A systems approach to rd investment

List of Tables Table 1-1: Critical review of traditional approaches ...5 Table 1-2: Benefits of the VaR approach ...6 Table 1-3: Thesis organisation...7 Table 2-1: Companies with the hig

A SYSTEMS APPROACH TO R&D INVESTMENT

NG CHU NGAH

NATIONAL UNIVERSITY OF SINGAPORE

2007

A SYSTEMS APPROACH TO R&D INVESTMENT

NG CHU NGAH

B.Eng.(Hons.), NUS

A THESIS SUBMITTED

FOR THE DEGREE OF MASTERS OF ENGINEERING

DEPARTMENT OF INDUSTRIAL AND SYSTEMS

ENGINEERING

NATIONAL UNIVERSITY OF SINGAPORE

i

Summary

R&D activities are increasingly recognised as the engine for corporate growth, yet they remain a challenge when it comes to valuation and selection R&D projects carry huge risks which make the potential high payoffs illusive, but these payoffs are precisely the incentive for the examination of project selection methodologies While there is no hard and fast rule to compare and select R&D projects, this report aims to propose possible improvements to the selection and management process

Our work builds on the concept of the strategy-monetary division of the Organisational Decision Support System (ODSS) and customises it into three segments:

(1) ONline R&D focus selection (2) OFFline project valuation, and (3) ONline portfolio selection

At the online level, Real Options (ROs) thinking is incorporated to specifically deal with the option to defer and its tradeoff of gaining competitive advantage We reason that this RO should make sense only if considered at the strategic level because of the existence of tradeoffs

Zeroing in on the individual projects at the offline stage, we separate risks and payoffs with a new perspective—the “within firm” and “beyond firm” distinction Decision Analysis (DA) is then brought in for the modelling of sequential decisions Since real

Summary projects are not as flexible as financial options when it comes to opportunities to exit, decision trees are sufficient in capturing the option to abandon

However, DA also has its shortcomings It depends largely on subjective expert opinions, and these are costly to obtain yet not reliable particularly at project outset as studies have shown Nonetheless, judgemental methods are unavoidable for R&D projects due to decreased liquidity (Drzik, 1996) We thus propose the borrowing of financial market data to replace, or at least complement, the subjective probabilities used in DA, especially at the initial project selection stage This recommendation relies

on the assumption that the eventual value of R&D projects would be reflected in the shareholder returns in financial markets, following the launch of the new products or technologies

Our scope is hence limited to five industries as identified by Foster and Kaplan (2001), where there appear to be a positive correlation between R&D investment and shareholder return They are namely pharmaceutical, pulp and paper, commodity and specialty chemicals, aerospace and defence, as well as oil extraction

For the treatment of the data, we extend the idea of collapsing the leaf-values into extremal values using the simplification rule, and adopted a suitable common financial risk measure—the Value at Risk (VaR) However, VaR has, in recent years, been

bi-discredited as an incoherent risk measure (Artzner et al., 1999) A similar risk measure

called the Expected Tail Loss (ETL)—the expectation of losses beyond VaR—turned out to be a possible remedy The purpose of both is the same and the calculation of

Comparing VaR with simulation, we note that VaR is like integrating real options into simulation As pointed out by scholars, simulation is useful but probably the extent of usefulness is limited to the central 80% of the information due to the consideration of options and management flexibility Thus, the interval between the VaRs allows us to focus on the essential information

On the technical side, the recent use of Extreme Value Distributions (EVDs) and Generalised Pareto Distributions (GPDs) to approximate VaRs is appealing in our study as they would allow direct simulation of the boundary quantiles Our results show that the GPD method is preferred over the parametric method for both the upper and lower-bound VaR This method would thus enable us to calculate a baseline for the payoff/loss estimation, while allowing decision makers to see the maximum potential of particular projects, thereby setting an investment limit before abandonment should be exercised

Acknowledgement

This thesis extends my B.Eng honours project Along the entire journey, I am grateful

to many people around me for their help, guidance, encouragement, and concern

First and foremost, I would like to express my sincere appreciation for my mentor, A/P Poh Kim Leng, who is always so patient and willing to take time off his hectic schedule to give me invaluable advice—work or non-work related alike—and to explain concepts foreign to me I thank him also for granting me the freedom and independence to explore possible research areas which I have interest in Having such

a supportive supervisor is indeed a blessing

Secondly, my heartfelt gratitude also goes to my internship supervisor, Mr See Chuen Teck, for his inputs and tips Equipped with practical experience and industrial knowledge, he is a great source of information and ideas Indeed, a discussion with him beats all blind research

Thirdly, a special mention and acknowledgement for Dr François Longin who had guided me through my first research training at ESSEC Business School in France in the year 2004 This invaluable experience gave me a peek into the vast world of Finance, and introduced me to the concept of Value-at Risk which serves as one of the preliminary sources of inspiration for this thesis

