Perspectives in Science (2016) 7, 10—18
Available online at www.sciencedirect.com
ScienceDirect
j ournal homepage: www.elsevier.com/pisc
Real options valuation with changing volatilityଝ
Miroslav Culíkˇ
VSB — Technical University Ostrava, Faculty of Economics, Finance Department, Sokolska trida 33,
701
21 Ostrava, Czech Republic
Received 25 September 2015; accepted 11 November 2015
Available online 10 December 2015
JEL Summary This paper aims at the valuation of real options with changing volatility. Volatility
CLASSIFICATION change is a typical feature of real investment projects, where the riskiness of cash flow gener-
G11; ated by the project can change significantly during the project life span. In this paper, there is
G12; explained how the problem of changing volatility can be considered if binomial lattice and repli-
G13; cation strategy is used for real option valuation. There are recombining and non-recombining
G31 lattice used and constant and increasing volatility are analysed and results compared. In situ-
KEYWORDS ation when volatility is changing, two approaches overcoming this problem are employed and compared.
Real options;
© 2015 Published by Elsevier GmbH. This is an open access article under the CC BY-NC-ND license Valuation; (http://creativecommons.org/licenses/by-nc-nd/4.0/). Risk-neutral; Transition probabilities;
Binomial lattice; Recombining; Non-recombining;
Volatility
Introduction Compared to the traditional passive valuation approaches
(NPV, IRR, etc.), real option approach takes into consid-
eration two important aspects: (a) riskiness of cash flow
Real options methodology is a relatively new approach for
generated by the assets and (b) flexibility, i.e. capability
solution of a wide range of valuation and decision-making
of management to change past decision or to make new
issues. Here, traditional methods and models used for finan-
ones in already undertaken projects. These future possi-
cial option valuation are used for real assets valuation.
ble decisions (depends on the future state of the world) are
modelled as a formal call and put options, which have their
ଝ value and can be exercised by company’s management. Real
This article is part of a special issue entitled ‘‘Proceedings of
asset value provided by the real option methodology appli-
the 1st Czech-China Scientific Conference 2015’’.
address: [email protected] cation is given as a sum of two components: present value of
http://dx.doi.org/10.1016/j.pisc.2015.11.004
2213-0209/© 2015 Published by Elsevier GmbH. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Real options valuation with changing volatility 11
directly measurable cash flows and flexibility value, which III. Management has flexibility.
captures managerial possibilities (real options). IV. Flexibility strategies (real options) are creditable and
As mentioned, assets value and future managerial oppor- executable.
tunities captured in real options are quantified by financial V. Management is rational in executing real options.
option valuation models. These models are based on some
assumptions, which are sometimes difficult to keep due to Future managerial investment opportunities captured in
the specific feature of real investments and real options real options and quantified by financial option valuation
captured in them (constant volatility, risk free rate, etc.). models represent the flexibility component (active part) of
That is why it is necessary to adjust these models to spe- the project value. The total project’s NPV then consists
cific conditions of a given project otherwise these are not of two components: the traditional static (passive) NPV of
applicable. directly measurable expected cash flows, and the flexibil-
ity value capturing the value of real options under active
management, i.e.,
Expanded NPV = standard (static, passive) NPV of directly measurable cash flows
+ flexibility value (value of real options from active management).
Valuation procedure by applying real options method-
The goal of this paper is the application of financial option
ology when discrete valuation model is applied can be
valuation models on the real asset under specific condition
described by the following steps:
typical for most real investments — changing risk (cash flow
volatility) during the expected life span.
1. Estimation of the type and parameters of the underlying
The paper is structured as follows. First, real option
asset random evolution (return mean, standard deviation
methodology is described and classification of financial
of returns);
option valuation models and application possibilities is
2. Simulation of the future random underlying asset evolu-
stated. In the subsequent part, mathematical background
tion for each discrete node of the tree;
for lattice valuation models is provided including the situa-
3. The option’s intrinsic value calculation (for given type of
tion of variable parameters. In the end, illustrative example
real option or portfolio of options) for each discrete node
is stated.
of the tree;
4. Flexibility and asset’s value quantification; and
Real options methodology and flexibility 5. Recommendation of the optimal decision.
valuation
For flexibility quantification, traditional models for finan-
cial options pricing are employed. These models can be
Real options methodology represents an approach where
classified as follows: (a) analytical (Black—Scholes model),
financial options pricing theory and models are applied on
discrete (binomial, trinomial, multinomial), and simulation real assets valuation.
(Monte Carlo).
There are many applications in corporate finance where
Because the real options are mostly the American options
the financial option valuation models are applied on the real
(decisions can be made at any time until the investment
assets valuation (real options approach). Black and Scholes
opportunity disappears), more possible decisions can exist
(1973) were the first authors to state that it is possible
at a given point of time, so the discrete valuation model on
to take the equity in a levered company as a call option
the basis of replication strategy is frequently applied.
on the company value. Since then, new pioneering work
appeared developing real option analysis; see for example
Merton (1973), Cox et al., 1979 or Brennan and Schwartz
Replication strategy for option valuation
(1985). During the last two decades, significant increase in
publishing activities on this topic is obvious. This area has
This approach relies on the fact that it is possible to set
been studied and developed by many authors, and new pos-
up a portfolio of the underlying asset and risk free borrow-
sible applications appear for solutions for a wide array of
ing, whose value replicates the payoff of the option for any
financial-decision and valuation problems. The key papers
state of the underlying asset value. Because there are two
and books that focus on the real options methodology appli-
assets (portfolio, option) providing identical payoffs, in the
cation are those of Dixit and Pindyck (1994), Smith and
absence of arbitrage opportunities, their current price must
Nau (1995), Trigeorgis (1999), Brennan and Trigeorgis (2000),
be the same. This makes it possible to work out the cost of
Copeland and Antikarov (2003), Grenadier (2000), Brach
setting up the portfolio and, thus, the option’s price.
