On fuzzy real option valuation

Transcription

1 On fuzzy real option valuation Supported by the Waeno project TEKES 40682/99. Christer Carlsson Institute for Advanced Management Systems Research, Robert Fullér Department of OR, Eötvös Loránd University and Institute for Advanced Management Systems Research, TUCS Turku Centre for Computer Science TUCS Technical Report No 367 October 2000 ISBN ISSN

2 Abstract Financial options are known from the financial world where they represent the right to buy or sell a financial value (mostly a stock) for a predetermined price (the exercise price), without having the obligation to do so. Real options in option thinking are based on the same principals as financial options. In real options, the options involve real assets as opposed to financial ones. To have a real option means to have the possibility for a certain period to either choose for or against something, without binding oneself up front. Real options are valued (as financial options), which is quite different with from discounted cashflow investment approaches. The real option rule is that one should invest today only if the net present value is high enough to compensate for giving up the value of the option to wait. Because the option to invest loses its value when the investment is irreversibly made, this loss is an opportunity cost of investing. However, the pure (probabilistic) real option rule characterizes the present value of expected cash flows and the expected costs by a single number, which is not realistic in many cases. In this paper we consider the real option rule in a more realistic setting, namely, when the present values of expected cash flows and expected costs are estimated by trapezoidal fuzzy numbers. Keywords: Option pricing; Possibilistic mean value; Possibility distributions; Possibilistic variance TUCS Research Group Institute for Advanced Management Systems Research

3 1 Possibilistic mean value and variance of fuzzy numbers Fuzzy sets were intorduced by Zadeh [6] in 1965 to represent/manipulate data and information possessing nonstatistical uncertainties. Let X be a nonempty set. A fuzzy set A in X is characterized by its membership function µ A : X [0,1], and µ A (x) is interpreted as the degree of membership of element x in fuzzy set A for each x X. The value zero is used to represent complete non-membership, the value one is used to represent complete membership, and values in between are used to represent intermediate degrees of membership. Frequently we will write simply A(x) instead of µ A (x). The family of all fuzzy (sub)sets in X is denoted by F(X). A fuzzy subset A of a classical set X is called normal if there exists an x X such that A(x) = 1. Otherwise A is subnormal. An α-level set of a fuzzy set A of X is a non-fuzzy set denoted by [A] α and is defined by [A] α = { {t X A(t) α} if α > 0, cl(suppa) if α = 0, where cl(suppa) denotes the closure of the support of A. A fuzzy set A of X is called convex if [A] α is a convex subset of X, α [0,1]. In many situations people are only able to characterize numeric information imprecisely. For example, people use terms such as, about $5,000, near zero, or essentially bigger than $5,000. These are examples of what are called fuzzy numbers. Using the theory of fuzzy subsets we can represent these fuzzy numbers as fuzzy subsets of the set of real numbers. More exactly, Definition 1.1. A fuzzy number A is a fuzzy set of the real line with a normal, (fuzzy) convex and continuous membership function of bounded support. The family of fuzzy numbers will be denoted by F. Let A be a fuzzy number. Then [A] γ is a closed convex (compact) subset of R for all γ [0,1]. Let us introduce the notations a 1 (γ) = min[a] γ, a 2 (γ) = max[a] γ In other words, a 1 (γ) denotes the left-hand side and a 2 (γ) denotes the right-hand side of the γ-cut. We shall use the notation [A] γ = [a 1 (γ), a 2 (γ)]. The support of A is the open interval (a 1 (0), a 2 (0)). Fuzzy numbers can also be considered as possibility distributions [3]. If A F is a fuzzy number and x R a real number then A(x) can be interpreted as the degree of possiblity of the statement x is A. 1

