An Alternative Real Option Approach to R&D Project Assessment

Transcription

1 Asia Pacific Managemen Review 14(1) (2009) 1-13 Absrac An Alernaive Real Opion Approach o R&D Projec Assessmen Her-Jiun Sheu a, Lieh-Ming Luo b,* a Deparmen of Finance, Naional Chi Nan Universiy, Taiwan b Deparmen of Inernaional Trade & Finance, Fu Jen Caholic Universiy, Taiwan Acceped June An organizaion can ake diversifying acions o reduce he oal risks of projec porfolios. For an R&D projec, he overall level of risk could be reduced wih beer managemen and/or appropriae diversifying. In his paper, we develop an alernaive real opion approach o assess R&D projecs in consideraion of diversifying invesmens. The disinguishing feaure of our model seing is ha wo major uncerainies of R&D invesmen, i.e., he marke and echnological risks, are combined o generae a specific underlying process of R&D projecs. The ne presen value rule and radiional real opion mehods are applied in evaluaing R&D projecs. I is found ha he oucomes of he radiional real opion mehods would be affeced by evaluaors expecaion upon he marke or echnological oulook. The proposed mehod incorporaing diversificaion effec could lead o more reasonable assessmens for R&D projecs since he influence of R&D firms subjecive views could be resolved. The valuecreaion effec from diversifying acions has been examined hrough he numerical analyses. The resuls indicae ha beer managemen for R&D invesmen diversificaion will increase he projec value. Keywords: R&D projec, diversificaion, real opions 1. Inroducion 1 A echnology firm faces wo ypes of risk, similar o sock invesor: unique risk and marke risk. The former is associaed wih he aciviies of an individual business and is reducible hrough diversifying acions. However he laer ha correlaes o he indusry or echnology caegory canno be reduced hrough diversifying. As suggesed by Boer (2000, 2003), sysemaic reducion of unique risk could be applied as a ool for a successful R&D. Because risks characerisics for R&D projecs are usually idiosyncraic, hese risks canno be offse compleely by a duplicae porfolio of oher asses. Alhough all he risk of R&D projecs could no be hedged fully, firms sill aemp o reduce he oal level of risks due o he common risk-aversion characerisics. Boer (1998, 2000) poined ou ha a manufacurer can reduce he unique risk of R&D projecs by aking diversified invesmens such as, invesing in oher similar projecs and joining exploring syndicaes. Thus i is worhwhile o sudy how such diversifying acions could influence he fair valuaion of an R&D projec. The purpose of his aricle is o evaluae he value of R&D projecs wih he opion pricing approach adoping he diversifying acions. * Corresponding auhor. lmluo@mail.fju.edu.w 1

2 The real opion approach has been widely applied o he valuaion of R&D projecs since real R&D opions are usually embedded in projecs or processes. Researchers recognized ha he radiional DCF model (such as NPV rule) could no cach he value of managerial flexibiliy embedded in an R&D projec. Even some reasons for relucance in he employmen of real opion approach for an R&D projec are elucidaed (Harmann and Hassan, 2006; Paxson, 2001). The advanage of real opion mehod ha capures such flexibiliy value has been proposed in he lieraure (e.g. Morellec and Zhdanov, 2005; Boer, 2000, 2003; Jensen and Warren, 2001; Perliz e al., 1999). Especially for he echnological projecs wih higher uncerainies, he adopion of real opion mehod is explicily superior o he radiional DCF models (e.g. MacMillan e al., 2006; McGrah and MacMillan, 2000; Morris e al., 1991; Michell and Hamilon, 1988). For an R&D projec, he more uncerainies associaed wih he fuure revenue sreams of a projec invesmen, he higher value of he managerial flexibiliy would be (Dixi and Pindyck, 1994 ). Addiionally, Saniago and Vakili (2005) proposed a simple decision model o examine how he increase in uncerainy and variabiliy impac he overall value and he value of managemen flexibiliy in R&D projecs, giving a conclusion ha increased variabiliy in general increases he value of he projec. Typically, here are five ypes of managerial flexibiliy, namely, defer opion, abandonmen opion, expansion opion, conracion opion, and swiching opion (Trigeorgis, 1998). Wih respec o he applicaion of real opion heory in assessing R&D programs, many sudies uilized he Black-Scholes model (BS model) or Cox-Ross-Rubinsein model (CRR model) o evaluae R&D projecs (Bowman and Moskowiz, 2001; Carer and Edwards, 2001; Boer, 2000; 2003; Miller and Arikan, 2004; Lewis e al., 2004). However, some assumpions of he aforemenioned models should be relaxed for he valuaion of echnological projecs. For example, i is assumed in he BS model ha he rae of reurns of he underlying asse follows lognormal disribuion which is no he usual characerisics in an R&D conex. Angelis (2000) proposed an alernaive mehod o assess real R&D opion values relaxing he lognormal-disribuion assumpion. The no-arbirage condiion, which is anoher crucial assumpion in he BS model or in