You are on page 1of 21

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/220210234

Using Binomial Decision Trees to Solve Real-


Option Valuation Problems

Article in Decision Analysis June 2005


DOI: 10.1287/deca.1050.0040 Source: DBLP

CITATIONS READS

152 396

3 authors:

Luiz Brando James S. Dyer


Pontifcia Universidade Catlica do Rio de Jane University of Texas at Austin
80 PUBLICATIONS 558 CITATIONS 126 PUBLICATIONS 5,744 CITATIONS

SEE PROFILE SEE PROFILE

Warren J. Hahn
University of Texas at Austin
24 PUBLICATIONS 353 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Copulas-based Approaches View project

Option Pricing View project

All content following this page was uploaded by Luiz Brando on 31 May 2014.

The user has requested enhancement of the downloaded file.


Decision Analysis informs
Vol. 2, No. 2, June 2005, pp. 6988
issn 1545-8490  eissn 1545-8504  05  0202  0069 doi 10.1287/deca.1050.0040
2005 INFORMS

Using Binomial Decision Trees to Solve


Real-Option Valuation Problems
Luiz E. Brando
IAG Business School, Pontifcia Universidade Catlica do Rio de Janeiro, Rio de Janeiro, RJ 22453-900, Brazil,
brandao@iag.puc-rio.br

James S. Dyer, Warren J. Hahn


McCombs School of Business, The University of Texas at Austin, Austin, Texas 78712
{jim.dyer@mccombs.utexas.edu, warren.hahn@phd.mccombs.utexas.edu}

T raditional decision analysis methods can provide an intuitive approach to valuing projects with managerial
exibility or real options. The discrete-time approach to real-option valuation has typically been imple-
mented in the nance literature using a binomial lattice framework. Instead, we use a binomial decision tree
with risk-neutral probabilities to approximate the uncertainty associated with the changes in the value of a
project over time. Both methods are based on the same principles, but we use dynamic programming to solve
the binomial decision tree, thereby providing a computationally intensive but simpler and more intuitive solu-
tion. This approach also provides greater exibility in the modeling of problems, including the ability to include
multiple underlying uncertainties and concurrent options with complex payoff characteristics.
Key words: decision analysis; real options; decision trees; binary approximations
History: Received on September 15, 2004. Accepted by Robert Clemen and Don Kleinmuntz on May 25, 2005,
after 2 revisions.

1. Introduction the cash ows and probabilities that give the cor-
Discounted cash ow (DCF) methods are commonly rect project values when discounted to each period
used for the valuation of projects and for decision and to each uncertain state. Project exibilities, or real
making regarding investments in real assets. One of options, can then be modeled easily as decisions that
the most important limitations of DCF is that it fails to affect these cash ows. This specication of project
account for the value of managerial exibility inher- uncertainties, cash ows, and decisions allows the
ent in many types of projects. The options derived problem to be modeled and solved using commer-
from managerial exibility are commonly called real cially available decision tree software familiar to the
options to reect their association with real assets decision analysis community. Our discussion expands
rather than with nancial assets. Although appeal- on the ideas presented originally by Brando and
ing from a theoretical perspective, the practical use of Dyer (2005) and illustrates the approach with several
real-option valuation techniques in industry has been examples.
limited by the mathematical complexity of these tech- While many of these ideas are relatively straight-
niques and the resulting lack of intuition associated forward and build on concepts suggested by Nau
with the solution process, or the restrictive assump- and McCardle (1991) and Smith and Nau (1995), we
tions required to obtain analytical solutions. hope to make this material more accessible to deci-
In this article we outline how traditional decision sion analysts and to encourage additional work on the
analysis tools can be used as an alternative to solve relationship between decision analysis and nance.
real-option valuation problems based on the ideas Triantis and Borison (2001) provide an assessment of
suggested by Copeland and Antikarov (2001) and fur- the use of options-based project valuation methods
ther illustrated in Copeland and Tufano (2004). We do in practice and conclude that a modest evolution is
this by using a binomial decision tree to determine occurring within some companies to support their
69
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
70 Decision Analysis 2(2), pp. 6988, 2005 INFORMS

adoption. In particular, Triantis and Borison antici- are maximized or expected losses are minimized.
pate increasing convergence among the various real- Examples of project exibilities include expanding
option approaches, particularly the decision-analytic operations in response to positive market conditions,
and option-pricing approaches. In that spirit we also abandoning a project that is underperforming, defer-
review some basic option-pricing concepts that will ring investment for a period of time, suspending
be familiar to many readers but that are nonethe- operations temporarily, switching inputs or outputs,
less included as a useful reference in the context reducing the project scale, or resuming operations
of this discussion. We will also take care to discuss after a temporary shutdown. The incremental value
the underlying assumptions and limitations of these of these options can only be determined using an
methods and to suggest when they might be a valu- option-pricing or decision analysis approach.
able addition to the decision-analysis tool kit when Option-pricing methods were rst developed to
used appropriately. value nancial options. However, the potential appli-
The remainder of the article is organized as follows. cation of these methods to the valuation of options
Section 2 reviews the traditional approaches to project on real assets was quickly identied, and hundreds of
valuation. Section 3 outlines a decision tree approach scholarly papers have been written on this topic. Nev-
to the real-option problem discussed by Copeland and ertheless, applications of real-option valuation meth-
Tufano (2004). Section 4 provides an extension of this ods to practical problems have been limited by the
approach to problems in which project cash ows mathematical complexity of the approach, by the
over time are explicitly modeled and used as the basis restrictive theoretical assumptions required, and by
for valuing real options. This approach is illustrated their lack of intuitive appeal.
in 5 with a numerical example. In 6 we conclude The mathematical complexity associated with real-
with a discussion of the limitations of this approach option theory stems from the fact that the gen-
and identify some areas for further research. eral problem requires a probabilistic solution to a
rms optimal investment decision policy, not only
2. Background on Project Valuation at present but also at all instances in time up to
With the DCF approach to valuation, the net present the maturity of its options. To solve this problem of
value of a project is calculated by discounting the dynamic optimization, the evolution of uncertainty in
future expected cash ows at a discount rate that the value of the real asset over time is rst modeled
takes into account the risk of the project. In practice, as a stochastic process. Then the value of the rms
this discount rate is often the weighted average cost optimal policy is a partial differential equation that
of capital (WACC) for the rm, based on the assump- is obtained as the solution to a value function rep-
tion that both the rm and the project share identical resented by Bellmans principle of optimality, where
market risks. While this assumption may be valid for appropriate boundary conditions reect the initial
projects that mimic the risks associated with the rm conditions and terminal payoff characteristics. When
as a whole, it may not be appropriate for unusual closed-form mathematical solutions are unavailable,
or innovative investment projects. In such cases, the which is usually the case for more complex problems
practitioner must exercise judgment in choosing an where the project may be subject to several sources
appropriate discount rate for the project. For a discus- of uncertainty and more than one type of option,
sion of the issues associated with the selection of a numerical methods and discrete dynamic program-
project discount rate and the calculation of the WACC, ming must be used to obtain a solution.
see Grinblatt and Titman (1998, Chapters 10 and 12). A discrete approximation to the underlying sto-
A major criticism of DCF is the implicit assump- chastic process can be developed to provide a trans-
tion that the projects outcome will be unaffected by parent and computationally efcient model of the
future decisions of the rm, thereby ignoring any valuation problem. The rst example of this approach
value that comes from managerial exibility. Manage- is a binomial lattice model that converges weakly to
ment exibility is the ability to make decisions dur- a lognormal diffusion of stock prices, developed by
ing the execution of a project so that expected returns Cox et al. (1979). A binomial lattice may be viewed as
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
Decision Analysis 2(2), pp. 6988, 2005 INFORMS 71