4 1 a-α a b b+β Figure 1: Trapezoidal fuzzy number. Definition 1.2. A fuzzy set A F is called trapezoidal fuzzy number with core [a, b], left width α and right width β if its membership function has the following form 1 a t if a α t a α 1 if a t b A(t) = 1 t b if a t b + β β 0 otherwise and we use the notation A = (a, b, α, β). It can easily be shown that [A] γ = [a (1 γ)α, b + (1 γ)β], γ [0,1]. The support of A is (a α, b + β). A trapezoidal fuzzy number with core [a, b] may be seen as a context-dependent description (α and β define the context) of the property the value of a real variable is approximately in [a, b]. If A(t) = 1 then t belongs to A with degree of membership one (i.e. a t b), and if A(t) = 0 then t belongs to A with degree of membership zero (i.e. t / (a α, b + β), t is too far from [a, b]), and finally if 0 < A(t) < 1 then t belongs to A with an intermediate degree of membership (i.e. t is close enough to [a, b]). In a possibilistic setting A(t), t R, can be interpreted as the degree of possibility of the statement t is approximately in [a, b]. Let [A] γ = [a 1 (γ), a 2 (γ)] and [B] γ = [b 1 (γ), b 2 (γ)] be fuzzy numbers and let λ R be a real number. Using the extension principle we can verify the following rules for addition and scalar muliplication of fuzzy numbers [A + B] γ = [a 1 (γ) + b 1 (γ), a 2 (γ) + b 2 (γ)], [λa] γ = λ[a] γ. (1) Specially, if A = (a, b, α, β) and B = (a, b, α, β ) are fuzzy numbers of trapezopidal form, and λ > 0 and µ < 0 are real numbers then we get A + B = (a + a, b + b, α + α, β + β ), λa = (λa, λb, λα, λβ), A B = (a b, b a, α + β, β + α ), µa = (µb, µa, µ β, µ α), (2) 2

5 Let A F be a fuzzy number with [A] γ = [a 1 (γ), a 2 (γ)], γ [0,1]. In [2] we introduced the (crisp) possibilistic mean (or expected) value of A as E(A) = 1 = γ(a 1 (γ) + a 2 (γ))dγ γ a1(γ) + a 2 (γ) dγ 2 1, γ dγ i.e., E(A) is nothing else but the level-weighted average of the arithmetic means of all γ-level sets, that is, the weight of the arithmetic mean of a 1 (γ) and a 2 (γ) is just γ. It can easily be proved that E : F R is a linear function (with respect to operations (1)). In [2] we also introduce the (possibilistic) variance of A F as ( 1 [a1 σ 2 (γ) + a 2 (γ) 2 [ ] ) a1 (γ) + a 2 (γ) 2 (A) = γ a 1 (γ)] + a 2 (γ) dγ 2 2 = γ ( a 2 (γ) a 1 (γ) ) 2 dγ. i.e. the possibilistic variance of A is defined as the expected value of the squared deviations between the arithmetic mean and the endpoints of its level sets, It is easy to see that if A = (a, b, α, β) is a trapezoidal fuzzy number then 0 and E(A) = 1 0 γ[a (1 γ)α + b + (1 γ)β]dγ = a + b + β α 2 6. σ 2 (b a)2 (b a)(α + β) (α + β)2 (A) = Probabilistic real option valuation Options are known from the financial world where they represent the right to buy or sell a financial value, mostly a stock, for a predetermined price (the exercise price), without having the obligation to do so. The actual selling or buying of the underlying value for the predetermined price is called exercising your option. One would only exercise the option if the underlying value is higher than the exercise price in case of a call option (the right to buy) or lower than the exercise prise in the case of a put option (the right to sell). 3

6 In 1973 Black and Scholes [1] made a major breakthrough by deriving a differential equation that must be satisfied by the price of any derivative security dependent on a non-dividend paying stock. For risk-neutral investors the Black-Scholes pricing formula for a call option is C 0 = S 0 N(d 1 ) Xe rt N(d 2 ), where d 1 = ln(s 0/X) + (r + σ 2 /2)T σ T, d 2 = d 1 σ T, and where C 0 = current call option value S 0 = current stock price N(d) = the probability that a random draw from a standard normal distribution will be less than d. X = exercise price r = the annualized continuously compounded rate on a safe asset with the same maturity as the expiration of the option, which is to be distinguished from r f, the discrete period interest rate, ln(1 + r) T = time to maturity of option, in years σ = standard deviation In 1973 Merton [5] extended the Black-Scholes option pricing formula to dividendspaying stocks as C 0 = S 0 e δt N(d 1 ) Xe rt N(d 2 ) (3) where, d 1 = ln(s 0/X) + (r δ + σ 2 /2)T σ T, d 2 = d 1 σ T where δ denotes the dividends payed out during the life-time of the option. Real options in option thinking are based on the same principals as financial options. In real options, the options involve real assets as opposed to financial ones. To have a real option means to have the possibility for a certain period to either choose for or against something, without binding oneself up front. For example, owning a power plant gives a utility the opportunity, but not the obligation, to produce electricity at some later date. Real options can be valued using the analogue option theories that have been developed for financial options, which is quite different with from traditional discounted cashflow investment approaches. In traditional investment approaches investments activities or projects are often seen as now or never and the main question is whether to go ahead with an investment yes or no. Formulated in this way it is very hard to make a decision when there is uncertainty about the exact outcome of the investment. To help with these tough decisions 4