he CRR model, is ignored in he lieraure relaed o real R&D opion value. Theoreically, he real opion mehod is based on coningen claims analysis. I is required for his approach ha he projec s risk could be replicaed compleely by he porfolio of some oher raded asses (Dixi and Pindyck, 1994). Because he risks of R&D projecs usually are idiosyncraic, such replicaions could no be carried ou. Therefore, many sudies assumed ha firms have risk-neural aiude oward R&D projecs wih proper discoun raes (Dixi and Pindyck, 1994; Simh and Nau, 1995; Trigeorgis, 1998; Huchzermeier and Loch, 2001; Bollen, 1999). Alhough no all risks for an R&D projec could be duplicaed compleely wih oher asses, firms would ake diversifying acions o reduce he risk while implemening an R&D projec. Diversificaion effecs on an R&D projec has been emphasized in he sudies of Boer (2000, 2002, 2003), bu an appropriae evaluaion mehod incorporaing such he effec has no ye been explored. This sudy will aemp o fill his gap. The main purpose hus is o develop an alernaive real opion mehod for R&D projecs considering he diversifying acions. We would also compare he resuls wih he radiional NPV rule and real opion mehod. I is found ha, under he incorporaion of diversificaion effecs, he proposed valuaion mehod could yield more reasonable oucomes. Numerical analyses will be done o demonsrae he radiional real opion mehods migh over-evaluae or underesimae of R&D projecs values. I should also be noiced ha ha beer managemen for diversificaion could increase he R&D projecs values. The remainder of his paper is arranged as follows. The diversifying acions for an R&D projec will be discussed in Secion 2. Secion 3 is devoed o develop a new valuaion approach incorporaing he diversificaion effec for an R&D projec. A simplified risk model describing he R&D evoluion process is esablished in his secion. In secion 4, some 2

3 numerical examples are provided o illusrae he proposed model. The difference among he proposed mehod and he radiional ones are shown. Furhermore, special managemen implicaions are drawn as well. Finally, conclusive remarks are presened in he las secion. 2. Diversifying acion of R&D firm 2.1 The source of uncerainy An R&D firm means he one ha is carrying on R&D invesmens in order o increase is marke value. R&D firms will face diverse ypes of risks when proceeding wih an R&D invesmen. Various kinds of uncerainy sources regarding o R&D invesmens could be synhesized ino wo ypes, namely, echnological uncerainy and marke uncerainy (MacMillan and McGrah, 2002). The echnological uncerainy refers o wheher new echnology can work, wheher complemenary echnology could be ready in ime, and wha echnological sandards will dominae he marke. The marke uncerainy involves he issues of wheher here is enough poenial marke demand, or wheher produc price will change in he fuure. These uncerainies will influence poenial revenue sreams of R&D firms indirecly or direcly. Incorporaing boh he echnological and marke uncerainies, Raynor and Leroux (2004) have proposed a concepual model of how o offers a sraegic hinking for opimal allocaion of echnology invesmen. The echnological risk is referred o as unique one in he sudy of Boer (2000). The sudy emphasized he imporance of separaing unique and marke risks in applying opions heory o R&D projecs, and suggesed ha he unique risk could be analysed using esimaed probabiliies from hisorical daabase and he marke risk could be modelled by well-known financial ools. Since he unique risk and he echnological risk have similar characerisics, in his aricle, we use he echnological and marke risks as wo major kinds of risks for an R&D projec. 2.2 Diversificaion effec on R&D Markowiz (1952) and his followers have demonsraed ha invesors can ake diversifying acions o eliminae he unique risk. However, here is an obvious disincion beween financial and echnology porfolio, ha is liquidiy. Since mos financial asses could be raded wih limied ransacional fricion, diversificaion could be achieved over various classes of securiies. Diversified porfolios of echnology asses are much more challenging o assemble and o liquidae han porfolios of eiher securiies or businesses. I may be difficul o obain hrough buying and selling as claimed by Boer (2002). Neverheless, diversificaion in he R&D porfolio can help us in maximizing he value for a given level of risk. A firm should ake proper acions o miigae negaive influence caused by hose aforemenioned risks. Bu R&D firms have no way o dodge he marke risk which can be affeced by he produc price or he condiions of marke demand. However hey could diminish he echnological risk of an R&D projec hrough projec diversificaions. Taking an invesmen of exploraory oil well as an example, an enerprise could reduce he echnological risk by invesing in differen exploiaion case or by ways of joining drilling syndicaes (Boer 2000). Thus R&D firms could make a consrucive acion o abae he losses caused by unfavourable siuaions in pracice. Analogous o diversified invesmen of sock marke, a firm could inves simulaneously in he rival companies developing he same echnologies or aking R&D projecs wih subsiuive echnology. Tha is, diversifying acion means ha R&D firm can inves in a wide variey of projecs so ha exposure o he risk of a paricular R&D projec could be limied. In oher words, diversifying acions could be done by invesing in oher asses wih a payoff paern ha offse exposure o a paricular source of risk. When unfavourable siuaions do ake place, a firm could obain addiional rewards o compensae is losses. Magniude of such compensaory reward obained in unfavourable siuaions would 3