a probability tree with binary chance branches, with of nature and in all future periods. The assump-
the unique feature that the outcome resulting from tion of the existence of a replicating portfolio under-
moving up u and then down d in value is the lies much of the initial work done in the eld of
same as the outcome from moving down and then up. continuous-time, real-option valuation by Brennan
Thus, this probability tree is recombining, since there and Schwartz (1985), McDonald and Siegel (1986),
are numerous paths to the same outcomes, which sig- Dixit and Pindyck (1994), and Trigeorgis (1996).
nicantly reduces the number of nodes in the lattice. The use of traditional option-pricing methods and
A binomial lattice and the corresponding binomial the replicating portfolio approach is complicated by
tree are shown in Figure 1, where S is the current mar- the fact that, for most projects involving real assets, no
ket price of the asset and q is the probability of an such replicating portfolio of securities exists, so mar-
upward move to Su. kets are incomplete. In this case, Dixit and Pindyck
The binomial lattice model can be used to accu- (1994) propose the use of dynamic programming
rately approximate solutions from the Black-Scholes- using a subjectively dened discount rate, but the
Merton continuous-time valuation model for nancial result does not provide a market value for the project
options, with the added advantage of allowing a solu- and its options.
tion for the value of early-exercise American options, The application of decision analysis to real-option
whereas the Black-Scholes-Merton model can value valuation problems seems natural because decision
only European options. trees are commonly used to model project exibility,
Unfortunately, the process of working through lat- but there has been limited work in this area (Howard
tices can be cumbersome and nonintuitive, espe- 1996). Nau and McCardle (1991) and Smith and Nau
cially for more complex applications to real assets, (1995) study the relationship between option pricing
which can involve several simultaneous and com- theory and decision analysis and demonstrate that the
pound options. The typical approach to using a lat- two approaches yield the same results when applied
tice involves nding a replicating portfolio at each correctly. Smith and Nau propose a method that inte-
node. This approach is based on traditional option- grates the two approaches by distinguishing between
pricing methods, which require that markets be com- market risks, which can be hedged by trading secu-
plete in the sense that there are enough traded assets rities and valued using option pricing theory, and
to allow the creation of a portfolio of securities whose private uncertainties, which are project-specic risks
payoffs replicate the payoffs of the asset in all states and can be valued using decision analysis techniques.

Figure 1 Recombining Binomial Lattice and Corresponding Binomial Tree

q Su

S
Su 3
Su 2
1 q Sd Su 2d

Su
Sud Su 2d

Su 3 Sud 2
Su2
Su S
Sud Su 2d
q Su 2d
Sud
S Sud 2
1q Sud 2 Sd
Sd Sud 2
Sd2
Sd 2 Sd 3
Sd 3

t=0 1 2 3 t=0 1 2 3
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
72 Decision Analysis 2(2), pp. 6988, 2005 INFORMS

Smith and McCardle (1998, 1999) illustrate how this practitioners. However, their presentation is based on
approach can be applied in the context of oil and gas the use of binomial lattices and the construction of
projects and provide a discussion of lessons learned market portfolios that replicate the risk characteristics
from applications to some case studies. of the project, and therefore it suffers from a lack of
The distinction between market risks and project- intuitive appeal. Adapting this method to use bino-
specic risks is often a very natural one in oil and gas mial decision trees provides transparency to its logic
exploration projects, since oil and gas prices are mar- and offers a link to decision analysis approaches to
ket risks, while the project-specic risks may be the real-options valuation.
probability of a dry hole or the probability distribu-
tion regarding the volume of reserves. The McCardle-
Nau-Smith approach (henceforth MNS) has a natural
3. A Decision Analysis Approach to
appeal in problem contexts such as these. However, Valuation
there are projects in other industries where the dis- Decision tree analysis (DTA) can be used to model
tinction between market risks and project-specic managerial exibility in discrete time by construct-
risks is not as sharp. ing a tree with decision nodes that represent deci-
Copeland and Antikarov (2001) have proposed a sions the manager can make to maximize the value
more general approach (henceforth CA) to valuing of the project as uncertainties are resolved over the
real options that may be applied to problems in cases projects life. This approach allows some of the lim-
where there is no market-traded asset. To obtain this itations of the static DCF approach to be overcome.
generality, they make the assumption that the present In fact, a nave approach to valuing projects with real
value of the project without options is the best unbi- options would be to simply include decision nodes
ased estimator of the market value of the project corresponding to project options into a decision tree
(the marketed asset disclaimer, or MAD assumption). model of the project uncertainties and solve the prob-
Under this assumption, the value of the project with- lem using the same risk-adjusted discount rate judged
out options serves as the underlying asset in the repli- to be appropriate for the original project without
cating portfolio, which implies that the markets are options.
complete for the project with options. If the changes However, the nave approach does not provide a
in the value of the project without options are then correct valuation of the real options. This is because
assumed to vary over time according to a random the optimization that occurs at the decision nodes
walk stochastic process, more formally called geomet- changes the expected future cash ows, thereby alter-
ric Brownian motion (GBM), then the options can be ing the risk characteristics of the project. Thus, the
valued with traditional option pricing methods. standard deviation of the project cash ows with ex-
These assumptions are conceptually similar to those ibility is different from that of the project without
adopted earlier by Luehrman (1998a, b) to ration- exibility, and the risk-adjusted discount rate for the
alize the direct application of the classic Black and project without options may not be appropriate after
Scholes (1973) option-pricing model to real options. the real options have been included in the model.
While Luehrmans approach has generally been dis- This observation has caused some authors to incor-
counted as too simplistic (Triantis and Borison 2001), rectly conclude that DTA cannot be used to value
the development by CA is more robust. For example, real options (e.g., see the discussion in CA 2001).
it allows for the modeling of project cash ows and However, as noted by Smith (1999), the differences
other project-specic risks to capture a more realistic between the DTA and nance approaches are largely
representation of the underlying problem, for the use matters of style, and DTA can readily be augmented
of stochastic processes other than the GBM, and for to incorporate market information about risk.
the separation of market and private risks. To adjust the nave approach, we can use the repli-
Copeland and Tufano (2004) have recently cham- cating portfolio method to determine the correct dis-
pioned this approach in an article in Harvard Busi- count rates for the project and thereby capture the
ness Review, guaranteeing it high visibility among market information about project risks. Let us rst
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
Decision Analysis 2(2), pp. 6988, 2005 INFORMS 73

assume there is a project of an unknown value V and Note that the expressions for A and B do not
a replicating portfolio of an amount A of a market- include the probability q of an up move in the stock
traded stock with a current price S and of B dol- price, which eliminates the necessity of trying to esti-
lars invested in a risk-free bond that pays an interest mate this variable. This is an important advantage
rate r. For simplicity, we assume that for a one-period of this approach to valuation, since it relies only on
model with probability q the stock price will move up information that can be calculated from market data.
to Su at the end of the period, and with probability If the value of q were known, then the appropriate
1 q it will move down to Sd, where u is a number discount rate for the project could be found by solving
greater than 1 reecting a proportional increase in the the relationship between the expected future value
stock value, and d = 1/u is a number smaller than 1 and the current value V of the project for k, as shown
reecting a proportional decrease. This approach can in Equation (1).
be extended to multiple time periods by simply con- quVu + 1 qdVd
tinuing to apply these same percentage changes to V= (1)
1+k
the values determined at the end of the one-period
Fortunately, there is an equivalent but simpler pricing
model, as we will show later.
algorithm that is analogous to the replicating portfolio
The value of this portfolio one period from now
approach and that avoids the need to estimate q or k.
will be ASu + B1 + r and ASd + B1 + r in the up and In this alternative approach, we account for the
down states, respectively, and we assume that the val- project risk by adjusting the up and down proba-
ues of the project in these same up and down states, bilities rather than by adjusting the discount rate.
Vu and Vd , are known. The dynamics of the stock, The discount rate in Equation (1) is set equal to the
the bond, and the replicating portfolio are shown in risk-free rate of interest r, which is known, and (1)
Figure 2. For these portfolio values to replicate the is solved for the value of the implied probability p
value of the project in each of the up and down states instead of the value k. Since the risk-free rate r will be
exactly, the appropriate values of A and B must be less than the risk-adjusted discount rate k, the derived
determined by solving a system of two equations in probability p will be less than the true probability of
two unknowns, Vu = ASu + 1 + rB and Vd = ASd + the up state, q.
1 + rB, which yield A = Vu Vd /u dS The solution for p is easily obtained by using the
and B = uVd dVu /u d1 + r. If the holdings A relationship V = AS + B and substituting the values
and B are the replicating portfolio for the project at the for A and B determined above. The resulting equation
end of the period, then by the basic no-arbitrage argu- for the current value of the project is:
ment of nance theory, their current price, AS + B,     
1+r d u 1 + r
must also be the price, or value, of the project V . V= Vu + Vd 1 + r or
ud ud
While this form of the replicating portfolio method
provides a market-based adjustment for the risk in the pVu + 1 pVd 
V= 
project, for a multiperiod and multistate project this 1+r
proves to be cumbersome computationally, since this where p = 1 + r d/u d. These values are often
exercise must be repeated for each node of the lattice. called risk-neutral probabilities because assets are