Acknowledgement

v

In addition, I would also like to express my appreciation to the department and the university for giving me the chance to fulfill my third and fourth year of my undergraduate studies at the Ecole Nationale des Ponts et Chaussées (ENPC) in France,

as well as the opportunity to present my work at the Asia Pacific Industrial Engineering Management (APIEMS) conference 2006 at Bangkok The various programs and experiences have been highly enriching

On this note, my sincere thanks also go to DSTA who supported and financed my studies in France and a return trip for the presentation of my internship at DSTA which required my return to ENPC during my research period; as well as the NUS Graduate Office for coordinating my return to the department

Finally, I dedicate this thesis as a gesture of thanks to my parents and brothers for their unyielding support and advice; to my friends particularly Zhili and Zhiyun for their concern for my progress and adaptation back to the NUS culture; and to my lab-mates who helped make my integration into the community a smooth and pleasant one

Table of Contents

Summary i Acknowledgement iv Table of Contents vi Terms and Abbreviations _ ix List of Figures _ x List of Tables xii

2.4 R&D Project Lifecycle 12

2.5 R&D Project Risks _ 14

2.6 Conclusion _ 16

Chapter 3 R&D Project Valuation Tools 17

3.1 Discounted Cash Flow 17

3.2 Decision Analysis _ 19

3.3 Simulation _ 23

3.4 Real Options Approach _ 24

3.5 Conclusion _ 30

Chapter 4 R&D Project Selection Cycle and ODSS 31

4.1 Offline: Individual Project Analysis 32

4.2 Online: R&D Capital Allocation _ 37

Table of Contents

vii

Chapter 5 Framework 41

5.1 A novel view of the R&D Project Lifecycle _ 41

5.2 Project Selection: ODSS modified _ 42

5.3 Project Management: Online Decisions 47

5.4 Consolidated framework 49

Chapter 6 Offline Project Valuation Considerations 50

6.1 Role of Real Options _ 50

6.2 Inconveniences of DA 51

6.3 Assumption 53

6.4 Risk Measure: Value-at-Risk (VaR) 55

Chapter 7 Value-at-Risk (VaR) 61

7.1 Calculation of the VaR by the Historic Method _ 61

7.2 Calculation of the VaR by the Parametric Method _ 62

7.3 Calculation of the VaR by the Classical EV Method _ 63

7.4 Calculation of the VaR by the Modern EV Method 70

7.5 How VaR Adds Value 73

Chapter 8 Case Example: 40 stocks from the NYSE 74

Table of Contents

Appendix A Financial Option Pricing 101

A.1 Option Pricing Models - Samuelson (1965) _ 101

A.2 Option Pricing Models - Black-Scholes model (1973) _ 101

A.3 Option Pricing Models – Merton (1973) 107

A.4 Option Pricing Models - Cox, Ross, & Rubinstein (1979) 108

Appendix B Fitting Extreme Value Distributions 116

B.1 Parameter Estimation by Maximum likelihood Method 116

B.2 Standard Error _ 118

Appendix C Calculating VaR using BestFit data 120

C.1 Gumbel 120

C.2 Weibull _ 121

Appendix D P-P plots for GEV and GPD fitting 122

Appendix E 5 * 8 Companies selected for study 134

E.1 Medical Laboratories and Pharmaceuticals _ 134

E.2 Pulp and Paper 138

E.3 Commodity and Specialty Chemicals _ 141

E.4 Aerospace and Defence _ 144

E.5 Oil Extraction 147

Appendix F Results from Case Study –VaR Max 149 Appendix G Results from Case Study –VaR Min 152

ix

Terms and Abbreviations

ETL / ES Expected Tail Loss / Expected Shortfall

EV(T/D) Extreme Value (Theorem/ Distribution)

GEV Generalised Extreme Value Distribution

LEAPS Long-Term Equity Anticipation Securities

ODSS Organisational Decision Support System

R&D Research and Development

Simulation Monté Carlo Simulation

List of Figures

Figure 2-1: Types of Innovation: Ritcher scale of innovation 8

Figure 2-2: Risk-payoff matrix showing position of R&D 11

Figure 2-3: Lifecycle of an R&D Project as proposed by Jensen & Warren (2001) 12

Figure 2-4: Decision criteria of R&D projects 14

Figure 2-5: Technical uncertainties arising at different stages of the R&D project 15