7 valuation methods as Net Present Value (NPV) or Discounted Cash Flow (DCF) have been developed. And since these methods ignore the value of flexibility and discount heavily for external uncertainty involved, many interesting and innovative activities and projects are cancelled because of the uncertainties. However, only a few projects are now or never. Often it is possible to delay, modify or split up the project in strategic components which generate important learning effects (and therefore reduce uncertainty). And in those cases option thinking can help. The new rule, derived from option pricing theory (3), is that you should invest today only if the net present value is high enough to compensate for giving up the value of the option to wait. Because the option to invest loses its value when the investment is irreversibly made, this loss is an opportunity cost of investing. Following Leslie and Michaels [4] we will compute the value of a real option by ROV = S 0 e δt N(d 1 ) Xe rt N(d 2 ) where, and where d 1 = ln(s 0/X) + (r δ + σ 2 /2)T σ T, d 2 = d 1 σ T ROV = current real option value S 0 = present value of expected cash flows N(d) = the probability that a random draw from a standard normal distribution will be less than d. X = (nominal) value of fixed costs r = the annualized continuously compounded rate on a safe asset T = time to maturity of option, in years σ = uncertainty of expected cash flows δ = value lost over the duration of the option We illustrate the principial difference between the traditional (passive) NPV decision rule and the (active) real option approach by an example quoted from [4]:... another oil company has the opportunity to acquire a five-year licence on block. When developed, the block is expected to yield 50 million barrels of oil. The current price of a barell of oil from this field is $10 and the present value of the development costs is $600 million. Thus the NPV of the project opportunity is 50 million $10 - $600 million = -$100 million. Faced with this valuation, the company would obviously pass up the opportunity. But what would option valuation make of the same case? To begin with, such a valuation would recognize the importance of uncertainty, which the NPV analysis effectively assumes away. There 5

8 are two major sources of uncertainty affecting the value of the block: the quantity of the price of the oil.... Assume for the sake of argument that these two sources of uncertainty jointly result in a 30 percent standard deviation (σ) around the growth rate of the value of operating cash inflows. Holding the option also obliges one to incur the annual fixed costs of keeping the reserve active - let us say, $15 million. This represents a dividend-like payout of three percent (i.e. 15/500) of the value of the assets. We already know that the duration of the option, T, is five years and the risk-free rate, r, is 5 percent, leading us to estimate option value at ROV = 600 e e = $251 million - $151 million = $100 million. It should be noted that the fact that real options are like financial options does not mean that they are the same. Real options are concerned about strategic decisions of a company, where degrees of freedom are limited to the capabilities of the company. In these strategic decisions different s takeholders play a role, especially if the resources needed for an investment are significant and thereby the continuity of the company is at stake. Real options therefore, always need to be seen in the larger context of the company, whereas financial options can be used freely and independently. 3 A hybrid approach to real option valuation Usually, the present value of expected cash flows can not be be characterized by a single number. We can, however, estimate the present value of expected cash flows by using a trapezoidal possibility distribution of the form S 0 = (s 1, s 2, α, β) i.e. the most possible values of the present value of expected cash flows lie in the interval [s 1, s 2 ] (which is the core of the trapezoidal fuzzy number S 0 ), and (s 2 +β) is the upward potential and (s 1 α) is the downward potential for the present value of expected cash flows. In a similar manner we can estimate the expected costs by using a trapezoidal possibility distribution of the form X = (x 1, x 2, α, β ), i.e. the most possible values of expected cost lie in the interval [x 1, x 2 ] (which is the core of the trapezoidal fuzzy number X), and (x 2 + β ) is the upward potential and (x 1 α ) is the downward potential for expected costs. 6