Figure 2 Dynamics of the Stock Price, the Bond Yield, and the Project Value
q q q
Su (1+r) A Su + (1+r)B

S 1 AS + B

1 q 1q 1 q
Sd (1+r) A Sd + (1+r)B

Stock Bond Replicating Portfolio


Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
74 Decision Analysis 2(2), pp. 6988, 2005 INFORMS

priced as if there is a risk-neutral representative net present value (NPV) calculation with the informa-
investor with an estimate of probability p for the up tion currently available, but the value of this plant in
state. future years is uncertain and is expected to change
This risk-neutral probability p can then be used in a over time. The investments at the end of years one,
binomial lattice or tree to calculate an expected value two, and three are options and will be made only if
given the future payoffs, and the risk-free rate can be they are justied by the revised estimate of the project
used to discount the future payoffs. This shifts the at that point in time.
problem of nding the appropriate risk-adjusted dis- To carry out an analysis of this problem some
count rate for a project to the problem of nding the assumptions must be made regarding the uncertainty
appropriate risk-neutral probabilities to use in calcu- in the future value of the project. A common assump-
lating the risk-neutral measure of value. Fortunately, tion regarding stock prices is that current prices
the latter problem is often easier to solve, since these already incorporate all relevant information available
risk-neutral probabilities may be available from mar- at this point in time, and that future changes will be
ket data or from assumptions based on theoretical the effect of random and thus unpredictable shocks,
arguments regarding the underlying stochastic pro- which are modeled as a random walk. This assump-
cess associated with the value of the project. More- tion and other arguments support the use of a GBM
over, the GBM with constant volatility is the most to model the dynamic uncertainty associated with
common assumption regarding the stochastic process stock prices (Hull 2003). CA (2001, Chapter 8) use
associated with the project value, and this implies that similar arguments to justify the use of the GBM to
the values of p and 1 p are constant and applied model changes in the value of a project over time
throughout the lattice or tree, whereas the values of A in some instances, and it is used in this example for
and B must be calculated for every node. simplicity.
We demonstrate this approach by solving an exam- However, the assumption of the GBM model may
ple from a recent article by Copeland and Tufano not be appropriate in all situations, and it is not a
(2004). In this example, a rm is considering a requirement of the CA approach. In the discussion
phased investment in a plant. An initial investment section we describe how alternate models of stochas-
of $60 million to cover the cost of permits and prepa- tic processes may be approximated using binomial
ration for the effort is due immediately. At the end of lattices or trees and used with this approach.
one year, a commitment of $400 million is required for The critical parameters required to model the GBM
the design phase of the new plant. Once the design is are the starting value, $1,000 million in this example,
completed one year later, the rm would have a two- the risk-free interest rate r, assumed to be 8% per year,
year window during which to make the nal invest- and the volatility, denoted as , which is the annual-
ment in the plant of $800 million, which would pay ized percentage standard deviation of the returns and
for construction. If the rm decides not to invest dur- is given as 18.23% in this example. This allows the
ing these two years, it then foregoes the opportunity computation of the values of p, u, and d, respectively
to build the plant. the risk-neutral probabilities and the up and down
From the real-options perspective, this investment proportional changes in the value per period illus-
opportunity is a compound option. The initial pay- trated in the previous example. With these param-
ment of $60 million gives the rm the option to con- eters, this continuous-time stochastic process can be
tinue with the project for one year, at the end of which approximated with a discrete time binomial lattice.
it has the option to invest an additional $400 million Copeland and Tufano assume this process represents
in the design phase. In turn, the completion of the the evolution of the project value, without options,
design phase gives the rm the option to construct over time and that this serves as the underlying asset
the plant at the end of year two or at the end of year (MAD assumption).
three. The idea behind the calculation of the parameters
The rm estimates that if the plant existed today it used in the binomial approximation of the stochastic
would be worth $1,000 million based on a traditional process is relatively simple. If the value of the project
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
Decision Analysis 2(2), pp. 6988, 2005 INFORMS 75

Figure 3 The Binomial Tree Approximation to the Geometric Brownian Motion

Time 3 Up [1,372]
Up [1,235] 0.673 1,372

Time 2 0.673 Down [953]


Up [1,111] 0.327 953
0.673 Up [953]
Time 3
Down [857] 0.673 953

Time 1 0.327 Down [662]


[1,000] 0.327 662

Time 3 Up [953]
Up [857] 0.673 953

Time 2 0.673 Down [662]


Down [772] 0.327 662
0.327 Time 3 Up [662]
Down [595] 0.673 662
0.327 Down [459]
0.327 459

is assumed to follow a GBM, then the estimate of its is technically correct (given their assumptions), it is
value at any point in time has a lognormal distribu- neither intuitively appealing nor computationally
tion. By equating the rst and second moments of a transparent.
binomial and a lognormal distribution, we can calcu- The same parameters can be used in a decision tree
late the corresponding values of u and d, and thus with binary chance nodes to yield an equivalent bino-
Vu = Vu and Vd = Vd, for each branch of the binomial mial tree for the project value, as shown in Figure 3.
approximation to ensure that the discrete distribution The values shown at each node in the tree are dis-
approximates its continuous counterpart in the limit counted Year 3 values, instead of the actual values at
as t becomes small. Adding the convenient speci- each point. However, it can easily be veried that this
cation that u = 1/d to the equations for matching binomial tree corresponds to the lattice developed by
the mean and variance of the GBM yields u = e t .
Copeland and Tufano. Notice, for example, that the
We then obtain the risk-neutral probability p = 1 +
value at the end of the up move in Time 1 and the
rt d/u d. In this example, we model three peri-
down move in Time 2 is exactly equal to the value at
ods and choose t = 1. Therefore, u = e01823 = 12,
the end of the down move in Time 1 and the up move
d = 083, and p = 0673. We emphasize again that only
in Time 2. These two nodes would be combined into
three parameters are needed to specify this discrete
one node in the corresponding binomial lattice.
approximation to the GBM estimate of the evolution
The advantage of using the corresponding bino-
of the uncertain project value over time: the estimate
of the current value of this project, the volatility of mial tree rather than a binomial lattice can now be
the returns from the project, and the risk-free rate. For illustrated. The real options in the project can sim-
details associated with this binomial approximation, ply be modeled with decision nodes in the tree.
see Cox et al. (1979) or Hull (2003). This results in the tree in Figure 4, which shows
Copeland and Tufano (2004) solve this problem that the expected value of the project with options
using a recombinant binomial lattice and obtain the is $11 million after subtracting the initial investment
value of the options by calculating a replicating port- cost. Notice, however, that the effort should be aban-
folio with values for A and B at each node in the doned if the expected value of the project is lower at
lattice. The value of the project at any point in this the end of the rst time period, one year in this case.
lattice is given by Vi j = V0 uij d j . While this approach The approximation to the GBM could be improved by
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
76 Decision Analysis 2(2), pp. 6988, 2005 INFORMS

Figure 4 Solution to the Real-Options Problem Using a Binomial Tree


Yes [118]
549
Invest 3? Yes [306]
Invest 2? Up [306] 737
Up [169] Time 3 0.673 No [430]
0.673 Defer [169] 0
Invest 3? Yes [113]
Down [113] 318
Time 2 0.327 No [430]
Invest 1? Yes [46] 0
Up [46] 370 Yes [259]
0.673 171
Time 1
Invest 3? Yes [113]
Invest 0? Yes [11]
Invest 2? Up [113] 318
[11] 60
Down [208] Time 3 0.673 No [430]
0.327 Defer [208] 0
Invest 3? Yes [404]
Down [404] 26
0.327 No [430]
0
No [ 60]

Time 2
Invest 1? Yes [ 275]
Down [60] 370
0.327 No [ 60]