Figure 2-6: Uncertainty in different phases of R&D project 16

Figure 3-1: Abstraction of DA process 19

Figure 3-2: Decision factors of R&D projects 20

Figure 3-3: Decision Tree differentiating risk probabilities and risk impacts 21

Figure 3-4: Uncertainty “dissected” 25

Figure 3-5: Identification of the option zone 28

Figure 5-1: R&D Project 42

Figure 5-2: Framework for R&D capital allocation 45

Figure 5-3: Efficient Frontier and AHP focused on clusters 46

Figure 5-4: Reversal to Online Capital Allocation during Project Life 48

Figure 5-5: Lifecycle of R&D projects: From selection to management 49

Figure 6-1: Consideration of extremes using Substitution Rule 52

Figure 6-2: Comparison of VaR and simulation 57

Figure 7-1: Illustration of the calculation of the VaR by the historic method 62

Figure 7-2: Schema illustrating the use of classical EV method to calculate the VaR (Longin, 1998) 64

Figure 7-3: Graphs of extreme value distributions 67

Figure 7-4: Illustration of the calculation of the VaR by the EV method 69

Figure 7-5: Contrasting between GEV (left) and POT (right) 70

Figure 8-1: Skeleton of Computations for a Single Company 77

Figure 8-2: Upper and lower VaRs as calculated using EVT 79

Figure 8-3: BLOCK-MAXIMA-of-20 Probability Plots – GEV Model vs Empirical 83

List of Figures

xi

Figure 8-5: Peaks-over-(95%)Threshold MAXIMUM Probability Plots – GPD Model

vs Empirical 88 Figure 8-6: Peaks-over-(95%)Threshold MINIMUM Probability Plots – GPD Model

vs Empirical 89

Figure A-1: PDF of lognormal distribution 105 Figure A-2: Binomial lattice 109 Figure D-1: BLOCK-MAXIMA-of-15 Probability Plots – GEV Model vs Empirical 122 Figure D-2: BLOCK-MINIMA-of-15 Probability Plots – GEV Model vs Empirical 123 Figure D-3: BLOCK-MAXIMA-of-10 Probability Plots – GEV Model vs Empirical 124 Figure D-4: BLOCK-MINIMA-of-10 Probability Plots – GEV Model vs Empirical 125 Figure D-5: BLOCK-MAXIMA-of-5 Probability Plots – GEV Model vs Empirical 126 Figure D-6: BLOCK-MINIMA-of-5 Probability Plots – GEV Model vs Empirical 127 Figure D-7: Peaks-over-(97.5%)Threshold MAXIMUM Probability Plots – GPD

Model vs Empirical 128 Figure D-8: Peaks-over-(97.5%)Threshold MINIMUM Probability Plots – GPD Model

vs Empirical 129 Figure D-9: Peaks-over-(92.5%)Threshold MAXIMUM Probability Plots – GPD

Model vs Empirical 130 Figure D-10: Peaks-over-(92.5%)Threshold MINIMUM Probability Plots – GPD Model vs Empirical 131 Figure D-11: Peaks-over-(90%)Threshold MAXIMUM Probability Plots – GPD

Model vs Empirical 132 Figure D-12: Peaks-over-(90%)Threshold MINIMUM Probability Plots – GPD Model

vs Empirical 133

List of Tables

Table 1-1: Critical review of traditional approaches 5

Table 1-2: Benefits of the VaR approach 6

Table 1-3: Thesis organisation 7

Table 2-1: Companies with the highest R&D intensity in the FT Global 500 10

Table 4-1: Five-phase framework for project selection propsed by Ghasemzadeh & Archer (2000) 31

Table 5-1: Modified ODSS framework for initial investment into R&D projects 43

Table 6-1: Overview of Project Valuation Considerations 50

Table 8-1: List of 40 companies chosen 75

Table 8-2: Comparisons of GEV VaR calculations with parametric approach 81

Table 8-3: Comparisons of GPD VaR calculations with parametric approach 86

1

“In the current risk management world, the hunt for risk takes three general forms One is the hunt for treasure - combing through markets for the next stellar investment management opportunity Second is the hunt for hazard - identifying unwanted risks in asset and liability portfolios through techniques such as simulation, stress-testing, and backtesting Third is the hunt for knowledge - forensic investigation of unexpected gains and losses.”

Tanya Styblo Beder The Great Risk Hunt

The Journal of Portfolio Management Special Issue, May, 1999, p 29

Corporate financial strategy points out that it is not only the earnings that sustain a firm, but also, indispensably, the growth which helps it thrive R&D activities are increasingly recognised as the engine for growth, but there remains no hard and fast rule for the allocation of budget on R&D Neither is there a definite way to compare and select R&D projects which essentially carry huge uncertainty It is also not clear how the uncertainty should be managed, particularly how to decide whether to abandon the project to prevent huge losses

The purpose of this work is thus to propose an overall selection and management framework that incorporates various tools and ideas, yet remaining logical and easy to understand This is in line with the observation by Cooper (1999) that the best companies use several complementary methods at the same time

