pharmaceutical industry in mind. Based on the above and the authors curiosity the following research questions are examined. What are the main theoretical differences between fundamental valuation models and real option valuation and their underlying assumptions with focus on the pharmaceutical industry? How do the different models differ in regards to practical implementation, considering their ease of use and strategic opportunities? What is the potential, with the case study and theoretical framework in mind, of using the Real Option valuation in the pharmaceutical industry? In order to answer the problem statement we have structured the thesis as seen in figure 1.1 below. The thesis will include both a theoretical and a practical research approach. The theoretical approach will consist of a review and evaluation of different valuation methods, having the pharmaceutical industry in mind. The theoretical evaluation will be based upon four defined criteria, which are used to analyse the theoretical differences and help structure a more in depth theoretical comparison of each valuation model. Each criterion is further used to assure that the underlying assumptions of each valuation model are discussed in relation to our research questions. In order to exemplify the discussed theory and differences in approach, the thesis includes a case study of a Novo Nordisk pipeline project. The case study is a practical approach, which will enable and ease the discussion of the different valuation approaches. Before a more in depth evaluation of each valuation approach, it is important to understand both the theoretical and practical differences between each valuation model. Page 2 of 92

Figure 1.1: Chapter structure of thesis Introduction Introduction Research questions Structure of thesis Methods of science Delimitation Introduction to industry Industry definition Size, R&D, patens Development phases Succes, costs and time data Criteria and different valuation methods Concept Uncertainty Strategic flexibility Usability DCF DTA Binomial Quadranomial Case study of pharmaceutical project NN9927 Project NN9927 Development Commercialisation DCF DTA Binomial Quadranomial Evaluation of theory and practical implementation Evaluation of DCF, DTA, ROA Quadranomial vs.dta Tradable vs. risk Conclusion Conclusion Future research Source: Own creation As seen in figure 1.1, this chapter is followed by an introduction to the pharmaceutical industry before the focus is on the evaluation of different valuation methods. In order to investigate the implementation of the valuation methods, a case study is conducted next, ultimately making it possible to answer the research questions proposed and making a conclusion. A more detailed explanation of, how the chapters are structured, will appear in the beginning of each chapter. 1.1 Methods of science To improve the arguments and choices made throughout this thesis, the following section is an explanation of our research approach and method of science. The purpose of this section is to improve the general understanding of our assumptions and clarify the limitations of our general research approach. Economic science is in general build on the basic of the positivistic paradigm 1, and financial theory is grown upon the concepts of this philosophical position. The positivistic paradigm is based on empiricism, the idea that observations and measurements are the essence of scientific endeavour, and that research produces facts that correspond to an independent reality (Eriksson & Kovalainen, 2011). The positivistic paradigm is based on strong 1 Positivism is a term coined by Auguste Comte (1898-1857), refers to an assumption that only legitimate knowledge can be found from experience (Eriksson & Kovalainen, 2011). Page 3 of 92

assumptions about empirical requirements and theoretical independence 2, which are often subject for criticism (Holm, 2011). These assumptions are difficult to satisfy in reality, and the assumptions may be violated to some extent. Based on our theoretical framework, chosen paradigm, and assumptions, it is likely that we influence our empirical data to verify our preexpectations. Our following results may therefore be biased by our own influence and should be interpreted in the context of the stated limitations and data input. The research approach, used in this thesis, is based on both the inductive and deductive research approach. Deduction is a form of reasoning, in which particular conclusions are formulated from general premises, and the inductive research logic concludes from the particular (Eriksson & Kovalainen, 2011). Both research logics are used in different parts of the thesis, and must be considered as combined through the research process. A research study based on both the inductive and deductive method is defined as the abduction logic method 3. The abduction research logic is used, when the research process consists of various forms of reasoning and logic exploratory (Eriksson & Kovalainen, 2011). Our research approach is using the deductive logic method in the answer of the first problem questions regarding the theoretical differences between the valuation methods. We use general financial theory to evaluate the specifics of the pharmaceutical industry and the stated criteria. The inductive method is primarily used in the case study of development project NN9927, in which we seek to evaluate upon the theory and the theoretical use of each valuation model. The financial framework used in the thesis is based on important assumptions about market efficiency and human rational behaviour. In the context of the positivistic approach, our study should be based on valuation models that aim to reach the assumptions of the positivistic requirements regarding valid empirical data and mathematical arguments. Our case study is based on the development of a single pharmaceutical product. The case study is built of both project specific knowledge and a large amount of industry data in order to answer our research question in the best way possible. In relation to our research questions, we focus on the theoretical differences and potential of each valuation method. The case study 2 Theoretical independence requires that observations must be unbiased (Holm, 2011). 3 The abduction research logic mentioned by Charles Sanders Peirce can be considered as the logic of exploratory data analysis (Eriksson & Kovalainen, 2011). Page 4 of 92