No [0]
0

adding additional periods of shorter duration at the ysis of the value of a stock option using the Black
expense of some computational burden. and Scholes model. Therefore, it is similar in spirit
This alternative approach yields the same optimal to the simplistic option valuation approach suggested
exercise policy and the project value of $71 million by Leuhrman (1998a). There are no allowances for
shown in Copeland and Tufano (2004) prior to sub- changes in the cash ows over time, for the fact that
tracting the investment cost. However, by using risk- the value of any project with a nite life will change
neutral probabilities in a decision tree, we did not as it is being executed or for options that occur during
need to solve for the replicating portfolio at each the operating life of the project.
node. Further, the optimal policy is obvious from the A traditional decision tree analysis of this same
graphic view of the decision tree, whereas it must be problem might include estimates of the uncertainties
inferred from a binomial lattice representation. associated with these underlying factors (sales vol-
The decision analyst might remain somewhat skep- umes, prices, etc.) in the calculation of the present
tical at this point, however, since this approximation values for the project, highlighting what Smith (1999)
to the value of the project over time is based on the has called an emphasis on modeling the sources of
GBM assumption, and the volatility of 18.23% was uncertainty in decision analysis versus an emphasis
simply given as one of the parameters for this prob- on modeling the dynamics of the uncertainty in real
lem. How might the volatility be derived in practice? options. As we shall see, however, these same sources
One might conjecture that the source of this volatil- of uncertainty can be used to estimate the volatility of
ity would be associated with uncertainties in some the project returns, and their impacts on cash ows
underlying factors, such as sales volumes, prices, over time can be modeled as well within this same
costs, and competitors actions. Further, this analysis generalized framework. The latter, in turn, allows the
is focused on the change in the estimated value of the representation of real options that may occur during
project and is very similar conceptually to the anal- the operating life of the project.
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
Decision Analysis 2(2), pp. 6988, 2005 INFORMS 77

4. Solving Real-Options Problems as a dream tree that cannot be solved because of its
with Binomial Decision Trees large size and suggest ways of trimming it.
Building on the work of Nau and McCardle (1991), An alternative to the construction of large trees
Smith and Nau (1995) suggested an approach for the with multiple uncertainties in each time period is the
valuation of real options using decision analysis tech- application of binomial decision trees to the approach
niques that differ in some signicant ways from the proposed by CA (2001), illustrated in the previous
one described above. The valuation procedure utilizes example. In the discussion that follows, we will let Vi
a separation of the project cash ows into two com- and Ci be random variables representing the uncer-
ponents, one subject to market risks and the other tain project values and cash ows in period i, and
subject to private risks. Market risks depend only on i and Ci will be their corresponding means. Realiza-
V
market states and can be hedged by creating a repli- tions of these random variables will be denoted with
cating portfolio of traded securities. Private risks are lower case, and the values associated with the dis-
project specic and thus cannot be hedged by trad- crete approximations to these random variables will
ing securities. The market component is then valued be denoted as Vij and Cij , where j indicates an out-
using market information (risk-neutral probabilities), come state.
while the private component is valued using subjec- In this development, we make the assumption that
tive beliefs and preferences (subjective probabilities the value of the project will evolve following a GBM
and certainty equivalents). In this approach, as long process but describe alternative assumptions in the
as all the market risks can be hedged with a marketed discussion. To show how this GBM assumption is
commodity or security, there is no need to estimate a utilized, let Vi be the value of a project at time
risk-adjusted discount rate for the project risks. period i and Vi+1 /Vi be its return over the time
This approach is generalized in an integrated roll- period between i and i + 1. Under the random walk
back method. The steps of the procedure are as fol- assumption, the logarithm of the random return z =
lows: (1) Calculate the NPV for all endpoints; (2) for lnVi+1 /Vi  is normally distributed, and we dene z
chance nodes with private uncertainties, use the and  2 as the mean and variance of this normal dis-
rms subjective probabilities and exponential utility tribution. The assumption that the distribution of the
function; and (3) for chance nodes with market uncer- logarithm of the project returns at any time is normal
tainties, use risk-neutral probabilities inferred from implies that the distribution of the project value at any
market information. Smith and Nau (1995) demon- time is lognormal. Therefore, Vi will be lognormally
strate this approach for the example of a plant in- distributed and can be modeled as a GBM stochas-
vestment with two underlying uncertainties: future tic process in the form dV= V dt + V dw, where
demand (which is correlated to a marketed security)  = z + 1/2 2 and dw =  dt is a standard Wiener
and plant efciency (private risk). process. For a discussion of the random walk assump-
In many projects, some uncertainties fall some- tion, see also Hull (2003) and Luenberger (1998).
where in between the notions of private and market The assumption that project returns follow a ran-
risks. For example, a pharmaceutical companys new dom walk is important for projects that involve sev-
drug development project may not include risks that eral uncertainties because it allows any number of
can be replicated by a traded asset, but the price of uncertainties in the project model to be combined
the product is clearly a market risk. Moreover, a into a single representative uncertainty: the uncer-
project may have numerous uncertainties to model. tainty associated with the stochastic process of the
Even if we can separate them into these two classes project value V . The parameters of this process can
and establish replication for each individual market be obtained from a Monte Carlo simulation of the
uncertainty, the underlying decision tree is compu- project cash ows. With these parameters, a discrete-
tationally unwieldy since we must include a sepa- time model using a binomial lattice or tree can be
rate chance node for each uncertainty in each time used to approximate the composite continuous-time
period. Smith and McCardle (1999) refer to the latter stochastic process as before.
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
78 Decision Analysis 2(2), pp. 6988, 2005 INFORMS

Consider a project that will last n periods that its returns. Under the MAD assumption, the present
requires an initial investment I to be implemented value of the project without options is taken as its
and that generates an expected cash ow Ci , i = market price, as if the project were a traded asset.
1 2     n in each of these periods. For simplicity we Assuming that markets are efcient, purchasing the
assume that the cash ows are paid instantaneously project at this price guarantees a zero NPV, and the
at the end of each time period in a manner analogous expected return of the project will be exactly the same
to stock dividends. as its risk-adjusted discount rate . As a result, the
The problem is modeled in three steps. First, the mean of the projects returns is exogenously dened.
expected present value of the project at Time 0 is cal-
culated. Next, a Monte Carlo simulation is performed Step 2
to combine several sources of uncertainty into a single The standard deviation of the returns, or volatility
representative uncertainty, which denes the stochas- of the project, can be estimated from a Monte Carlo
tic process for the value of the project. The third and simulation of the project returns. In this process, key
last step is to construct a binomial tree to model the project uncertainties are entered as simulation input
variables in the project cash ow pro forma worksheet,
dynamics of the project value using the parameters of
so that each iteration of a simulation of the worksheet
the stochastic process and to add the decision nodes
provides a new set of future cash ows ci , i = 1     n,
to model the projects real options.
from which a new project value v1 at the end of the
These rst two steps are identical to those pro-
rst period is computed from (2):
posed by CA (2001). For the third step we provide
an alternative solution methodology based on a bino- 
n
ci
v1 = 
mial tree that offers computational advantages and 1 + i1
i=1
a more intuitive logic. For completeness, we briey
summarize the rst two steps below and then discuss Then a sample of the random variable z can be deter-
our proposed modications of the third step in more mined using the relationship
 
detail. V
z = ln 1  (3)
0
V
Step 1
The expected present value of the project at Time 0, where z = Ez is the mean of the distribution of the
0 , is determined using the traditional DCF method
V project returns between Time 0 and Time 1. The esti-
and without considering any managerial exibility. mate of the standard deviation of z, denoted as s,
This requires the estimation of the appropriate risk- is obtained from the simulation results. The project
adjusted discount rate for the project without options volatility  is then dened as the annualized percent-
and introduces an element of judgment into this age standard deviation ofthe returns and is estimated
valuation approach (which we shall discuss subse- from the relationship s/ t, where t is the length
quently). These cash ows are then discounted at this of the period in years used in the cash ow pro forma
worksheet. If the time period between V1 and V 0 is
estimated risk-adjusted discount rate  to obtain the
expected present value of the project in each period: one year, then  = s.