Chapter 1 Introduction

The Real Options (RO) approach is theoretically attractive with its lofty ambition to give a numerical value to management flexibility Unfortunately, it is mathematically complex as fundamental issues like volatility remain unresolved; and arguably lopsided because it fails to recognise the tradeoff between having a business option and gaining strategic competitive advantage

Despite being greatly motivated by real options which exist in the entire project lifecycle, we realise that real options are vastly different from financial options, which explains why a direct use of option pricing is not pragmatic, and thus the real option methodology cannot stand as a valuation tool

However, an important issue that real options highlight is the strategic concerns of

embarking on R&D projects Indeed, as Torkkeli et al (2003) pointed out, there is a

straightforward relationship between strategic orientation and R&D directions through technology selection: technology selection is based on company’s strategic goals and targets, and technologies selected draw new R&D directions for the company This means that strategic consideration is of paramount significance in R&D project selection

1.2 Proposed Improvements

Persuaded that strategy cannot be fully quantified but that RO has its merits, we adapt the separation into online and offline segments as in the Organisational Decision

Chapter 1 Introduction

3

Support System (ODSS) for project portfolio selection (Ghasemzadeh & Archer, 2000;

Tian et al, 2002)

The customised ODSS framework consists of three segments:

(1) ONline R&D focus selection (2) OFFline project valuation, and (3) ONline portfolio selection, which allow the separation of strategy from project valuation This also enables a proper consideration of the relevant ROs at appropriate levels

At the online level, we incorporate real option thinking (Myers, 1977; Kester, 1984; Faulkner, 1996; Kulatilaka and Perotti, 1998) into the framework Specifically, the option to defer should be considered at the management level in view of the existing or potential competition

Apart from the option to defer to be considered at the online stage, other relevant options can effectively be examined with the aid of a decision tree during the offline project valuation stage An in-depth exploration of various project valuation models accentuates the usefulness and clarity of modelling the project life with a decision tree

In fact, it makes redundant the continuous hedging as proposed by RO and stands out

as a sufficient starting point for projects which cater for sequential decision nodes

Nevertheless, Decision Tree Analysis, commonly referred to as DA, has its shortcomings In our study, we would prefer to distinguish the tool i.e Decision Tree from the methodology – DA, which depends on subjective expert opinions, and these

Chapter 1 Introduction are costly to obtain in terms of time and money yet not reliable as studies have shown Nonetheless, judgemental methods are unavoidable for R&D projects due to the decreased liquidity (Drzik, 1996)

We highlight the deliberate separation of risks and payoffs through the boundary of

“within firm” and “beyond firm”, and propose the borrowing of financial market data

to replace, or at least complement, the subjective probabilities used in DA, especially

at the initial project selection stage This recommendation relies on the assumption that the eventual value of R&D projects would be reflected as shareholder return in financial markets, following the launch of the new products or technologies These values are conditional on the success of the R&D projects

The above assumption scopes out five industries, as identified by Foster and Kaplan (2001), where there appear to be a positive correlation between R&D investment and shareholder return The five industries are namely pharmaceutical, pulp and paper, commodity and specialty chemicals, aerospace and defence, as well as oil extraction

Chapter 1 Introduction

5

Table 1-1 summarises the tools and context at the traditional level, as well as the proposed amelioration and paradigm shifts

Table 1-1: Critical review of traditional approaches

Traditional approaches Paradigm shift & Improvements

Tools – Real Options

- the concept of risk neutrality can be

applied to take care of choice of

Tools – Decision Analysis

DA depends on expert opinions for leaf

values and probabilities

Expert opinions are usually flawed (Meadows, 1968; Souder, 1978, 1969), but might be remedied by the adoption of the financial risk measure, VaR, to quantify values of R&D projects in particular industries identified

Context – at the valuation level

- Consideration of R&D projects at year

0 and projecting into 3 stages (Jensen

Context – at the strategic level

ODSS separates strategy and valuation

Our findings show that the Generalised Pareto Distribution (GPD) approach is preferred over the parametric method for both the upper and lower bound VaR This method would thus enable us to calculate a baseline for the loss estimation, while allowing decision makers to see the maximum potential of particular projects, thereby setting an investment limit before abandonment should be exercised

Table 1-2 summarises the advantages of VaR over the traditional approaches

Table 1-2: Benefits of the VaR approach

Traditional approaches How VaR value-add

Calculate Expected Value of individual

projects (NPV, DA, RO) for comparison

Use VaR to identify interval of possible returns and thus capture volatility

DA depends solely on expert opinions Financial data as captured by VaR as an

objective complement VaR need not assume a return distribution, yet also gives a fuller profile Direct simulation of boundary quantiles possible

Simulation to take into account fuller risk

profile

VaR is akin to RO incorporated into simulation

Chapter 5 introduces our framework from project conception to selection and finally to management