4 express he diversificaion effec. In oher words, hrough he invesmen in oher R&D projecs, an R&D organizaion is jus mimicking in holding anoher asse, and would ge compensaory reward while unfavourable siuaions occur. The ne effec of such diversifying behaviours would be regarded as obaining he asse, which is similar o insurance. So a firm could be compensaed for poenial losses hrough holding such asses. The diversificaion effec could be measured by magniude of he compensaory reward. 3. Valuaion framework 3.1 Se-up of risk model In his paper, boh marke and echnological risks are regarded as major R&D risk source. The marke risk involves he uncerainy of he fuure revenue sreams ha are relaed o projecs iself briefly. Technological risk may involve he exen in which he R&D projecs migh be failed, or in which he developed echnology migh be replaced by oher new dominan echnologies afer successful commercializaion. In order o consider he marke and echnological risks joinly, he evoluion of an R&D projec should be modelled as a sochasic process influenced simulaneously by hese wo uncerainy facors. In a ypical model seing, he marke risk always is reaed as he only facor describing he evoluion of R&D projec and he echnological risk is regarded as an independen variable which could be esimaed wih evaluaors judgmen or hisoric daa (e.g. Saniago and Vakili, 2005; Boer, 2003; Bowman and Moskowiz, 2001). In oher words, hese sudies assume ha he echnology risk is independen from marke one. However he wo risk facors may correlae posiively each oher. For example, a bull sock marke may simulae beer R&D performance because he projec paricipaors will receive greaer bonuses in his siuaion. Since boh risks in an R&D projec can occur simulaneously a he same ime, all saes occurring in he same period should be combined. I may be more suiable and comprehensive for he valuaion of R&D projecs o incorporae he fac ha an R&D firm may encouner simulaneously boh he uncerainies ino he framework seing. A simplified wo-phase model would be applied o describe such a process as exhibied in Figure 1. The firs phase is devoed o he R&D for new echnology. Assume ha here are wo saes of naure, i.e., success and failure, in he firs phase before an R&D projec is compleed. If being failed in he firs phase, he R&D projec will be disconinued, and is solvency value will be lef o V s (V s <V 0 ). On he oher hand, if he oucome is successful in his phase, a firm could inves furher for commercializaion. In he nex phase, a firm will encouner anoher risk ype, ha is, he commercialized echnology may be replaced by anoher new echnology. Two saes in his phase are herefore defined. One of he saes is unfavourable, in which he commercialized echnology is replaced, and he marke is dominaed by oher subsiuive echnologies. The oher one is favourable, in which he commercialized echnology can esablish produc sandards and dominae in he marke. Thus he disinguishing feaure of our model seing is ha wo major uncerainies of R&D invesmen are combined o generae a specific underlying process of R&D projecs. Such he seing would become more reasonable since he boh risks may no need o be reaed as ones ha ake place in differen periods. Alhough he model seing is more complex, i may more closely approximae he real world. The convenional reamen on building he marke risk for an R&D projec is analogous o financial opions. Typically, i is supposed ha price or rae of reurns of underlying asse follows a specific sochasic process. The opion value could be derived according o noarbirage pricing heory. However, R&D projecs usually do no have he corresponding underlying asses as he financial opions do. Thus he general reamens usually assume ha he fuure revenue or ne presen value (NPV) associaed wih he R&D program follows a specific sochasic process. A cerain linkage relaionship beween his evoluion process and 4