n
Ci Step 3
t =
V  (2)
1 + it With the project volatility determined as indicated
i=t
above, and given the initial expected project value V0 ,
The expected present value of the project will de- a binomial lattice can be constructed to model the
crease in each period as t increases if the cash ows stochastic process for project value. The volatility
for
are all positive, due to the payout of the cash ows each time period in the binomial lattice is  t,
in each period. Thus, for a project with nite life, the where t is the time period used in the lattice. This
nal value of the project will be 0. is the approach illustrated by CA (2001).
The lognormal distribution of the projects value In contrast to the CA approach, we use a binomial
can be dened by the mean and standard deviation of tree and express the project value in terms of a more
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
Decision Analysis 2(2), pp. 6988, 2005 INFORMS 79

basic variable: the project cash ows. To do this, we at the risk-free rate to arrive at the present value of
use the cash ow payout rate, #i = Ci /V i , to calcu- the project at Time i = 0.
late the cash ows that are paid out at the end of The use of project cash ows in this approach
each time period as a function of the project value. provides a greater level of detail in modeling the
We assume that the cash ows will vary over time, operation of the project and the effects of manage-
reecting the uncertainty in the project value, but that rial decisions. For example, these cash ows could
they will remain a constant fraction of the residual ramp up over the early years of a project as sales
value of the project in each time period. These cash are forecasted to grow and decrease at an increasing
ows Ci j  will therefore be a function of the project rate at the end of the project life-cycle. As another
value and the stochastic process that drives the bino- example, a model of the development of an oil eld
mial model. The primary advantage of this approach could show lumpy increases in production as new
is that it provides greater exibility in the modeling wells are added, and then show a decrease in pro-
of the real options of the project. duction that would follow a decline curve. The model
To obtain the cash ows, we begin by building allows simple abandon options to be included in the
the tree of precash ow payout values. These val- tree and expansion and contraction options that can
ues are calculated according to the following equa- be modeled as percentage changes in the underlying
tions, where the superscripts u and d correspond to cash ows. For example, the option to sell a half inter-
the up and down state values and the state subscript est in the project could be modeled as a 50% reduc-
is suppressed: tion in subsequent cash ows, or the option to expand
operations could be modeled as a percentage increase
Viu = Vi1 Vi1 #i1 u in cash ows.
Vid = Vi1 Vi1 #i1 d The use of these cash ows, rather than project
values, allows the easy use of decision trees rather
The logic of this relationship should be transparent. than binomial lattices to evaluate project options. As a
Vi1 is the value of the project in the previous state, result, the evaluation of real options can be carried out
and Ci1 = Vi1 #i1 is the cash ow paid out at the conveniently using off-the-shelf decision tree soft-
end of the period, which reduces the project value in ware and allows options to be included in the models
the subsequent states. using decision nodes that are a natural part of this
There are no cash ows in the initial period i = 0, problem representation.
since the project has not yet been initiated, so #0 = 0.
For i = 1, V1u = uV0 and V1d = dV0 . For all subsequent 5. An Example Problem
periods, the cash ow payout rate is assumed to be We illustrate this approach to the evaluation of real
constant across states in each period but variable in options by solving for the value of an oil production
time, so the cash ows in each period are a xed pro- project using commercially available decision analysis
portion of the value of the project in that period and software, DPL . While a decision tree representation
state, as noted above. That is, in DPL does not take advantage of the recombining
Ci Ci j feature of binomial lattices and thus results in larger
#i = = j (4) trees than necessary, it is a convenient and exible
Vi Vi j
modeling tool that provides a simple and intuitive
Therefore, the discounted cash ow in each period/ visual interface.
state is simply given by The example project has estimated reserves of
Vi j #i 90 million barrels, and the initial production level
Ci j =  (5) of 9 million barrels declines by 15% per year over
1 + ri
its 10-year operating life. The variable operating cost
Thus, (5) provides the branch values in each chance starts at $10 per barrel in Year 0 and grows at 2% per
node of the binomial tree. Since risk-neutral probabil- year. Oil price starts at $25 per barrel and grows at
ities are being used, these cash ows are discounted 3% per year. There is also a $5 million per year xed
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
80 Decision Analysis 2(2), pp. 6988, 2005 INFORMS

Table 1 Base Case Expected Cash Flows for the Project

Year 0 1 2 3 4 5 6 7 8 9 10

Remaining reserves 900 810 734 668 613 566 526 492 463 439
Production level 90 77 65 55 47 40 34 29 25 21
Variable op cost rate 102 104 106 108 110 113 115 117 120 122
Oil price 258 265 273 281 290 299 307 317 326 336
Revenues 2318 2029 1776 1555 1362 1192 1044 914 800 700
Production cost 968 846 740 648 569 500 440 388 343 304
Cash ow 1350 1183 1036 907 793 692 604 526 457 396
Prot sharing 337 296 259 227 198 173 151 131 114 99
Net cash ows 1012 887 777 680 595 519 453 394 343 297
PV of cash ows 4040 4445 3776 3177 2640 2156 1717 1318 951 613 297
Cash ow payout rate 0.228 0.235 0.245 0.258 0.276 0.302 0.344 0.414 0.559 1.000

cost that is not shown in the table. The appropriate deviation of the project returns (3) to obtain an esti-
risk-adjusted discount rate is assumed to be 10% per mate of the project volatility, which was determined
year, and the risk-free rate is 5% per year. We initially to be  = 466%. This estimate of the project volatil-
determine the expected value of the future cash ows, ity was calculated directly from the simulation as
which are shown in Table 1. All values are in millions explained earlier, and since the time periods are one
of dollars. year in length in this example, this is the annualized
The Year 0 present value of the expected cash ows volatility of the project returns. The project volatility
is $404.0 million, which was calculated using the risk- may be signicantly different from the volatility of the
adjusted discount rate of 10% per year. This is used underlying project uncertainties because of the effects
as the best estimate of the current market value of of operational leverage. In this example, the impact
the project without options (base case). The required of price uncertainties on project cash ows may be
up-front investment is $180 million, so the projects magnied by the subtraction of operating and xed
NPV is $224.0 million. The project value at the end costs.
of each year may be determined using Equation (2), The nal assumption is that these returns are nor-
along with the corresponding cash ow payout rate #i mally distributed; consequently, the project values are
in each period using (4). For example, the cash ow lognormally distributed and can be modeled as a
payout rate in Year 1, #1 , is 1012/4445 = 0228, as GBM with constant volatility. The binomial approx-
shown in Table 1. imation to the GBM process may be modeled using
In the next step, project uncertainties that may have the DPL software. The input parameters are the
some correlation with the market are inserted into Year 0 value of the project, the volatility , the risk-
this deterministic model to perform a Monte Carlo free rate of return r, and the project cash ow payout
simulation on the project cash ows. We assume that ratios. The values of u, d, and the risk-neutral prob-
the project has two primary sources of market uncer- ability p are incorporated into the model and com-
tainty, price and variable operating costs, which fol- puted according to the formulas dened previously.
low a GBM stochastic diffusion process with a mean The cash ows in the DPL model are computed
annual rate of increase of 3% and volatility of 15% using (5), and the value of the project is determined
for the price process and of 2% and 10%, respec- by applying the usual procedures of dynamic pro-
tively, for the variable costs process. We could have gramming implemented in a binomial tree and dis-
made additional input variables to this model uncer- counting the expected cash ows at the risk-free rate
tain and included correlations or other relationships of return.
among them without any impact on the subsequent This construction of the tree guarantees that the
computational burden. present value obtained with this model is the same as
After a large number (e.g., 10,000) of iterations, the one calculated with the spreadsheet, as illustrated
the Monte Carlo simulation will provide the standard in Figure 5, where only the rst four of the ten periods
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
Decision Analysis 2(2), pp. 6988, 2005 INFORMS 81