Next, we shall then focus on the offline stage for the alternative treatment of individual project valuation Chapter 6 explains all the considerations taken into account Chapter

7 dives into the various approaches to calculate the financial risk measure, VaR Chapter 8 gives a case example showing the calculation of VaRs from the New York Stock Exchange, as well as a discussion of the results

Finally, Chapter 9 concludes and states further work that might be worth exploring Table 1-3 illustrates the logical flow of the thesis by parts

Table 1-3: Thesis organisation

Background (R&D projects) Framework Offline segment

Chapter 2: the projects

Chapter 3: valuation tools

Chapter 4: selection cycle

Chapter 5 Chapter 6: considerations

Chapter 7: VaR Chapter 8: Case Example

Chapter 2 R&D Projects

“ a company may die a quick death if it does not manage its critical risks, it will certainly die a slow death if it does not take enough risks.”

James Lam

Enterprise: Risk Management, Wiley (2003), p 273

From a business point of view, innovation is beneficial to allow the renewal of the Foster’s S-curve (1986) of industry earnings so as to maintain sustainable growth One basic type of innovation is product innovation, which involves the introduction of a new good or service that has been substantially improved According to Foster and Kaplan (2001), innovation can be classified in increasing impact of wealth creation and

newness as incremental, substantial or transformational (c.f Figure 2-1) Different

level of innovation requires a different managerial treatment In our study, we are interested in the latter two types of innovation which usually stem from research and development (R&D) activities

Figure 2-1: Types of Innovation: Ritcher scale of innovation

innovation Incremental innovation

100

10

1

Chapter 2 R&D Projects

9

R&D activities refer to future-oriented, long term projects in science and technology that aim for breakthrough innovations, and are crucial in ensuring competitiveness in our ever-progressing society International studies have consistently demonstrated the positive correlation between R&D investment intensity and company performance measures such as sales growth, wealth creation efficiency and market capitalisation in the sectors where R&D is important In particular industries like pharmaceutical, pulp and paper, commodity and specialty chemicals, aerospace and defence, as well as oil extraction, it appears that the companies which concentrate on sustained growth through investment in R&D are most likely to achieve increased shareholder return 1(Foster and Kaplan, 2001)

R&D results in valuable inventions, ideas and designs which can be sources of potential value when it comes to gaining competitive advantage A variety of Intellectual Property Rights including patents and trademarks exists to help a company protect these valuable assets

In the 2005 R&D Scoreboard2, the US continues to score highly in "R&D intensity" – the ratio of R&D to sales American companies invested 4.5 % of sales revenues in R&D, compared with 4.0 % for Japanese and 3.3 % for European companies

1

2nd group witnesses no correlation: soaps and detergents; medical and surgical equipment; telecommunications 3rd group incredibly sees a negative correlation: computer hardware, software; semiconductors

2 http://www.innovation.gov.uk/rd_scoreboard/index.asp

Chapter 2 R&D Projects

Table 2-1 shows the top 15 companies ranked by R&D intensity These firms are amongst the 500 largest companies in the world by market capitalisation As can be inferred from the table, the US is strongly represented in the three big R&D-intensive industry sectors: pharmaceuticals, IT hardware and software In contrast, Europe is relatively weak in IT and related fields, while Asia lacks a vibrant pharmaceutical sector

It should be noted that an R&D intensity of over 15% is considered remarkable and companies under this category usually gain a reputation for being high technology companies

Generally, high-tech firms prosper in markets of extreme demands, such as medicine, scientific instruments, safety-critical mechanisms (aircraft) or high technology military armaments The extreme needs justify the high risk of failure and consequently high gross margins from 60% to 90% of revenues Most industrial companies however get only 40% of revenues

Table 2-1: Companies with the highest R&D intensity in the FT Global 500

Sector Growth of

Market Cap

Country

1 Computer Associates(7)* 21.5% £0.4bn +8% S +9% USA

2 Electronic Arts, (20) 20.2% £0.3bn +24% S +22% USA

3 Analog Devices (6) 19.4% £0.3bn +14% H -8% USA

4 Eli Lilly(12) 19.4% £1.4bn +15% P%< -17% USA

5 Schering-Plough (17) 19.4% £0.8bn +9% P +13% USA

6 Amgen (10) 19.2% £1.1bn +23% P +38% USA

7 Adobe Systems (n) 18.7% £0.2bn +12% S +31% USA

8 Juniper Networks (n) 17.8% £0.1bn +35% H +9% USA

9 AstraZeneca (13) 17.7% £2.0bn +10% P +2% UK

10 Merck (34) 17.5% £2.1bn +26% P -34% USA

11 Genzyme (12) 17.2% £0.2bn +20% P +56% USA

Chapter 2 R&D Projects

11

13 Ericsson (4) 16.7% £1.7bn -25% H +37% Sweden

14 ST Microelectronics (24) 16.5% £0.8bn +25% H -8% Netherlands

15 Roche (28) 16.3% £2.3bn +7% P +42% Switzerland

* Position in the equivalent list from the 2004 R&D Scoreboard (n = not in FT Global 500)