5 he marke condiion or he average indusrial sock price is also assumed. Therefore he sochasic process of he revenue sreams could be esablished according o he characerisics of indusrial sock price or marke demand. For example, i could be assumed ha here is a reasonable relaion beween he revenue volailiy of R&D projec and ha of indusrial sock price (Bowman and Moskowiz, 2001; Boer, 2003). Wih he hisorical daabase, we can esimae he growh rae and volailiy of revenue sreams for an R&D projec. By his way, a suiable sochasic process describing he marke risk could be esablished. In his research, i is assumed ha he marke risk of an R&D projec follows a Binomial ree. We will use σ o denoe he volailiy of he revenue of an R&D projec and δ, 1/δ o denoe he upward and downward mulipliers for he Binomial ree respecively. Because uncerainy comes from boh he marke and echnological risks, here are four possible saes a he end of he firs period, i.e. successful R&D in beer marke siuaion, failed R&D in beer marke siuaion, successful R&D in worse marke siuaion and failed R&D in worse marke siuaion. The iniial revenue associaed wih R&D projecs is supposed o be V 0. The four saes could be expressed respecively wih symbols as (δv 0, S), (V s, F), (V 0 /δ, S) and (V s, F), where he firs variable in parenheses represens he magniude of revenue; S and F denoes he success and failure of R&D respecively. If R&D is compleed successfully, a firm could choose o commercialize he new echnology or no o, depending on marke condiion a ha ime. Oherwise, his R&D projec would be erminaed. Once he new echnology commercialized decision is made, a firm mus pu furher in he invesmen cos. The firm will face oher risk associaed wih echnology replacemen in he second phase. If new superior echnologies appear and replace he exising echnology, he exising marke share would be affeced. We assume ha he marke revenue would be reduced down o a specified proporion of original scale, θ, if such he unfavourable siuaion occurs. Consequenly, here will be eigh saes a he end of he second period, as shown in Figure 1. They are (δ 2 V 0, D), (θδ 2 V 0, R), (V 0, D), (θδv 0, R), (V 0, D), (θv 0, R), (V 0 /δ 2, D) and (θv 0 /δ 2, R) respecively, wih he firs variable in parenheses denoing he magniude of revenues; D represening he siuaion in which he exising echnology could be dominan in he marke, and R denoing ha i is replaced by oher dominan echnologies. (δ 2 V 0, D) (θδ 2 V 0, R) (δv 0, S) (V 0, D) Saring poin (V S, F) (V 0 /δ, S) (V S, F) (θv 0, R) (V 0, D) (θv 0, R) (V 0 /δ 2, D) (θv 0 /δ 2, R) Figure 1: The simplified sochasic process of an R&D projec 5

6 3.2 Valuaion mehod Theoreical developmen of opion valuaion in financial economic has laid he foundaion for he exension of opion pricing approach o he evaluaion of invesmen projecs embedding some managerial flexibiliies. The radiional NPV mehod assumes an invesmen projec will be operaed coninuously unil he end of he pre-esimaed invesmen duraion. The valuaion crierion is based on he presen value of expeced fuure revenue sreams and invesmen cos inpu hrough discouning procedure wih a risk-adjused rae, i.e. NPV = ~ E[ CF I ] T = 0 (1 + radj ) where C ~ F is cash inflow in period, r adj is a discoun rae adjused for risks associaed wih he invesmen, T is expeced duraion of he invesmen, I is he required invesmen cos in period, and E[ ] is he expecaion operaor wih respec o subjecive esimaed probabiliy. The NPV rule, which suggess he invesmen opporuniy wih discouned cash flows exceeding discouned coss should be chosen, helps decision makers choose beween wo alernaives: accep (if NPV > 0) or rejec (if NPV < 0) an opporuniy. Among alernaive valuaion approaches proposed as remedies for he saic NPV approach, he mos prominen one is he decision ree analysis (DTA). DTA can help managemen o srucure invesmen decision-making by mapping ou feasible invesmen acions coningen on all saes in a hierarchical manner. Projec invesmen is modelled as a sequence of decisions for choosing among alernaive courses of acion, where he consequence of each alernaive acion depends on uncerain evens ha can be described probabilisically on he basis of pas informaion. The value of he invesmen projec is hen esimaed by discouning oucomes along opimal decision pah by he risk-adjused rae, i.e. (1) V = ~ E[ CF COST T = 0 (1 + radj ) Y ] (2) where E[ Y ] is expecaion operaor condiioned on ime. Whereas he convenional NPV rule assumes iniial invesmen sraegy is unchanged hrough he invesmen duraion, DTA recognizes ha he implici operaional flexibiliies are embedded in he projec and allows he iniial sraegy o be alered during he invesmen duraion. However his approach suffers he same difficuly as saic NPV when i comes o deermine he risk-adjused discoun rae. Subjecive probabiliy disribuion has o be adoped o achieve he adjusmen needed o compensae risk bearing, hen probably resuling in inconsisen invesmen valuaion (Copeland and Anikarov, 2001). Problems wih he saic NPV rule and he dynamic decision ree analysis can be easily solved wih coningen claims analyses (CCA). Theoreically, he opion pricing approach for financial insrumen is based on CCA. This approach requires ha he projec s risk is replicaed compleely by he porfolio of some oher raded asses as menioned in he firs secion. This mehod echnically also adops he decision ree analysis, in which invesmen involves a sequence of decisions of choosing among alernaive courses of acion, and each decision is reaed as an opion. However, i differs from DTA, in which he no-arbirage argumen adoped coningen claims analyses ensure invesmen valuaion is independen of evaluaor s risk preference. The risk-neural assumpion required in DTA can be relaxed. In paricular, he coningen claims approach uses he marke prices of exising asses o infer sae prices, and hence ge risk-neural probabiliies associaed wih every fuure node implici 6