design is thereby inspired by the extensive case study approach 4. This design emphasise the theoretical differences between the different valuations models, rather than analysing the specific case project. As mentioned, it is likely that we influence our findings, which also influence the question if the case study can be generalized to similar studies. Based on our input data, we believe our findings can be generalised and used for other pharmaceutical project, since our data consists of mostly generalisable industry data. 1.2 Delimitations Before proceeding, we find it important to state the delimitations in the thesis. A clear explanation of these will help narrow the otherwise large field of study and give the reader a more precise indication of, where the focus of the thesis is. Of the many traditional valuation models used in corporate finance, we have chosen to focus on a few specific models. We have chosen to focus on the standard Discounted Cash Flow model (DCF), Decision Tree Analysis (DTA), and lastly Real Option Theory. The focus of the Real Option theory is predominantly on the binomial- and quadranomial lattice models rather than formula based methods, since these are the most used methods in the practical literature (Bogdan & Villiger, 2010; Mun, 2002). The thesis focuses on the American Food and Drug Administration (FDA) regulation of the pharmaceutical industry. The reason for this choice is mainly because of the importance and size of the U.S. drug-market. It is the largest market for most pharmaceutical drugs. In the European Union, the European Medicines Agency (EMA) conducts regulation and approval of new drugs. In the broad perspective the requirements and regulation are similar to FDA regulation, hence it would not make any noticeable difference to conclusions of the thesis if the focus was any different. In the case study there will only be a limited focus on the strategic analysis in regards to the DCF valuation of the project. We are aware that a thorough and detailed strategic analysis is necessary in a useful valuation, but due to the fact that the DCF is limited to a single project and not a main part of the problem statement, the strategic analysis is reduced to a minimum. 4 Extensive design aims at elaboration, testing or generation of generalizable theoretical constructions by replicating or comparing cases (Eriksson & Kovalainen, 2011). Page 5 of 92

The case study takes start in one of Novo Nordisk s pipeline products. The product is called NN9926 and is an oral GLP-analogue focused on treating Type II diabetes. The product is currently in Clinical Phase I. The sales forecast and cost data will be based on a literature walkthrough of pharmaceutical industry sales and costs data. Page 6 of 92