† S = Software & computer services; H = IT hardware, P = Pharmaceuticals & biotechnology

Technology investments are interesting both from the management and the financial perspectives If aligned with corporate strategy, R&D projects often grant the possibility of pursuing an avenue in several months or a couple of years Each successful innovation may be used as a building block for further R&D efforts, enabling creation of sustainable competitive advantage through a cohesive R&D program that blends and builds upon previous results

From the capital budgeting point of view however, the only certain aspect of R&D projects is the investment sum which unfortunately can be rather significant The potential payoff may be high, but that is contingent on the combination of many factors, including (internal) technical maturity, (external) market competition, and the extent of innovation (radical/ incremental)

Figure 2-2 pinpoints the position of R&D in a risk-payoff matrix

Figure 2-2: Risk-payoff matrix showing position of R&D

Chapter 2 R&D Projects

2.4 R&D Project Lifecycle

An R&D project can be divided into several stages of differing objectives and concerns, and thus each stage analysed separately Jensen and Warren (2001) identified three stages as depicted in Figure 2-3:

1 Scientific research – Basic, Applied, Prototype/pilot plant

2 Development : the second phase is the product development option, which, for convenience, is modelled as a perpetuity

3 Commercial introduction

Figure 2-3: Lifecycle of an R&D Project as proposed by Jensen & Warren (2001)

(a) Scientific research

Following the UK Department of Trade and Industry and the Secretary of State, Scientific research is defined in the statute as "any activities in the fields of natural or applied science for the extension of knowledge" Activities are scientific research if they involve:

- the application of new scientific principles in an existing area of research, or

- the application of existing principles in a new area of research

Launch Scientific

research Development

Chapter 2 R&D Projects

13

Research can be further categorised into basic or applied research, the first more for knowledge formation, and the second for knowledge application At the corporate level, applied research tends to be the dominant interest, in conformance to the accountability

of funds Essentially, this is an exploration phase that determines the viability of the project Within this stage of initial research possibly lie several sub-decisions as investment funds are allocated but not paid out at one instant

(b) Development

Development involves the methods for proceeding as well as all the stages of design and manufacturing required to yield a working product

Again from the UK HM Revenue & Customs definitions, “Experimental Development

is systematic work, drawing on existing knowledge gained from research and/or practical experience, that is directed to producing new materials, products and devices;

to installing new processes, systems and services; or to improving substantially those already produced or installed which will lead to an extension of knowledge.”

(c) Launch

After successful development, if the market condition is deemed favourable for the introduction of the product, the company will proceed into this commercialisation stage

Chapter 2 R&D Projects

The risks of R&D projects are typically characterised by the huge investment outlay and a long timeframe to potential results As illustrated in Figure 2-4, the factors contributing to the uncertainty of an R&D program include technical, marketing and political factors, as well as the types and stages of the innovation

Figure 2-4: Decision criteria of R&D projects

2.5.1 Technical Uncertainty

A primary and apparent concern in R&D project selection is the inherent technical feasibility of the projects If the required competence and resources are not available, the project cannot exist since it would only be doomed to fail Other considerations may include the commitment of the research and project team as well as the existence

of a product champion

Socio-political factors Technical

uncertainty

Market uncertainty

Economic feasibility

Project feasibility

Chapter 2 R&D Projects

15

It is worthwhile to distinguish the factors at the upstream and downstream stages of a project These can be viewed as constraints and intangible output respectively, the latter of which can be translated as the intrinsic worth of the research For example, the development of a particular technology knowledge, potential for strategic positioning, but is not a tangible output of the project

In Figure 2-5, we see the position of the various technical considerations on a timeline

Figure 2-5: Technical uncertainties arising at different stages of the R&D project

2.5.2 Market Uncertainty

Emerging from the research stage, a technically successful product may still end up in

a commercial failure if it fails to overcome the market hurdle From Brockhoff and Chakrabarti (1988), it has been shown that while the typical rates of technical success exceed 50%, that of commercial success are less than 50% Clearly then, market factors cannot be neglected in R&D project selection

Market factors come into play at the downstream stages of R&D projects and thus involve high uncertainty In the earlier stages, it is more difficult to define the market

as the research team seeks to identify potential demands Decision makers need to anticipate the future market dynamics when considering the factors like:

Timeline

Constraints (resources)

Intangible output (learning / indirect discovery)

Chapter 2 R&D Projects

- Degree and nature of competition: time to market, substitute products

- Market size: number of potential users

- Market success: proportion of adoption by potential users

- Product life cycle

- Availability of raw materials

Indeed, as the R&D project develops, there are varying degrees of concerns with regards to the aforementioned technical and marketing uncertainties However, as depicted in Figure 2-6, we see the impact of the gradual shift from technical to marketing concerns as the R&D proceeds towards commercial launch

introduction Technical

Having examined and understood the importance of R&D projects, the different stages

of the projects and the uncertainties involved at separate stages, we shall next move on

to review some of the tools that could be used to evaluate them

17

“It is always wise to look ahead but difficult to look further than you can see.”