7 in he decision ree. These risk-neural probabiliies are hen used o value cash flows from alernaive decisions and he opimal decision can be chosen. Thus, he value of he invesmen projec can be calculaed by discouning expeced cash flows ha derive from he opimal decision sraegy by he risk-free ineres rae, i.e. V = ~ Eˆ[ CF COST T = 0 (1 + r ) Y ] Where Eˆ[ Y ] is expecaion operaor wih respec o he risk-neural probabiliies condiioned on ime and r is he risk-free ineres rae. We will evaluae an R&D projec s value using CCA. Given he sochasic process of an R&D projec described in he previous subsecion, we could derive he risk-neural probabiliies of he sochasic process of an R&D projec applying he no-arbirage condiion. The deailed compuaion procedure for he risk-neural probabiliy is described below. Suppose ha he sock price follows Binomial process. Is volailiy is denoed by σ s (he sandard deviaion of he sock price), accompanied by upward-moving facor, u, and downward-moving facor, d, d = 1/u. For he underlying process esablished, here are consequenly hree basis asses a firm could choose o form is porfolio, namely, he indusrial sock porfolio, he compensaory reward derived from he diversificaion effec, and he risk-free asse. Under he no-arbirage condiion, he risk-neural probabiliies of he whole evoluion process of an R&D projec could be derived. According o he esimaed cash flows occurring a each sae, he fair value of an R&D projec could be achieved by backward procedure wih respec o he risk-neural probabiliies. Consider he i-h period in Figure 1 for i = 1~2. I is assumed ha a firm could obain reurn rae of compensaory reward, r i, for he i- h period while unfavourable siuaions occurring. The magniude of r i depends on he diversificaion effec, for i = 1~2. According o no-arbirage heory, he risk-neural probabiliies for i-h period, Q i = (q i1, q i2, q i3, q i4 ), mus saisfy he following equaion sysem: (q i1 + q i2 + q i3 + q i4 ) / ( 1 + r f ) = 1, (uq i1 + uq i2 + dq i3 + dq i4 ) / ( 1 + r f ) = 1, [(1 + r i ) q i2 + (1 + r i ) q i4 ] / ( 1 + r f ) = 1, wih r f as he risk-free ineres rae. The general soluion of his equaion sysem could be derived as Q * i = ((1 + r- d) /(u - d) -1 /(1 + r i ) + α i, 1 /(1+ r i )- α i,1- (1+ r f - d) /(u - d)- α i, α i ) wih max[0, 1 / (1 + r i ) - (r - d) / (u - d)] < α i < min[1 - (1 + r f - d) / (u - d), 1 / (1 + r i )] for i = 1~2. Obviously, he derived risk-neural probabiliy measure is no unique for each period. Making up Q i for i = 1~2, he risk-neural probabiliy se of he whole process of an R&D projec could be obained. Incorporaing he esimaed revenue sreams and cos inpus, he fair value of an R&D projec could be calculaed by (3) applying he backward procedure wih respecive o he risk-neural probabiliy se. However, since he risk-neural probabiliy se for his process seing is no unique, an R&D projec s value obained would be bounded by a range. We will choose he maximum of he inerval as he measuremen of projec s value. (3) 7

8 4. Numerical analyses 4.1 Illusraive example In order o illusrae and o provide a beer undersanding of he risk model, he scenario ha a firm will carry ou an R&D projec is assumed. The evoluion process of his R&D projec is simplified ino a wo-phase model, as shown in Figure 1. Since he goal of he numerical analyses is o yield comparaive properies among he proposed mehod and he radiional ones, his seing migh no oally fulfil he real world cases. The reamen employed here will no affec he resuls for generalized cases. For simpliciy, each phase will las for a year. Before calculaing he projec s value, i is necessary o choose relevan parameers appropriaely. In order o use appropriae numbers of parameers, we refer o numerical example employed by Boer (2003) and Faulkner (1996). Suppose ha iniial invesmen of $5 million is required, he projec will be compleed one year laer, and he oal commercializaion cos is abou $20 million in he nex phase. The revenue associaed wih he projec is assumed o be 30 million. Furhermore, he volailiy of average sock price wihin he same indusry (σ s ) is 25%. Wih a ypical reamen, we also assume a specific relaionship beween he volailiy of R&D revenue sreams and ha of he average indusrial sock price. The average sock price in indusrial secion usually reflecs he overall performance of he indusry. I is expeced ha an individual R&D projec has higher oal risks han he whole indusrial secion. The risk for an R&D projec aiming especially a emerging echnologies would be much higher. We will se he volailiy of an individual R&D projec σ o be 50%. I follows ha δ = e 0.5. In addiion, he probabiliies for all possible saes should be esimaed in advance. Since he esimaed probabiliies of