2 Introduction to the pharmaceutical industry The following will introduce some very general characteristics of the pharmaceutical industry. Further greater focus is pointed towards the drug development process, and the chapter will serve as reference point throughout the thesis. The chapter will last discuss cost, success rates and time in relation to the industry, which will be used in the practical implementation of each valuation method. 2.1 Definition First, a definition of the pharmaceutical industry will be necessary for the further analysis and data collection. In order to understand and define, what the pharmaceutical industry is, one must look at two closely related terms the pharmaceutical industry and the biotech industry. The two terms are not straightforward and are often used indiscriminately. Therefore, it can also be fairly difficult and rather confusing to precisely classify these terms. Some make a strict distinction, while others do not distinguish between the two terms 5. In order to understand the different terms, it is important first to look at how pharmaceuticaland biotech firms develop their respective drugs. Biotech companies use biotechnology 6 in the drug development process, whereas conventional pharmaceutical companies predominantly rely on chemical-based synthetic processes to develop new drugs (Ferrara, 2011). Secondly, the size and scope of operations are often different. Pharmaceutical companies usually have the resources and capabilities to produce new drugs at large scale and successfully market them thereafter. This is often not the case for biotech companies, since they are generally smaller and more specialized in the research and development process. After initial development of a new drug they typically sell or license the rights to produce and sell that drug to a larger pharmaceutical company (Ferrara, 2011). Another difference is related to the threat of generic products 7. Biotech firms generally face less generic competition due to the fact that it is usually more difficult, time consuming, and costly to develop a new drug using biotechnology than using a traditional drug development process (Ferrara, 2011). Even though there are some clear differences between pharmaceutical and biotech firms, a strict separation of them may not be as forthright as one should think. This is mainly because 5 For statistical purposes no separating of the terms are often used. 6 Biotechnology is the manipulation of microorganisms (such as bacteria) or biological substances (like enzymes) to perform a specific process. 7 Generic competition: Competition from look-a-like drugs that roughly offer the same efficacy but at a much lower price. Page 7 of 92

many larger firms, which generally are perceived as pharmaceutical firms, i.e. Pfizer, Novo Nordisk, and Eli Lilly use biotechnology in the development of new products as well. In sense having a biotech company in-house 8. This fact makes a separation of the two types of firms into two completely separate industries rather unnecessary, since the development of new drugs face the same regulations and requirements, and most data available do not distinguish between the two terms. A common and simple definition of the pharmaceutical industry is an industry comprised of firms engaged in the discovery, manufacturing, and sales of drugs, biologics, vaccines and medical devices (Ferrara, 2011). In the following sections, when the industry is mentioned, it is based on the just stated definition of a pharmaceutical company. 2.2 Size and sales In the United States and Europe the pharmaceutical industry plays a major role in the society and economy. The pharmaceutical industry is the second largest US export sector and a major employer, estimated to directly provide jobs to 655,000 people in the US. In total, directly and indirectly, the sector supports over 3.1 million jobs nationwide in the US (Ding, Eliashberg, & Stremersch, 2014). The concentration of global sales is another clear indication of the importance for these regions. In 2008 the US accounted for 48 per cent of the total global sales, while Europe accounted for 29 per cent, between them sharing roughly 80 per cent of the total global sales of pharmaceuticals drugs (Boldrin & Levine, 2008). The US drug market and the appertaining medical legislation of FDA thereby have a major influence on the global pharmaceutical industry. Putting these percentages into perspective, the global sales potential of the pharmaceutical market was in 2009 estimated to be US$ 837 billion and today estimates go as high as US$ 1,1 trillion (Ding et al., 2014). The pharmaceutical market is estimated to be worth close to US$ 1,6 trillion in 2020 (PWC, 2012) 9. The rapidly growing and aging world population is the main driver for the increased demand for pharmaceutical drugs. 2.3 Research & development A unique characteristic of the pharmaceutical industry is the amount of resources allocated to Research and Development hereafter (R&D). Some estimates indicate that pharmaceutical firms are accountable for 19 per cent of all R&D spending worldwide and that US pharmaceutical R&D spending alone make up 36 per cent of the total pharmaceutical R&D in 8 In-house meaning that a part of the company is doing research by using biotechnology and combining the two approaches. 9 See appendix 5 for industry forecast. Page 8 of 92