Winston Churchill

Observer, Sayings of the Week , July 27, 1952

R&D projects are typically characterised by huge investment outlay and a long timeframe to potential results The inherent technical uncertainty of the process, as well as the inevitable market uncertainty of the unrealised research product in a future point in time, makes valuation elusive Since only the investment costs are certain, a risk-averse valuation would tend to produce negative net present values and thus discourage calculated risk-taking

3.1 Discounted Cash Flow

Traditionally, the Discounted Cash Flow (DCF) method is used to evaluate the attractiveness of an investment opportunity by estimating the value of the project, and this tool can also be used to value companies DCF analysis uses future cash flow projections and discounts them to arrive at a present value The weighted average cost

of capital (WACC) is often used as the discount rate, but much work has also been done on the choice of this factor

Chapter 3 R&D Project Valuation Tools

If the present value of the potential gains is significantly higher than that of the investment cost (Net Present Value >> 0), the opportunity may be potentially lucrative and there should be no hesitation in giving corporate approval

r

CF DCF

1 (1 )

where CF i = Cash Flow of year i

r = discount factor (WACC)

DCF models are powerful and simple to use and understand but they do have shortcomings Being a mere mechanical valuation tool, it is subject to the axiom

"garbage in, garbage out" Small changes in inputs can result in large changes in the NPV and reverse the recommendation With the advance into IT age, computer simulations have been introduced and integrated into the model Scenario and Sensitivity analysis then serve to present a more complete picture of the investment opportunity

However, decision makers cannot rely primarily on this static approach to select projects In fact, in the name of corporate strategy, many projects have been given the green signal despite their low NPVs A scrutiny of the DCF analysis reveals its applicability to “cash cow” investments with independent point considerations that command relatively low flexibility, while exposing its unsuitability to value risky opportunities where there appear to be two extra aspects neglected by the analysis:

Chapter 3 R&D Project Valuation Tools

Figure 3-1 gives a simplified illustration of the DA process

Figure 3-1: Abstraction of DA process

Decision Analysis Process

Choice (what we can do)

Information (what we know)

Preferences (what we want)

Inputs

Foundation

Decision Theory – rationality in simple yet uncertain situations

Systems Modelling Methodology – treatment of complexity and dynamics

Output Performance measures

Consequence variables

Environment: Uncertain, Complex, Dynamic,

Competitive, Finite

Chapter 3 R&D Project Valuation Tools

3.2.1 Intuition and Reasoning in DA

DA introduces a paradigm shift from avoiding uncertainty to optimising in the face of uncertainty An immediate visible merit of the decision framework was the use of Decision trees which make multi-stage decisions more natural and logical While the initial idea was to come up with a framework that can quantify risk and time preferences, this method does allow other strategic concerns, such as the option to abandon and the option to defer, as visible in Figure 3-2 Furthermore, the value of the options can be upper-bounded by the value of free perfect information

Figure 3-2: Decision factors of R&D projects

Yet, upon reflection, the guiding intuition in DA still treats uncertainty as undesirable The coinage of “Value of Information (VI)” implies the preference for decision-making under deterministic scenarios This is logical as we will reason with the aid of the illustration of the decision tree in Figure 3-3:

Chapter 3 R&D Project Valuation Tools

21

Figure 3-3: Decision Tree differentiating risk probabilities and risk impacts

In fact, VI is the value of resolving the uncertainty of having various scenarios, U and

D In other words, VI points out the value of having a deterministic p In the upcoming section on Real Option, we will further discuss the lacking dimension of this treatment

of uncertainty

3.2.2 Issue of Subjective probabilities

Undeniably, Decision trees are particularly useful to model and analyse complex sequential investment decisions Nevertheless, there remains one main implementation hurdle while a fresh problem is being introduced The first issue concerns the determination of an appropriate discount rate to roll back the tree The second relates

to the use of subjective probabilities

While it is meaningful to use probability as an indicator of success, the numerical value can only be either assigned by experts or estimated from past projects However,

as R&D aims for breakthrough or innovation, there is no historical data to justifiably quantify the technical uncertainty of a new product