differen sae will be influenced by subjecive expecaions of he firm, various aiudes regarding o he fuure prospec of he R&D projec including normal, opimisic and pessimisic will be assumed. For he normal case, he probabiliies ha he marke condiion urns o be beer and o be worse are se o 0.5 and 0.5 respecively. Besides, he probabiliies of he R&D s success and failure are assumed o be 0.5 and 0.5 respecively. Assuming ha marke risk and echnological risk are independen, i hen follows ha he probabiliies of four saes a he end of he firs period are 0.25, 0.25, 0.25 and 0.25 respecively. In he second phase, he probabiliies of echnology dominaion and echnology replacemen are supposed o be 0.7 and 0.3 respecively. I follows ha he probabiliies for he eigh saes in he second period are , , , , , , and respecively. If he commercialized echnology is replaced by oher echnologies in he second period, assume he marke share of producs will shrink o onefifh of he original scale wih he revenue reduced proporionally, i.e. θ = 0.2. Wih hese relevan parameers seing, he probabiliies and esimaed cash flows for each sae as shown in Figure 1 could be calculaed. For simpliciy, le he risk-free rae o be zero and he required reurn of he R&D projec based on is risk level be fixed o 15%. And assume ha solvency value (V s ) of he projec is also zero if failed R&D. We will evaluae he projec using radiional NPV rule and real opion mehod (i.e. decision ree mehod) wihou diversificaion effecs. Wih he required reurn as he discoun rae, NPV of he projec is million. When abandonmen is he only opion embedded in he R&D projec and he marke condiion is no good (or R&D is failed), he firm will erminae his projec. The esimaed ne value of he projecs (he value of projec minus all coss) is million. Consider he nex case wih boh he abandonmen and defer opions incorporaed, he firm will no make commercializaion ill he marke condiion urns ino be beer as well as he new echnology can dominae in he marke. In his case, he ne value of his projec increases o 0.56 million. 8

9 We will hen consider oher ypes of firm s aiudes oward he fuure oulook, namely opimisic and pessimisic. For he opimisic case, assume ha he probabiliies of he R&D s success and failure are 0.7 and 0.3 respecively based on he firm s view of he echnological oulook. On he oher hand, he associaed probabiliies for echnology oulook in he pessimisic case are se o 0.3 and 0.7 respecively. The probabiliies of he marke condiion urning o beer and o worse are se o 0.7 and 0.3 respecively for he opimisic case while hey are se o 0.3 and 0.7 respecively for he pessimisic case. In accordance wih aforemenioned valuaion procedure, we can also obain he ne values of he R&D projec embedding various opion ypes for he wo cases. The valuaion resuls of he projec are summarized in Figure NPV Abandon opion only Boh abandon and defer opions Pessimisic Normal Opimisic Figure 2: The valuaion resuls of he R&D projec by he radiional real opion mehod for various firms expecaion on R&D prospec. The radiional real opion mehods implicily assume ha an R&D firm is risk-neural wih respec o he esimaed probabiliies. Evaluaion of an R&D projec wih diversificaion effec will be explored uilizing coningen claims analysis. The risk-neural assumpion would be relaxed. The disinguishing characerisic of he proposed valuaion mehod is o consider he fac ha a firm could sill obain compensaory rewards hrough diversifying invesmens while unfavourable siuaions occur. The diversificaion effec is likened o possessing anoher asse, which is similar o geing insurances. The magniude of he compensaed reward would represen he diversificaion effec. Accordingly, we suppose ha if an R&D failed, he firm could obain he compensaory reward wih he reurn rae of 50% in he firs period. This reurn rae for he compensaory reward depends on he diversificaion effec. The greaer he diversificaion effec is, he higher he reurn rae will be. We consider he diversificaion effec varies from 20% o 100%. The seing of abou 50% for he middle degree is appropriae because i implies ha he firm can obain average reurn of 25% from he diversificaion effec in he normal case. According o he daabase of reurn hisory in American sock marke (1926~2003), he average reurn of small company sock is abou 18%. Since he risk level of a projec porfolio usually is much greaer han small company, he expeced reurn of a projec porfolio should be higher han ha of small company. Therefore he seing for he range of diversificaion effec may be raher reasonable and appropriae. Wih he similar argumen, i is supposed ha he firm could obain he rae of reurn of 50% in second period in case he new echnology is replaced by ohers. While incorporaing he diversificaion effec ino he valuaion of an R&D projecs, he hree real opion ypes, no opion, abandonmen opion only, as well as boh he abandonmen 9