Figure 5 Binomial Tree Model of Value of Project Without Options


T4
High [1,143]
T3
0.437 204.1
High [858.3]
T4
0.437 168.9
Low [637.1]
T2
0.563 80.37
High [613.2]
0.437 139.6 T4
High [534.7]
T3
[422.6] 0.437 80.37
Low
0.563 66.51 T4
Low [335.5]
T1
0.563 31.65
[404]
T4
High [450.1]
T3
0.437 80.37
High [338]
0.437 66.51 T4
Low [250.9]
T2
Low 0.563 31.65
[241.4]
T4
0.563 54.98
High [210.5]
T3
0.437 31.65
Low [166.4]
T4
0.563 26.19
Low [132.1]
0.563 12.46

of the tree are shown. Tree building can be greatly and in others the buy-out option is exercised. The
simplied by developing a standard template for a value of the project with these real options is
binomial tree for any given number of time periods. increased to $444.9 million, as shown in Figure 6.
The inputs to the binomial tree can also be linked to a More options and time periods can be added in a
spreadsheet using software packages such as DPL . straightforward manner.
This binomial tree represents the underlying asset As noted earlier, additional market uncertainties
and can now be used to evaluate real options. Sup- could be added to the simulation model and would
pose the project can be divested in the fth year of its increase the volatility estimate for the project if not
life for a price of $100 million. The rm might specif- negatively correlated with the other risks. As a result,
ically want this option if it is averse to risks later in the value of the options would increase relative to the
the project life. Given the binomial tree representa- project base value because of the increase in volatility.
tion, this option can be evaluated by simply inserting In a manner consistent with the approach Smith
a decision node in Year 5 that models the managerial and Nau outlined earlier, we can also add private
exibility that exists in the fth year of the project. uncertainties to this problem. For example, suppose
Additional options can be evaluated by adding the the oil production in this example is driven by an
appropriate decision nodes in the tree. For example, underlying aquifer, and there is uncertainty about the
suppose the rm can also buy out its partner (assume level at which the oil-water interface exists. When
the partner holds a 25% interest) in Year 5 at a cost of this interface reaches the well, it will begin produc-
$40 million. Since the rm already owns 75% of the ing water, and operations will be shut down. This is
project, purchasing the remaining 25% represents an an example of an uncertainty that has zero correlation
increase in value of one-third. A new present value with any marketed security. We can model this uncer-
for the project is then computed using the same risk- tainty in the decision tree by adding chance nodes in
neutral probabilities, as illustrated in Figure 6, where the appropriate time periods in the tree and increas-
again, not all nodes are expanded. ing the probability as time goes on and the limit of
In some of the states the option to abandon by oil production is approached. A decision tree for this
divesting ownership in the project will be exercised, addition to our model is shown in Figure 7, where we
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
82 Decision Analysis 2(2), pp. 6988, 2005 INFORMS

Figure 6 Value of Project with Option to Buy Out and Divest

T4
High [1,225]
0.437 204.1 T6
Buyout [981.5]
Option
31.34
Low [981.5] T6
High Continue [910.4]
T1 0.437 117.1
0.437 T5 Divest [681.4]
[444.9] High [801.7] 78.35 T6
0.437 97.08 Buyout [662.1]
Option
31.34 T6
High [662.1]
T3 Continue [653.1]
T4 0.563 46.11
High [909.6] Divest [610.4]
0.437 168.9 Low [664.8] 78.35 T6
0.563 80.37 Buyout [603.2]
Option
31.34 T6
Low [603.2] [594.2]
Continue
T5 0.437 46.11
Divest [551.6]
Low [558.4] 78.35 T6
0.563 38.23 Buyout [477.5]
Option
31.34 T6
High [523.6] [492.9]
Continue
0.563 18.16
T2 Divest [523.6]
High [653.7] 78.35
T1
[444.9] 0.437 Buyout [879.1] T6
139.6 Option
31.34 T6
Low [879.1] Continue [808]
0.437 117.1
T5 Divest [579]
High [699.4] 78.35 T6
Buyout [559.7]
0.437 97.08 Option
31.34
High [559.7] T6
Continue [550.7]
0.563 46.11
T4 Divest [508]
T3 High [562.4] 78.35 T6
Low [454.8] 0.437 80.37 Buyout [500.8]
Option
0.563 66.51 31.34 T6
Low [500.8] Continue [491.8]
0.437 46.11
T5 Divest [449.2]
Low [456] 78.35 T6
0.563 38.23 Buyout [375.1]
Option
31.34 T6
High [421.2]
Continue [390.5]
0.563 18.16
Divest [421.2]
T4 78.35
Low [371.2]
0.563 31.65
Low [282.6] T 2
0.563 54.98

assume that water can only reach the well after ve simply removing the decision branch for the abandon
years of operation. option.
As we would expect, adding this uncertainty re- Thus far, in considering the private uncertainty, we
duces the overall value of the project ($428.0 million, have assumed the rm is risk neutral. This may be
solved tree not shown), since the occurrence of water reasonable for a very large rm that has exposure to
in the well terminates the cash ows but the exposure many such projects. However, a small rm with a
to this downside loss is greatly limited by our option limited number of such capital investments may be
to divest. Without this option, the project value would risk averse, rather than risk neutral. As the cost of
fall to $397.1 million. This value is easily calculated by the investment increases, a risk-averse rm will have
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
Decision Analysis 2(2), pp. 6988, 2005 INFORMS 83

Figure 7 Project with Real Options and a Private Risk


T1 T2 T3 T4 T5

High High High High Low


(T 1*d1)/(1+rf )^1 (T2*d2)/(1+rf )^2 (T3*d3)/(1+rf )^3 (T4*d4)/(1+rf )^4 (T5*d5)/(1+rf )^5
a
Low Low Low Low High
(T1*d1)/(1+rf )^1 (T2*d2)/(1+rf )^2 (T3*d3)/(1+rf )^3 (T4*d4)/(1+rf )^4 (T5*d5)/(1+rf )^5

Option
T6 W6 T7 W7 T8
Buyout
b
40/(1+rf )^5 High High High
(T6*d6)/(1+rf )^6 No (T 7*d 7)/(1+rf )^7 No (T8*d8)/(1+rf )^8
Continue b c
a b Low Low
Low
(T6*d6)/(1+rf )^6 (T 7*d 7)/(1+rf )^7 (T8*d8)/(1+rf )^8
Divest
Yes Yes
100/(1+rf )^5

W8 T9 W9 T10

High High
No (T9*d9)/(1+rf )^9 No (T 10*d 10)/(1+rf )^10
c
Low Low
(T9*d9)/(1+rf )^9 (T 10*d 10)/(1+rf )^10
Yes
Yes

a decreasing marginal value for the project because it 6. Discussion


loses the ability to diversify its risks. The risk aversion The objective of showing the developments in the pre-
of such a rm can be modeled by assessing its utility vious sections was to illustrate how binomial deci-
function. For this example, we assume the rms util- sion trees can be used to solve real-option problems
ity function is the exponential form U c0  c1      cT  = using the approach suggested by CA (2001). To make

Tt=0 expct /RTt , where ct and RTt are the cash this discussion as simple and transparent as possible,
ows and risk tolerances, respectively, in each period. we have focused on their basic approach as it is pre-
We use RT0 = $100MM and increase each subsequent sented in their textbook. However, this approach can
risk tolerance over time to reect the rms time pref- be modied to include the use of alternate stochastic
erence for cash ows, as indicated by a 10% discount processes rather than the GBM, and therefore it pro-
rate. An effective risk tolerance for each period can vides additional exibility.
then be calculated as described by Smith and Nau In practice, there are a number of issues that should
(1995) and entered into the chance nodes for the pri- be considered in an attempt to apply this method-
vate risks for the calculation of the certainty equiva- ology within a decision-analysis framework. Like all
lent for the project. modeling approaches, this framework has its limita-
The rms effective risk tolerance is applied to tions, but it also has some exibility that should be
chance nodes for the private risk only, so the risk- recognized. We will organize this discussion to focus
neutral view is retained for the chance nodes in the on the assumptions required by this approach, and
tree that are risk-adjusted by the risk-neutral proba- on ways in which this model might be extended. As
bilities. This change in the model results in a drop we shall see, the CA approach implemented using
in value measured by the certainty equivalent to binomial decision trees can be viewed as complimen-
$400.5 million. Although the value has been further tary to the decision analysis approach to solving real-
reduced, the risk-averse rm is protected by the aban- options problems developed by McCardle, Nau, and
don option. Smith.
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
84 Decision Analysis 2(2), pp. 6988, 2005 INFORMS