the world (Ding et al., 2014). These figures give a clear picture of an industry that is heavily engaged in R&D processes but also indications of an industry that is extremely reliable on the R&D process. According to the PhRMA, the Pharmaceutical Research and Manufacturers of America, members of the organisation currently have close to 3,000 drugs in different stages of development in their pipelines. Of the 3,000 different drugs under development naturally most of pipeline-drugs lay within the groups that have some of the largest sales in the US. The top three categories are oncologics, respiratory agents, anti-diabetics and lipid regulators 10 (Ding et al., 2014). Industry measures on cost, time, and success rates of drug development will be presented later in this chapter. 2.4 Patents Innovation and R&D is undoubtedly connected with patents and intellectual property rights. The close connection between the two is fairly obvious, when looking at the resources pharmaceutical companies allocate to drug development and the threat from generics. Pharmaceutical companies are of course then dependent on the protection that patents offer. This thesis will not go in to detail about different patent-systems and their mechanics, but instead focus on some few specifics about the industry. Pharmaceutical companies are, as other industries, protected by 20 years of patent protection but beyond that there are some specific characteristics about the industry (Clift, 2008). Unlike other R&D-intensive industries, pharmaceutical firms do not have the option first to disclose their findings at the final stage of development, but have to reveal their discovery very early in the development. This is mainly due to requirements from government agencies and the fact that much development involves human trails, as will be elaborated later (Lehman, 2003). The lengthy R&D time period, which is also presented in the following sections, also constitute a special case for the pharmaceutical industry. The time between filing the patent and releasing the product to the market reduces that exclusivity a patent otherwise provides. In 1984 the US government introduced a special act called the Hatch-Waxman Act, which enabled the extension of a pharmaceutical patent with up to five years (Clift, 2008). The median or mean peak sales of pharmaceutical drugs are undoubtedly connected with their patent periods. Research suggests that the average percentage decline in sales the first four years after patent expiry is 31, 28, 20, and 20 per cent respectively (Grabowski, Vernon, & DiMasi, 2002). 10 Lipid regulators are regulators that affect the levels of lipid, such as cholesterol and fat, in the blood. Page 9 of 92

2.5 Drug development In order to understand how pharmaceutical companies develop new drugs and some of the challenges they face, the next section will describe a typical development process of a new drug. It is worth mentioning that the process described follows the regulation of the US Food and Drug Administrations (FDA). Importantly, creating new drugs in the twenty-first century is no longer a series of accidental, serendipitous breakthroughs. Instead, a long and systematic process requiring steadfast commitment, diligence, and meticulous work has taken the place of the previous haphazard experimentation (Ding et al., 2014, page 26) In order to get an overview of the development process figure 2.1 is used to describe the process of development in different steps. It shows in crude terms the different drug development stages from discovery of lead compounds through clinical phases (CP) and final approval of the drug. Figure 2.1: The drug development process Discovery CP I CP II CP III NDA - Approval Post Approval Source: Own creation Discovery: The first step in the development of a new pharmaceutical drug is the study of a disease at a molecular and cellular level. The discovery process is a complex process and requires both the involvement of chemists and biologists. This stage is highly time-consuming and uncertain, which results in a large amount of abandoned entities (Kellogg & Charnes, 2000). More detailed, the stage involves the analysis of basic cellular processes at both a healthy and pathologic state, and by comparing the two different states several diseaseresponsible actors are identified as possible drug targets. Before any of the discovered compounds can be tested in a human body, several tests in vitro and in vivo, test in tubes, and in living cells must be conducted. Furthermore, researchers must find the most appropriate and safe dose of drug for the further tests in animals (Ding et al., 2014). Several candidate drugs that seemed successful are often abandoned due to problems of low efficiency, toxicity, Page 10 of 92