10 and defer opions, would be considered. I is obvious ha he risk-neural probabiliy of he whole projecs evoluion would no be influenced by he firm's subjecive expecaion, i.e., he esimaed probabiliies. Therefore, no maer wha he firm s aiude oward he fuure marke prospec (normal, opimisic or pessimisic) is, he valuaion resuls for various cases would be he same. According o he calculaion procedure in he previous secion, he fair projec s value could be calculaed and he resuls are summarized in Figure Pessimisic Normal Opimisic No opion Abandon opion only Boh abandon and defer opions Figure 3: The valuaion resuls of he R&D projec by he proposed mehod wih he diversificaion effec for various firms expecaion on R&D prospec. The compensaory reward received by he firm in unfavourable siuaions depends on he diversificaion effec. Greaer managemen effeciveness for diversifying will creae higher compensaory reward. The impac of various degree of diversificaion effec on he projec value should also be sudied. For simpliciy, we sill assume ha he compensaory reward is he same for he firs period and he second period. The values of each real opion ype are calculaed when he degree of diversificaion effec varies from 20% o 100% wih a sep of 20%. In oher words, we le he corresponding compensaory rewards be 20%, 40%, 60%, 80% and 100% respecively. The evaluaion resuls for various degree of diversificaion effec are summarized in Figure No opion Abandon opion only Boh abandon and defer opions Lower Degree Higher Degree Figure 4: The valuaion resuls of he R&D projec by he proposed mehod for various degrees of diversificaion effec. 10

11 4.2 Discussion Referring o Figures 2 and 3, i is found ha he oucomes derived by he proposed approach are obviously differen from hose by he radiional real opion mehods (DTA). I is obvious ha he esimaed probabiliies have quie heavy influence on he valuaion oucomes by he radiional real opion mehods, as could be seen from Figure 2. The oucomes obained by he proposed mehod, which is based on no-arbirage condiion or CCA, acually are independen from he esimaed probabiliies. In oher words, he assumpion ha he R&D firm is risk-neural wih respec o he esimaed probabiliies is no required for he proposed model. The radiional real opion mehods usually assume ha evaluaor has riskneuraliy wih respec o he esimaed probabiliies. Thus is valuaion oucomes are correlaed significanly wih esimaed probabiliies. Take as an example, if he esimaed probabiliy ha marke condiion urns beer is higher, he projec s value calculaed will be higher. If he aiude oward an R&D prospec is more opimism, is esimaed value will be larger. Alhough he radiional real opion mehod is recognized as a superior valuaion ool han DCF model due o is capabiliy o capure he value of managerial flexibiliy, he aiude oward an R&D projec is no appropriaely considered. When a firm s aiude oward R&D projec is over-opimisic, he esimaed value of projecs would be overevaluaed. On he oher hand, while being over-pessimisic oward he fuure oulook, he firm may underesimae he projec s value. In addiion, he value-creaion effec from projec diversificaion has been examined hrough he numerical analysis. According o Figure 4, i should be noiced ha a projec s value obained by he proposed mehod is influenced by he magniude of compensaory reward. A greaer diversifying effec would generae higher value of an R&D projec. Moreover, managemen inefficiency for diversifying invesmen will decrease a projec s value. They are he wo sides of he same coin. In oher words, he diversifying acions of R&D firms can no only eliminae he risks of projec, bu also can enhance he projec value. As demonsraed by Boer (2002) In he echnological world, he risks are scarier. Who is o predic wheher Inerne conen will ener he home in 2010 predominanly via he ubiquious wised pair of phone wires (which have proved wih DSL [digial subscriber line] o be far more versaile han firs esimae), by cable, by saellie, by microwave, by fiber opics, or over elecrical power lines? An accurae forecas would be invaluable o he companies compeing in his markeplace and o privae invesors evaluaing elecommunicaions socks. Bu he oucome of he bale of broadband echnologies has no been deermined and will ye be influenced by housands of R&D programs, invesmen decisions, regulaory decisions, and oher facors ha have ye o play ou. False confidence is a risk, oo, and o couner his risk, company such as 3M has asked each of heir businesses o proacively idenify a pacing projec one ha has he capabiliy of changing he basis of compeiion and o fund i. The resuls imply ha R&D firms should aach grea imporance o he diversifying invesmen for R&D projecs in order o increase is marke value, which is consisen wih he viewpoin of Boer (2002). Even hough he resuls indicae ha he valuaion oucomes generaed by he proposed mehod is independen o he esimaed probabiliies, we need o make addiional esimaion for he compensaory rewards ha can occur in unfavourable siuaions while uilizing his mehod. Bu he diversificaion effec ha firms can echnically derive should be aken ino accoun for he valuaion of an R&D projec. Finally, i is also found ha no maer which kind of mehod is adoped, projec s value will increase wih he volailiy coefficien (σ). This resul coincides wih he opion pricing heory. In oher words, i implies ha a higher uncerainy for he fuure revenue sreams associaed wih an R&D projec will induce a higher managemen-flexibiliy value (Dixi and Pindyck, 1994 ). 11