The MAD Assumption risk-adjusted discount rates, which would result in


The use of the MAD assumption by CA as the basis a similar discounting approach to the one suggested
for creating a complete market for an asset that by CA. Therefore, when used with proper judgment
is not traded could lead to signicant errors, since regarding the pricing of market risks, the MNS and
the approach is based on assumptions regarding the CA approaches will use similar modeling inputs.
value of the project without options that cannot be
tested in the market place. Since identical copies of Market vs. Private Risks
the project are not freely traded in the market, this The risks that are included in the simulation model
should be recognized as a strong modeling assump- used by CA (2001, Chapter 9)price, quantity, and
tion to justify the use of risk-neutral pricing for project variable costsmay all have some correlation with
options. the market and therefore be considered market risks.
The choice of the discount rate for the project with- The use of the simulation model as a basis for esti-
out options is left to the discretion of the analyst, mating the projects value without options and its
and the use of the WACC will not be appropriate volatility should be restricted to include only risks
for all projects even though it is commonly used in that arguably have some correlation with the market.
practice. Therefore, it is important to realize that the CA (2001, Exhibit 10.1) illustrate how to include
issue of selecting a risk-adjusted discount rate for private risks in their analysis as well, and treat them
the project is not resolved by this methodology. independently in a manner similar in spirit to the
Under ideal conditions, the MNS approach avoids approach suggested by Smith and Nau (1994). That
this problem by dividing risks into market and is, these risks are kept separate in their event tree,
private categories and by using information from but the solution is still carried out using replicating
market-traded commodities (oil prices in the case of portfolios at each node. This same problem can be
Smith and McCardle 1999) or from a correlated stock solved using DPL , where the discrete approximation
price (Smith and Nau 1995) to estimate the risk- to the underlying stochastic process is kept separate
neutral probabilities for these risks. We agree that from chance nodes representing the private risk. This
this should be done when such market information approach is also illustrated by the incorporation of the
is available, and in fact it can be incorporated into private risk associated with the oil-water interface in
the CA simulation model was well. For example, the the example in the previous section.
stochastic process for oil price in the example pro-
vided in the previous section could easily be speci- GBM Assumption
ed using market information (e.g., see Schwartz and The GBM assumption is a standard one in nance as
Smith 2000). It might also be possible to nd market an estimate of the price or value of a market-traded
replication to approximate the cost process, in which asset. As indicated earlier, CA provide a rationale and
case the appropriate risk-adjusted rate for the project some empirical results to support this assumption as
would be the risk-free rate, and this would be logi- a reasonable one to consider for estimating the future
cally consistent with the MNS approach. value of a project. However, they also recognize that
As a practical matter it may be difcult to iden- this assumption may not be appropriate for every
tify replicating portfolios of market-traded assets for project. For example, they discuss the use of the bino-
all market risks in a project. For example, the risks mial lattice to value options on projects that follow an
associated with a pharmaceutical companys new arithmetic Brownian motion (ABM) in instances when
drug might include marketing costs, market size, and the change in the assets value is assumed to be addi-
price, and it may be impractical to estimate repli- tive rather than multiplicative and project value may
cating portfolios of market-traded assets for each of go negative.
them. In such a case, Smith (personal communication, However, there is considerable exibility in mod-
May 2002) suggests estimating the risk premiums for eling the underlying stochastic process with a bino-
these risks by considering their correlations with the mial tree. If the primary uncertainty associated with
market and effectively estimating their appropriate an asset is thought to be mean reverting, as in the
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
Decision Analysis 2(2), pp. 6988, 2005 INFORMS 85

case of oil or other commodity prices, then Hahn and conditional probability distributions for the uncertain-
Dyer (2004) show how a binomial tree may be used to ties in each period where they appear. If the uncer-
approximate such mean-reverting models as the one- tainties were correlated, this approach would become
factor Ornstein-Uhlenbeck process or the two-factor even more challenging. But if one were careful about
Schwartz and Smith (2000) process. CA (2001, Chap- exploiting the recombining nature of the resulting
ter 9) also discuss the use of a mean-reverting stochas- trees, it could still be manageable.
tic process within this framework. The representation of the individual market risks
We have illustrated the use of simulation to esti- with separate chance nodes might provide additional
mate the volatility associated with a project by cal- insights into the way the optimal exercise strategies
culating its value at the rst period only and then for the options depend on a key uncertainty, and
assuming that it remains constant over the life of this might be lost when these uncertainties are com-
the project, as required by the GBM assumption. bined into a single stochastic process using the CA
That assumption could be veried by calculating the approach. The choice of one approach or the other
volatility at other time periods during the simulated should depend, we suggest, on both the nature of the
project life using Equation (3), modied to adjust for problem and the preferences of the modeler.
the appropriate time period. This would be especially
relevant if some of the risks in the simulation model Binomial Lattice vs. Binomial Trees
were changing over its life in idiosyncratic ways. For We have discussed how binomial trees with risk-
example, if the production rate were decreasing over neutral probabilities may be used to provide discrete
time, the volatility might be decreasing as well. Other time approximations to the stochastic processes that
uncertainties may not occur until several time periods are often used in the valuation of real options. While
have passed, such as those associated with a planned this approach is suggested by CA, they emphasize
investment decision or new product introduction, or the use of binomial lattices and replicating portfolios.
they may even be modeled as jump processes. We believe that most decision analystsand most
This heteroskedasticity could be incorporated by managers without technical training in real options
changing the volatility in the binomial tree at the would nd a problem representation based on bino-
appropriate time periods, which would be imple- mial trees to have more intuitive appeal.
mented by corresponding changes in the values Even for a simple model such as the one illus-
of u, d, and the risk-neutral probability p in these time trated in the previous section, the decision tree very
periods. CA (2001, p. 342) recognize this possibility quickly becomes large. In most practical problems
and note that the stochastic process could be mod- the complexity of the decision tree will be such that
eled with a binomial tree rather than with a binomial full visualization will be impossible. However, even
lattice. While it may be possible to develop a recom- large problems with literally millions of endpoints for
bining lattice with changing volatility over time, this the tree can be solved using this approach. Brando
introduces additional complexity into the calculation (2002) provides an example of the application of this
of the probabilities on the branches. It is relatively methodology to the evaluation of options associated
straightforward, however, to model a heteroskedastic with a highway project in Brazil that includes 20 time
process using the decision tree approach that we have periods and several different options, resulting in a
illustrated. decision tree with 2 109 endpoints that is solved
The obvious alternative to the CA approach is to within practical computational times.
use the model and distributions from the Monte Carlo If only the expected value of a project is needed, it
simulation to build a traditional decision tree with is not necessary to expand the binomial tree beyond
chance nodes for each uncertainty in each period the point at which the last option is introduced as a
and value the options in the problem without using decision node, since the expected value of this expan-
the GBM approximation or one of the extensions sion is known at that point. This is illustrated in Fig-
mentioned above. This may lead to a more complex ure 4, where the binomial chance node for the third
model and would require the estimation of a set of period is not expanded if the decision to invest in the
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
86 Decision Analysis 2(2), pp. 6988, 2005 INFORMS

new plant is made. This could provide some compu- degree of accuracy in the estimate is required. How-
tational efciency in some applications. ever, the binomial tree is certainly a practical com-
While an n period recombining binomial lattice has putational tool for n = 20 periods and could even
a total of n + 1n + 2/2 nodes, an equivalent bino- be used for larger numbers of time periods, up to
mial tree has 2n+1 1 nodes, which represents a sig- approximately the n = 30 periods suggested by Hull.
nicant difference for large values of n. Therefore, A characteristic of the binomial method is that the
we conducted a simple comparison of the computa- convergence is not smooth and oscillates around the
tional performance of a real options problem modeled true value (Clewlow and Strickland 1998, p. 20). For
with the binomial tree versus the binomial lattice. The this reason it may be desirable to make several com-
problem we selected was the example problem used puter runs with binary decision trees of different time
by Copeland and Tufano (2004) and solved using a periods and average the results. On the other hand,
binomial tree with n = 3 time periods in our earlier estimates of value in real-options problems may not
example. require the same accuracy that is typically demanded
We created a binomial lattice to solve this problem when using lattices to value nancial options.
using a VBA code and compared its performance to The lattice also provides a representation of the
the corresponding binomial tree representation solved problem that is visually more compact. The optimal
using DPL (version 6). While other commercial deci- exercise decisions can be indicated by shading or for-
sion tree software, such as PrecisionTree , could be matting values shown in the lattice, and it may be
used for example problems in a classroom setting, we easier to see thresholds, e.g., exercise if the value
believe that the inuence diagram interface in DPL exceeds some specic number, in the lattice. However,
is useful for modeling problems of realistic size using binomial lattices do become complex when dealing
this approach. We made no effort to optimize the com- with multiple uncertainties, path-dependent uncer-
putational efciency of the DPL software, and sim- tainties or payoffs, and complex options. These prob-
ply used the default settings. lems can be handled more conveniently with binomial
According to Hull (2003), in practice solving a trees. For example, compound options can be mod-
binomial lattice with n = 30 usually gives reasonable eled simply by adding additional decision nodes to
results, so we used this as the upper limit for our the binomial trees.
range. The results were obtained with an IBM T40 lap- According to Triantis and Borison (2001), the choice
top computer using a 1.5 GHz processor and 256 K of a binomial lattice or tree structure by analysts in
RAM, and are shown in Figure 8. practice often reects the background of the individ-
As indicated in Figure 8, a well-constructed lattice ual as well as the complexity of the project being eval-
is much more computationally efcient, which may uated. Binomial lattices are typically used by those
be very important in large problems or when a high with nance training who are looking at relatively
straightforward investment problems.
Perhaps a more relevant comparison of the com-
Figure 8 Comparison of Computational Efciency: Tree vs. Lattice
putational efciency of the binomial tree based on
10,000
Lattice the CA approach would be with the probability tree
Computational time (minutes)