U

D

p

1-p

Chapter 3 R&D Project Valuation Tools Furthermore, the use of subjective probabilities also calls into question its reliability Statistical surveys had been carried out, and as Meadows (1968) found out, the initial estimates made at project outsets were “not accurate enough to be employed in project evaluation formulas.” Similarly, Souder (1978, 1969) found significance that initial estimates of projects’ success probabilities are not reliable, but those made at a later stage of the projects are highly reliable Thus, it can be concluded that estimates of success probabilities may not be very useful in project selection, but can be otherwise useful at subsequent stages of the project lifecycle

Notwithstanding the criticism on subjective probabilities however, Drzik (1996) proposed a gradual shift to judgmental methods as markets liquidity decreases He suggested that different risks be positioned along a liquidity continuum: from smooth, choppy, icy to frozen As markets become less liquid, we should place increasing emphasis on deferred compensation

Thus, even though subjective probabilities may not be reliable due to the long time projection of cost and profit estimation, this very cause explains that the flaw is inevitable and that the flawed approach might well be the best we can have

Chapter 3 R&D Project Valuation Tools

This can be carried out by drawing (pseudo-)random numbers to determine the outcomes and payoffs The procedure is repeated for a statistically large number of times to ensure validity of the results For multi-stage R&D projects, simulation can still be useful but a limitation is that it does not permit interaction with the analyst Decisions to be made at certain milestones need to be pre-programmed into the simulation

In relation to real option thinking, Myers (1976) reminds that managerial responses to contingencies can void tails of output distributions Therefore, it may be advisable to focus on the range of values covered by the central 80% of the probability distribution

as generated using simulation

Chapter 3 R&D Project Valuation Tools

3.4 Real Options Approach

A few years after the introduction of the DA methodology, another group of researchers who dabbles with financial option pricing came up with another tool that aims to simplify project valuation and reconcile corporate strategy and finance As Myers (1987) put it, “smart managers do not accept positive (or negative) NPVs unless they can explain them.” Indeed, managers recognise the insufficiency of standard valuation tools in the face of uncertainty as they always turn out low or negative NPVs for investments that align with strategic corporate vision There must then be some implicit value hidden in the uncertainty

3.4.1 Intuition and Reasoning in Real Options Analysis

As mentioned earlier, instead of the traditional risk avoidance, DA adopts a new notion

of “optimising in the face of uncertainty” This is extended further to a new paradigm

of “uncovering option value hidden in uncertainty” with the RO approach Resuming the discussion on treatment of uncertainty, the RO analysis focuses on the uncertainty

of the scenario itself instead of the probability of having the scenario Referring back

to Figure 3-3, while DA deals with the uncertainty of p, RO cash in on the uncertainty

of the value of U and D

RO dissects uncertainty and reminds that uncertainty does not equal to risk On the contrary, with the presence of management flexibility in sequential investment decisions, the investment opportunities present a set of options that magnify upside

Chapter 3 R&D Project Valuation Tools

25

Conceivably, options add value as they provide opportunities to take advantage of an uncertain situation as the uncertainty resolves itself over time

Figure 3-4: Uncertainty “dissected”

3.4.2 Understanding Financial Options

As ROs is essentially the application of option pricing to non-financial valuation, it would be useful to introduce some terms related to financial options

In financial terms, an option is a contract giving the buyer the right but not the obligation to buy or sell an underlying asset at a pre-determined price on or before a certain date Options are derivatives because they derive their value from an underlying asset

• An option to buy an asset is a call option, while the option to sell an asset is a put option There are thus four types of participants in options markets, namely buyers of calls, sellers of calls, buyers of puts, and sellers of puts Buyers are often referred to as holders and sellers are also referred to as writers

Upside potential

Downside risk

Uncertainty

Chapter 3 R&D Project Valuation Tools

• The price at which an underlying stock can be purchased or sold is called the strike or exercise price

• The date before or on which the option is valid is called the maturity or expiry date Long term options are known as LEAPS If the option can only be exercised on the maturity date, they are called European Options The other type of options is American options which can be exercised anytime before or

on the expiry date

• An option is said to be in-the-money if an immediate profit can be made by exercising the option If a loss would be incurred from its immediate exercise, the option is out-of-the-money Options that do not offer any deviation are at-the-money These characterise the intrinsic value of the options, but the second component – the time value – is not as readily evaluated

• The total cost of an option is called the premium, which is determined by factors including the stock price, strike price, and time remaining until expiration

A stock option contract represents 100 shares of the underlying stock, and investors use them both to speculate and hedge risk

For an in-depth understanding of the mainstream theory and the seminal works of Financial Option Pricing, the reader can refer to Appendix A Financial Option Pricing

3.4.3 Real Options Defined

The term “real options” was coined by Stewart Myers in 1977 as “opportunities to purchase real assets on possibly favorable terms.” It referred to the application of