or poor absorption (Bogdan & Villiger, 2010). Upon completion of the drug discovery process researchers must prepare for the more critical stages in the innovation process drug development through clinical trials on humans. Before the candidate drug can be tested on humans, the researchers must hand in an Investigational New Drug application (IND) to the national drug agency. The application must include all evidence of the previous steps and must proof that the candidate drug fulfils all present requirements and standards. Clinical phases: In clinical phase I (CP I) the candidate drug is tested in humans for the first time, and the study is usually conducted on a group of volunteers of 20 to 100 healthy persons (Bogdan & Villiger, 2010). The test is conducted to establish a more precise dosage and to gather more documentation about the absorption, distribution, embolic, and excretion effect of the human body. Short-term side-effects are studied and the desired effects of the drug are compared to the established treatments to determine if the drug provides a better alternative (DiMasi & Paquette, 2004; Ding et al., 2014). If this is fulfilled, then it will move on to clinical phase II. In phase II a larger group of 100-300 people having the target disease or condition are tested. The purpose is to define the most appropriate dose and further to prove the effectiveness of the drug. The drug must proof its effectiveness, since the tested group of people in phase I were healthy people (Bogdan & Villiger, 2010). Researchers strive to understand if the drug has good efficacy, and whether the drug has any short-term side effects. Further, the drug must prove a clear benefit over existing treatments in terms of efficacy, safety, and delivery (Bogdan & Villiger, 2010; DiMasi & Paquette, 2004). If the clinical phase II of the drug is successful, it is taking forward to a large-scale test of 500-20.000 patients having the target disease, the clinical phase III. A large and diversified group of people, often from different nations, are included as a diversified test group is necessary for the following studies (Bogdan & Villiger, 2010). The aim of the large-scale phase III is to confirm and significantly prove the effectiveness of the treatment under different conditions. For establishing a significant evidence of efficacy and safety, the drug must be comparatively tested against different placebo options and other standard treatments. If the drug succeeds in being safe and effective the pharmaceutical company can file a New Drug Application (NDA) to the drug agency requesting approval (Ding et al., 2014). Page 11 of 92

NDA: The NDA must include all data, documentation, and statistics of the testing and proof of the safety and efficacy of the drug. The approval must also include proposals for specific labelling and manufacturing. A NDA may consist of more than 100.00 pages of data and evidence. According to Bogdan & Villiger (2010) it is very likely that the regulatory authority asks for further clinical trials or even rejects the marketing approval. Post approval: After the pharmaceutical drug has been approved by the drug agency and sold to the market, the test and research process has to continue. Pharmaceutical companies must continue to monitor and observe carefully for newly found adverse and long-term side-effects (Bogdan & Villiger, 2010). The company completes periodic reports to the drug agency on quarterly basis the first three years and annually afterwards (Ding et al., 2014). To sum-up the process of developing a new pharmaceutical drug, it is a complicated timedemanding process affected by high uncertainty in each development stage. The next section will present a litterateur walkthrough of some of the latest research within this field. 2.6 Success rates, cost, and time The two previous sections have introduced the pharmaceutical industry and explained the clinical phases of pharmaceutical drug development. The following section will investigate and discuss the available data of clinical success rates, R&D costs, and time effort in developing new drugs. This section is essential for the thesis, since these findings will be the source of the industry data used in the later case study. Each section will start with a list of relevant authors and continue with an explanation of these and conclude with a table that summarises the findings. From our research, we find that only few studies have investigated costs and success rates of the pharmaceutical industry and that they differ significantly. The literature is further complicated by the reluctance of the industry to publish confidential industry data of drug development. Most data in the literature is derived from the Tufts Centre for the Study of Drug Development (CSDD), where data is provided by big unnamed pharmaceutical companies (Bogdan & Villiger, 2010). According to Light & Warburton (2011), the CSDD has received substantial industry funding for years and is a repository, where companies submit Page 12 of 92