9 1 s -α s +β 1 s 1 s 2 2 Figure 2: The possibility distribution of present values of expected cash flow. In these circumstances we suggest the use of the following formula for computing fuzzy real option values where, FROV = S 0 e δt N(d 1 ) Xe rt N(d 2 ), (4) d 1 = ln(e(s 0)/E(X)) + (r δ + σ 2 /2)T σ T, d 2 = d 1 σ T. E(S 0 ) denotes the possibilistic mean value of the present value of expected cash flows, E(X) stands for the the possibilistic mean value of expected costs and σ := σ(s 0 ) is the possibilistic variance of the present value expected cash flows. Using formulas (2) for arithmetic operations on trapezoidal fuzzy numbers we find FROV = (s 1, s 2, α, β)e δt N(d 1 ) (x 1, x 2, α, β )e rt N(d 2 ) = (s 1 e δt N(d 1 ) x 2 e rt N(d 2 ), s 2 e δt N(d 1 ) x 1 e rt N(d 2 ), (5) αe δt N(d 1 ) + β e rt N(d 2 ), βe δt N(d 1 ) + α e rt N(d 2 )). Example 3.1. Suppose we want to find a fuzzy real option value under the following assumptions, S 0 = ($400 million,$600 million,$150 million,$150 million), 1 S 0 millions E(S )= Figure 3: The possibility distribution of expected cash flows. r = 5% per year, T = 5 years, δ = 0.03 per year and X = ($550 million, $650 million, $50 million, $50 million), 7

10 1 X E(X)= millions Figure 4: The possibility distribution of expected costs. First calculate σ(s 0 ) = (s 2 s 1 ) 2 i.e. σ(s 0 ) = 30.8%, and 4 + (s 2 s 1 )(α + β) (α + β)2 + = $ million, 6 24 E(S 0 ) = s 1 + s 2 2 E(X) = x 1 + x β α 6 + β α 6 = $500 million, = $600 million, furthermore, ( ln(600/500) + ( ) /2) 5 N(d 1 ) = N = 0.589, 5 N(d 2 ) = Thus, from (4) we get that the fuzzy value of the real option is FROV = ($40.15 million, $ million, $88.56 million, $88.56 million). 1 FROV millions Figure 5: The possibility distribution of real option values. The expected value of FROV is $ million and its most possible values are bracketed by the interval [$40.15 million, $ million] the downward potential (i.e. the maximal possible loss) is $48.41 million, and the upward potential (i.e. the maximal possible gain) is $ million. 8

11 Suppose now that X 0 = (x 1, x 2, α, β ) denotes the present value of expected costs. Then the equation for fuzzy real option value (4) can be written in the following form FROV = S 0 e δt N(d 1 ) X 0 N(d 2 ) where, d 1 = ln(e(s 0)/E(X 0 )) + (r δ + σ 2 /2)T σ T In this case we get, d 2 = d 1 σ T FROV = (s 1, s 2, α, β)e δt N(d 1 ) (x 1, x 2, α, β )N(d 2 ) = (s 1 e δt N(d 1 ) x 2 N(d 2 ), s 2 e δt N(d 1 ) x 1 N(d 2 ), αe δt N(d 1 ) + β N(d 2 ), βe δt N(d 1 ) + α N(d 2 )). Example 3.2. Suppose we want to find a fuzzy real option value under the following assumptions, S 0 = ($400 million,$600 million,$150 million,$150 million), r = 5% per year, T = 5 years, δ = 0.03 per year and X 0 = ($550 million,$650 million,$50 million,$50 million), Then from (6) we get that the fuzzy value of the real option is FROV = ( $6.08 million, $ million, $92.12 million, $92.12 million). (6) FROV millions Figure 6: The possibility distribution of real option values. The expected value of FROV is $60.72 million and its most possible values are bracketed by the interval [ $6.08 million, $ million] the downward potential (i.e. the maximal possible loss) is $98.16 million, and the upward potential (i.e. the maximal possible gain) is $ million. Summary. In this paper we have considered the real option rule for capital investment decisions in a more realistic setting, namely, when the present values of expected cash flows and expected costs are estimated by trapezoidal fuzzy numbers. 9

12 References [1] F. Black and M. Scholes, The pricing of options and corporate liabilities, Journal of Political Economy, 81(1973) [2] C. Carlsson and R. Fullér, On possibilistic mean value and variance of fuzzy numbers, Fuzzy Sets and Systems, 2001 (to appear). [3] D.Dubois and H.Prade, Possibility Theory (Plenum Press, New York,1988). [4] K. J. Leslie and M. P. Michaels, The real power of real options, The McKinsey Quarterly, 3(1997) [5] R. Merton, Theory of rational option pricing, Bell Journal of Economics and Management Science, 4(1973) [6] L.A. Zadeh, Fuzzy Sets, Information and Control, 8(1965)

13

14 Turku Centre for Computer Science Lemminkäisenkatu 14 FIN Turku Finland University of Turku Department of Mathematical Sciences Åbo Akademi University Department of Computer Science Institute for Advanced Management Systems Research Turku School of Economics and Business Administration Institute of Information Systems Science