Tree required by the MNS method. If there is only one mar-


100
ket uncertainty in the corresponding tree, and a tri-
1 nomial chance node is created with estimates of high,
medium, and low outcome values, for example, then
0.01 after 10 periods it would contain 88,573 nodes, com-
pared with 66 for the binomial lattice and 2,047 for
0.0001 the binomial decision tree. Of course the use of the
trinomial chance nodes would provide more precision
0.000001
3 6 9 12 15 18 21 24 27 30 in the estimation of the stochastic process associated
Number of periods with the risk, so a smaller number of periods might
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
Decision Analysis 2(2), pp. 6988, 2005 INFORMS 87

be used. If there were two market uncertainties in the may be the preferred approach if the number of mar-
problem, as in the example in the previous section, ket risks is limited, as in oil and gas exploration. Indi-
the MNS probability tree with trinomial chance nodes vidual modeling skills and preferences would also be
would contain almost 4 billion nodes after 10 peri- a major consideration.

ods. In general, it would contain 1 + ni=1 xm i chance We agree with Triantis and Borison (2001) that
nodes, where x is the number of branches at each there should be a convergence of real-option eval-
node, m is the number of market risks, and n is uation models between nance and decision analy-
the number of periods. A seasoned analyst would sis. In their recent summary article, Smith and von
never try to build such a tree and would nd ways Winterfeldt (2004) also call for more research on
to trim it to a manageable size. For example, Smith the links between decision analysis and nance. The
and McCardle (1998, 1999) discuss the use of dynamic recognition of the similarities between the use of bino-
programming formulations and lattices in such set- mial decision trees and the use of binomial lattices for
tings. Nevertheless, it should be clear that the MNS solving real-option problems offers a rich opportunity
approach will generate very large dream trees as for further research.
well and cannot be applied naively to projects on a Our comparisons between the CA and MNS ap-
period-by-period basis. proaches have been based on observations and mod-
In practice, we think that the decision analyst eling experiences rather than on a rigorous theoreti-
should be aware of the trade-offs between the use of cal analysis, and we acknowledge that more could be
the binomial tree and the binomial lattice to model done to explore these ideas. Likewise, our computa-
real-option problems and recognize that there may be tional comparisons were merely suggestive of more
situations in which one or the other would be pre- rigorous work that could be done to investigate the
ferred. Similar considerations would apply to the use computational properties of these methods.
of the MNS approach as well. Based on our experience in modeling ABM and
GBM processes, we have also developed a binomial
decision tree approach that can be applied to model
Summary mean-reverting stochastic processes (Hahn and Dyer
We have shown an approach for solving real-option 2004). In this spirit there may be more to be gained
valuation problems with decision analysis methods by reviewing other work on binomial lattices that has
that is consistent with nance-based methods used appeared in the nance literature and adapting some
in practice. This approach provides a straightforward of these models into a decision analysis framework.
yet exible way to implement real-option valuation All of the spreadsheets and DPL models for the
techniques using off-the-shelf decision analysis soft- example problems in this paper are available in the
ware. Additional computational efciencies may be Online Supplements section of the Decision Analysis
obtained by using specially coded algorithms to solve web page.
binomial lattices, although at the cost of having to
forgo the simple user interface offered by decision tree
References
programs such as DPL and the advantage of visual
Black, F., M. Scholes. 1973. The pricing of options and corporate
modeling and a logical representation. liabilities. J. Political Econom. 81 637659.
The CA approach can be used to create models that Brando, L. 2002. An application of real-option theory in dis-
are consistent with the ideas developed by MNS. The crete time to the valuation of a highway concession project in
Brazil. Unpublished doctoral dissertation, Ponticia Universi-
primary difference between these two approaches is dade Catolica do Rio de Janeiro, Rio de Janeiro, Brazil.
in the treatment of the market risks in the models. CA Brando, L., J. Dyer. 2005. Decision analysis and real options: A dis-
suggest reducing them to one stochastic process by crete time approach to real option valuation. Ann. Oper. Res.
135(1) 2139.
focusing on their impacts on cash ows. MNS model Brennan, M., E. Schwartz. 1985. Evaluating natural resource invest-
these individual risks in each time period. This sug- ments. J. Bus. 58(2) 135157.
gests that the CA approach might be an appropriate Copeland, T., V. Antikarov. 2001. Real Options. Texere LLC,
New York.
choice if there are several market risks and several Copeland, T., P. Tufano. 2004. A real-world way to manage real
time periods in a model, whereas the MNS approach options. Harvard Bus. Rev. 82(3) 9099.
Brando et al.: Using Binomial Decision Trees to Solve Real-Option Valuation Problems
88 Decision Analysis 2(2), pp. 6988, 2005 INFORMS

Cox, J., S. Ross, M. Rubinstein. 1979. Option pricing: A simplied McDonald, R., D. Siegel. 1986. The value of waiting to invest. Quart.
approach. J. Financial Econom. 7 229263. J. Econom. 101(4) 707727.
Clewlow, L., C. Strickland. 1998. Implementing Derivative Models. Nau, R., K. McCardle. 1991. Arbitrage, rationality and equilibrium.
John Wiley & Sons, Chichester, UK. Theory Decision 33 199240.
Dixit, A., R. Pindyck. 1994. Investment Under Uncertainty. Princeton Schwartz, E., J. Smith. 2000. Short-term variations and long-term
University Press, Princeton, NJ. dynamics in commodity prices. Management Sci. 46 893911.
Grinblatt, M., S. Titman. 1998. Financial Markets and Corporate Strat- Smith, J. 1999. Much ado about options? Decision Anal. Newsletter
egy, Irwin/McGrawHill Companies. 18(2) 48.
Hahn, W., J. Dyer. 2004. A discrete-time approach for valuing real Smith, J., K. McCardle. 1998. Valuing oil properties: Integrating
options with underlying mean-reverting stochastic processes. option pricing and decision analysis approaches. Oper. Res. 46
Working paper, The University of Texas, Austin, TX. 198217.
Howard, R. 1996. Options. R. Zeckhauser, R. Keeney, J. Sebenius, Smith, J., K. McCardle. 1999. Options in the real world: Lessons
eds. Wise Choices. Harvard Business School Press, Boston, MA, learned in evaluating oil and gas investments. Oper. Res. 47
81101. 115.
Hull, J. 2003. Options, Futures and Other Derivatives. Prentice Hall, Smith, J., R. Nau. 1995. Valuing risky projects: Option pricing the-
Upper Saddle River, NJ. ory and decision analysis. Management Sci. 14(5) 795816.
Luehrman, T. 1998a. Investment opportunities as real options: Get- Smith, J., D. von Winterfeldt. 2004. Decision analysis in manage-
ting started on the numbers. Harvard Bus. Rev. 76(4) 5167. ment science. Management Sci. 50 561574.
Luehrman, T. 1998b. Strategy as a portfolio of real options. Harvard Triantis, A., A. Borison. 2001. Real options: State of the practice.
Bus. Rev. 76(5) 8999. J. Appl. Corporate Finance 14 824.
Luenberger, D. 1998. Investment Science. Oxford University Press, Trigeorgis, L. 1996. Real OptionsManagerial Flexibility and Strategy
New York. in Resource Allocation. MIT Press, Cambridge, MA.

View publication stats