their closely guarded figures on R&D. As several of the following studies are based on data from CSDD, the studies of cost estimates are critically examined. 2.6.1 Success rates The definition of the success rate is the chance that a pipeline-drug entering a development phase reaches the next phase (Bogdan & Villiger, 2010; Ding et al., 2014). In the following the average cumulative success rate is defined as the success rate from first in man to marketing of the drug (Kola & Landis, 2004). The publication by Ding et al. (2014) advocates that success rates are complex to understand and interpret upon, as the success rates associated with passing each stage are different among different drug candidates. Meaning, that probabilities of success vary quite substantially within different pharmaceutical classes. Some of the most cited authors in the literature and industry are the publications by DiMasi & Grabowski (2007) and DiMasi, Feldman, & Wilson (2010), who both are associated with the before mentioned CSDD. The publication from DiMasi & Grabowski (2007) finds the success rates of each clinical phase to be the following; CP I 71 per cent, CP II 44,2 per cent and CP III 68,5 per cent. Based on these, the clinical cumulative success rates is 21,5 per cent that a new drug would be able to reach the market. The later publication by DiMasi et al. (2010) studied a larger sample of drugs. The sample consisted of 1.738 different compounds, from the 50 largest pharmaceutical companies in the US 11. They find that approximately one in six or 17 per cent of all new drugs were approved in the period of 1994-2009. Their results show that failures occurred earlier in the clinical phases. The success rate in each clinical phase was; CP I 67 per cent, CP II 41 per cent, and CP III 63 per cent, and 90 per cent for approval. The most recent study on pharmaceutical drug development success rates is an article published by DiMasi et al. (2014) on behalf of CSDD. The study present the following success rates; CP I 59,52 per cent, CP II 35,52 per cent, CP III 61,95 per cent, and last the approval phase 90,35 per cent. The cumulative success rate is 11,83 per cent, which is much lower, compared to the previous studies. Compared to the two previously mentioned studies the 11 The size was measured in sales numbers Page 13 of 92

success rate of clinical phase I and II is substantially lower. The above results are presented in table 2.1. Table 2.1: Clinical success rates Discovery CP I CP II CP III NDA Total Authors / Succes per cent DiMasi & Grabowski, 2007 71% 44,2% 68,5% - 21,5 % DiMasi et al., 2010 67% 41% 63% 90% 15,58% DiMasi et al., 2014 59,52% 35,52% 61,95% 90,35% 11,83% Source: Own creation As seen in table 2.1, the studies of success rates show some different results. The differences in percentage are caused by the different measurements, pharmaceutical classes or other statistical estimations used by the authors. Given that the approval phase is not a part of the total success rate in DiMasi & Grabowski (2007) the estimations is higher compared to the two other studies. If we adjust the estimate with the approval phase from DiMasi et al. (2010) and DiMasi et al. (2014), which is around 90 per cent, we estimate a total success rate of 19,3 per cent. 2.6.2 Costs Several studies have analysed the development process of the pharmaceutical industry but only few of them have studied the costs of the R&D process. A significant variation in the cost estimations among the studies complicates the findings. Through the literature we find that drug development costs range from US $75mio to $4billion dollars (PWC, 2012). Most studies lean towards the higher end of the range, as seen in the table below. The significant cost variation makes it necessary to examine the following literature critically, as mentioned in the introduction. The significant cost variation can often be explained by two components (Ding et al., 2014). The first component is a question of the definition of R&D costs. As just explained in the previous section on success rates, the pharmaceutical industry is influenced by low success rates meaning that several drugs are abandoned before a successful drug is discovered and developed. Some argue that the R&D cost should include cost of both the successful and the Page 14 of 92