12 5. Conclusive remarks An alernaive valuaion mehod incorporaing he diversificaion effec for R&D projecs has been developed in his paper. Since diversificaion could reduce he unique risk of an R&D projec, i may be necessary as well as imporan o ake he diversificaion effec ino accoun while a firm assesses an R&D projec. This facor acually can affec he rue value of projecs. This research provides a modified valuaion mehod wih he consideraion of noarbirage condiion required in opion pricing heory. Through he numerical analyses, i could be seen ha he proposed mehod could yield more reasonable and reliable oucomes. The numerical oucomes are also compared wih hose by he radiional real opion mehods. The radiional real-opions evaluaion assessmen is ap o be affeced by a firm s subjecive expecaion on fuure marke and echnology condiion. I migh cause over- or under- esimaion of a projec s value for some cases. Considering he diversificaion effec, he value of an R&D projec derived by our mehod will overcome he defici caused by firm's subjecive view. Besides, i is also found ha he diversificaion effec could acually affec he value of R&D projecs. I is shown ha more effecive managemen for diversified invesmens could increase he value of a projec. This managemen implicaion is consisen wih he prescripion alhough he diversifying acions would come a a cos, he combinaion of successively increasing cos and sharply reduced relaively risks will power value creaion by Boer (2003). References Angelis, D.I. (2000) Capuring he opion value of R&D. Research Technology Managemen, Jul-Aug, Boer, F.P. (1998) Traps, pifalls and snares in he valuaion of echnology. Research Technology Managemen, Sep-Oc, Boer, F.P. (2000) Valuaion of echnology using real opions. Research Technology Managemen, Jul-Aug, Boer, F.P. (2002) The Real Opions Soluion. John Wiley & Sons, New York. Boer, F.P. (2003) Risk-adjused valuaion of R&D projecs. Research Technology Managemen, Sep-Oc, Bollen, N.P.B. (1999) Real opion and produc life cycles. Managemen Science, 45(5), Bowman, E.H., Moskowiz, G.T. (2001) Real opion analysis and sraegic decision making. Organizaion Science, 12(6), Carer, R., Edward, D. (2001) Financial analysis exends managemen of R&D. Research Technology Managemen, Sep-Oc, Copeland, T., Anikarov, V. (2001) Real Opions. TEXERE, New York. Dixi, A K., Pindyck, R.S. (1994) Invesmen under Uncerainy. Princeon Press, Princeon, NJ. Faulkner, T.W. (1996) Applying opion hinking o R&D valuaion. Research Technology Managemen, May-June, Harmann, M., Hassan, A. (2006) Applicaion of real opions analysis for pharmaceuical R&D projec valuaion Empirical resuls from a survey. Research Policy, 35(3), Huchzermeier, A., Loch, C. (2001) Projec managemen under risk: Using he real opion approach o evaluae flexibiliy in R&D. Managemen Science, 47(1), Jensen, K., Warren, P. (2001) The use of opions heory o value research in he service secor. R&D Managemen, 31(2),

13 Lewis, N., Enke, D., Spurlock, D. (2004) Valuaion for he sraegic managemen of research and developmen projecs: he deferral opion. Engineering Managemen Journal,16(4), MacMillan, I.C., McGrah, R.G. (2002) Crafing R&D Projec Porfolios. Research Technology Managemen, Sep-Oc, MacMillan, I.C., Puen, A.B., McGrah, R.G., Thompson, J.D. (2006) Using real opions discipline for highly uncerainy echnology invesmens. Research Technology Managemen, Jan- Feb, Markowiz, H.M. (1952) Porfolio selecion. Journal of Finance, 7(1), McGrah. R.G., MacMillan, I.C. (2000) Assessing echnology projecs using real opions reasoning. Research Technology Managemen, Jul-Aug, Miller, K.D., Arikan, A.T. (2004) Technology search invesmen: Evoluionary, opion reasoning, and opion pricing approaches. Sraegy Managemen Journal, 25(5), Michell, G.R., Hamilon, W.F. (1988) Managemen R&D as a sraegic opion. Research Technology Managemen, May-June, Morris, P.A., Teisberg, E.O., Kolbe, L.A. (1991) When choosing R&D projecs, go wih long shos. Research Technology Managemen, Jan-Feb, Morellec, E., Zhdanov, A. (2005) The dynamics of mergers and acquisiions. Journal of Financial Economics, 77(3), Paxson, D.A. (2001) Inroducion o Real R&D opions. R&D Managemen, 31(2), Perliz, M., Peske, T., Schrank, R. (1999) Real opions valuaion: The new fronier in R&D projec evaluaion? R&D Managemen, 29(3), Raynor, M.E., Leroux, X. (2004) Sraegic flexibiliy in R&D. Research Technology Managemen, May-Jun, Saniago, L.P., Vakili, P. (2005) On he value of flexibiliy in R&D projecs. Managemen Science, 51(8), Smih, K.W., Nau, R.F. (1995) Valuing risky projecs: Opion pricing heory and decision analysis. Managemen Science, 41(5), Trigeorgis, L. (1998) Real Opion: Managerial Flexibiliy and Sraegy in Resource Allocaion. MIT Press, Cambridge, MA. 13