abandoned drugs, while others do not. The second component is the allocation of opportunity cost due to the long time horizon of the development process. The average time of a drug development is 12 years and the opportunity cost thus has a significant influence on the total cost allocation (Ding et al., 2014). Both components will be explained more in the following examples. Similar to the literature on success rates the publication by DiMasi & Grabowski (2007) is one of the most cited by the industry. The authors estimated in 2007 an average out-of-pocket cost per new drug of US$672 million and after capitalising that at an opportunity cost of 11 per cent the total pre-approval cost was estimated to $1318 million dollars. The paper by Light & Warburton (2011) criticises these estimates and argues that since none of the drugs were titled or specified in therapeutic classes, it is not possible to verify these results. The authors further argue that the estimates do not include any R&D tax adjustments, and that they should be based on median numbers rather than a mean, which decrease the influence of extreme outliers. Most criticised is the allocation of 11 per cent opportunity cost, which doubles the total cost estimations from $672 million to $1318 million dollars. Even if one accept the use of opportunity costs, US government guidelines call for using 3 per cent and not 11 per cent opportunity cost. As a part of their critic, they calculate their own cost estimations based on the data of DiMasi & Grabowski (2007) and found a median cost ranging from US$180-231 million dollars. In 2012 Price Waterhouse Cooper estimated the development cost of an average pharmaceutical company (PWC, 2012). They found a significant variation in costs depending on therapeutic classes similar to other studies. Based on average costs and average attrition rates in each phase of the R&D process, the cost of the R&D process was estimated to be around US$701 million dollars per pharmaceutical product. More detailed numbers for each R&D process were given as; preclinical and development $87 million, CP I for $130 million, CP II for $190 million, CP III for $268 million, and lastly $26 million dollars in approval. The publication did not include the preceding calculations for the estimations and therefore not possible to discuss further. The 2010 publication by Bogdan & Villiger (2010) estimated the cost of each clinical phase in a small and medium sized pharmaceutical company. Compared to the other literature, the Page 15 of 92

estimations by Bogdan & Villiger (2010) provide some of the lowest cost estimates. This is mainly due to the fact that the estimations are made for smaller companies. More essential questions of both opportunity cost and failure costs are not clarified, and the reader must assume that both components have been left out. This could also explain the low cost measures in each phase of development. The total average cost accounts for: research and discovery US$ 4-6 million, CP I for $1-5 million, CP II for $3-11 million, CP III for $10-60 million, and last the approval phase for $2-4 million dollars. The maximum total cost denotes to $86 million dollars, which is incredible low compared to the other findings. The latest report by CSDD DiMasi et al. (2014) estimates the average capitalised cost to be US$2.2558 million dollars, allocated as $1.098 million cost in pre-human and US$1.460 million dollars in clinical testing. The cost estimate is calculated using a 11,4 per cent opportunity cost, which almost doubled the out-of-pocket costs from US$ 1.395 million to 2.558 million dollars. The above results are presented in table 2.2. Table 2.2: Cost of drug development Discovery CP I CP II CP III NDA Total Authors / Costs million dollars DiMasi & Grabowski, 2007* 150 522 672 DiMasi & Grabowski, 2007 439 879 1318 Light & Warburton, 2011 180/231 PWC, 2012 87 130 190 268 26 701 Bogdan & Villiger, 2010 6 5 11 60 4 86 DiMasi et al., 2014 1098 1.460 2.558 Source: Own creation *Out-of-pocket costs not capitalised (DiMasi & Grabowski, 2007) As seen in table 2.2 the different cost estimates differ quite substantially. Cost figures range from US$ 86-2.558 million. When taking the different methods of approach and component assumptions in each of the studies into consideration, the variation may not be as surprising as first noticed. Page 16 of 92

2.6.3 Time According to Ding et al. (2014) a successful pharmaceutical company must be able to balance returns, uncertainty, and not least time. The majority of the income of a pharmaceutical firm is generated from drugs with patent protection, which will fade out as soon the patent ends (Ding et al., 2014). The time of research and development has a significant impact on earnings in the pharmaceutical industry, since a long development process reduces time of patent protection (as mentioned earlier on page 9). In 2010 Bogdan & Villiger (2010) found that small and medium sized pharmaceutical companies had an average development time of; discovery and research 30-42 month, CP I for 18-22 month, CP II for 24-28 month, CP III for 28-32 month, and approval for 16-20 month. The maximum length of the R&D process is 12 years, which is similar to Ding et al. (2014) and a minimum time of nine and a half year. The latest publication by the CSDD DiMasi et al. (2014), which is based on ten unknown pharmaceutical and 106 new drugs in the US, estimates the following time length of the total development process; discovery 31,2 month, CP I 19,8 month, CP II 30,3 month, CP III 30,7, and last approval 16 month. The total time length from discovery to approval was observed to be 10,6 years (DiMasi et al., 2014). In 2011 Kaitlin & DiMasi (2011) completed a study on the length of clinical phases and the approval time on new drugs in the US. The data was gathered using the before mentioned CSDD database from the period 1980 to 2009. They grouped the data in to brackets of fiveyear periods to see the development in clinical phase- and approval time. The earlier results from their study seem less relevant for this thesis, why only the newest results will be presented here. They did not distinguish between CP I to III but grouped them in one category, but they had separate data for the approval time. The total clinical phase time was in the last period of data, 2005-2009, 76,8 month, while the approval time comprised to 14,4 months. The total time from first in man testing (clinical phase I) to approval was 91,2 months (7,6 years). Page 17 of 92

Table 2.3: Clinical time length Discovery CP I CP II CP III NDA Total Authors / Time in months Ding et al., 2014 12 yr. Bogdan & Villiger, 2010 30-42 18-22 24-28 28-32 16-22 12 yr. DiMasi et al., 2014 31,2 19,8 30,3 30,7 16 10,6 yr Kaitlin & DiMasi, 2011 76,8 14,4 7,6y r. Source: Own creation As illustrated in table 2.3, the variation in time length is rather low. From the above, we find the typical time length of a drug development to be 7,6 to 12 years. It is worth noticing that the publication by Kaitlin & DiMasi (2011) finds a shorter period of drug development. This is mainly because the discovery phase is not included in the development period. Furthermore, the study does not divide the clinical phases in to separate phases, which makes it less usable and expressive. 2.7 Sum-up This chapter has presented some of the main characteristics of the pharmaceutical industry. Characteristics that give an indication of an industry that generally face high uncertainty and risk. From complex patent legislation to the very risky processes and clinical phases pharmaceutical companies go through in the development new drugs. Combining this with the just presented actual industry measures on R&D-cost, drug development success-rates, and time needed to develop a new approved drug, it is clear that valuation of pipeline-projects within this industry calls for much consideration and attention. Do the well-known fundamental valuation methods suit the characteristics of the industry or are other valuation methods more suitable when valuating projects within this industry? During the theoretical discussion of different valuation methods, continuous references will be made to this chapter s description of the characteristics of the pharmaceutical industry. Page 18 of 92

3 Valuation methods in the pharmaceutical industry The previous sections have introduced the pharmaceutical industry and some few significant characteristics of it. To answer the problem statement, this section will analyse and discuss some of the theoretical differences between fundamental valuation models and the more complex real options valuation methods and their underlying assumptions. Each valuation method will be theoretically discussed and evaluated upon four defined criteria. The potential of each valuation methods will be evaluated after the practical implementation. The decisions of financial analysts in corporate finance, in a very simplistic way, can be divided into two overall decisions. First one is the investment decision, which is choosing the projects that have positive NPV. Second one is the finance decision, which is deciding on how to finance the selected projects. All decisions have the overall goal of maximizing the market value of a given portfolio of investments. These decisions are undoubtedly connected with the strategy a company pursues. And in an increasingly uncertain global marketplace, strategy and strategic flexibility are becoming more important for firms in order to capture the advantage of future opportunities, and limit losses of any unfavourable developments (Smit & Trigeorgis, 2004). Having the previous chapter in mind, it is clear that the pharmaceutical industry to a high degree face such challenges. Therefore, the purpose of this chapter is to present, analyse, and discuss possible project-valuation methods for the pharmaceutical industry in order to continue with an evaluation of the practical implementation through a case study and finally make some statements regarding the optimal method to use. As mentioned in the introduction, the perspective of view is of financial analysts, meaning it is an outside-in perspective. Also, as mentioned earlier several studies have previously found that financial analysts favour the present value approach and that the real options approach is hardly ever used (Block, 2007; C. V. Petersen & Plenborg, 2012). And, this is where the thesis has its merits an investigation of which valuation methods are most suitable in theory and practice when valuing pharmaceutical drug development projects. Page 19 of 92