(2002), Trigeorgis and Schwartz (2001), Trigeorgis and Smith
The following symbols are used for the option’s price
(2004) or Damodaran (2006).
derivation: h is the quantity (number of units) of the under-
Real options methodology application on real projects is
lying asset, St is the price of the underlying asset at t, Bt is
justifiable only if:
the monetary amount of the risk free borrowings, Ct is the
option’s price at t, t is the value of the replication port-
I. There is risk.
folio, Rf is the risk free rate, and u (d) is the proportional
II. Risk drives project value.
increase (or decrease) in the price of the underlying asset.
ˇ 12 M. Culík
t t + dt t t + dt t t + dt
The cost of setting up the replication portfolio consisting
of h units of the underlying asset and risk free borrowings at time t is:
= h · St + Bt, (2.1)
t
and in the absence of arbitrage opportunities it must hold n
that the value of the replication portfolio and the option’s
Figure 1 Discrete stochastic lattices (left-binomial, middle-
price must be identical:
trinomial, right-multinomial).
=
Ct = h · St + Bt. (2.2)
t
Real option valuation under changing volatility
If the underlying asset moves up at the time t + dt, it
holds for the portfolio value:
There are a lot methods and approaches in finance theory, u
u u dt
which are applicable for option pricing. These methods and
= = · + · + Ct+dt h St+dt Bt (1 Rf ) , (2.3)
t+dt
approaches range from analytical equations (Black—Scholes
model), lattice models (binomial, trinomial, multinomial),
and if it moves down:
simulation (Monte Carlo) to using partial-differential equa- d
d d dt
=
= · + · + tions (finite difference method). Ct+dt h St+dt Bt (1 Rf ) . (2.4)
t+dt
Generally, real options can be quantified by applying any
Under the assumption that the payoff of the European of these approaches. Due to specific features of real options
call option at maturity T is equal to its intrinsic value, discrete lattices are mostly employed. The reason is as fol-
u u u d d d
= = − = = − lows:
that is CT IVT max(ST X; 0) and CT IVT max(ST
X; 0); then, by solving the set of Eqs. (2.3) and (2.4) for
h
B
and and substituting this number into (2.2), we get for — easy calculation, interpretation,
the option’s price: — easily accommodate most types of real options problems,
— valuation of both plain vanilla (call, put) and exotic
dt d
· + −
dt u St (1 Rf ) St+dt (Bermudian, Asian, etc.) options,
· + = ·
Ct (1 Rf ) Ct+dt u d
− — managerial strategic decisions are made rather at dis-
St+dt St+dt
crete time moment than continuously,
dt
u — valuation of multinomial real options (more possible deci-
− · +
d St+dt St (1 Rf )
+ ·
Ct+dt u d , (2.5) sions are available at given time moment), −
St+dt St+dt
— valuation of real options with multiple sources of risk,
— valuation of real options with variable parameters
where (.) on the right-side of (2.5) represents the risk-
(changing volatility, exercise price, risk free rate, etc.).
neutral probabilities of up and down movements.
Formula (2.5) can be reduced to
u u d u −dt 1
Simulation via discrete lattice
= · + · − · +
Ct [Ct+dt p Ct+dt (1 p )] (1 Rf ) . (2.6)
It is apparent from (2.6) that the European option’s price
Discrete lattice is a stochastic process, where the stochas-
at t is equal to its expected payoff at the subsequent period
tic variable can change only after passing certain time-step
t
+ dt discounted at the risk-free rate.
(stepping time) and can take on given number of new val-
The procedure for the American options is similar to that
ues. A given time period (T − t), during which the stochastic
of the European options, i.e., we work back through the tree
process is simulated, is divided into finite number of time
from the end to the beginning, and, moreover, we are testing
steps where dt is the length of one discrete time step (time
at each node whether the early exercise is optimal. The
interval). For any discrete moment at time t has the stochas-
value of the American call option at maturity (end nodes of
tic process at t + dt (i.e. after passing stepping time) finite
the tree) is the same as for the European options; at earlier
possible number of values which can take on. According to
nodes, the option’s price is greater than its expected payoff
the number of values at the end of stepping time we work
at the subsequent period t + dt, discounted at the risk-free
with the following processes: binomial (at t + dt takes on
rate or the payoff (intrinsic value) IVt from early exercise,
two values), trinomial (at t + dt takes on three values) or
i.e., multinomial
(at t + dt takes on n values), see Fig. 1.
dt d u dt
· + − − · +
u St (1 Rf ) St+dt d St+dt St (1 Rf ) −dt
= + +
Ct max Ct+dt u d Ct+dt u d (1 Rf ) ; IVt ,
− −
St+dt St+dt St+dt St+dt and after simplification,
u u d u −dt
= · + · − · + Ct max[(Ct+dt p Ct+dt (1 p )) (1 Rf ) ; IVt].
1
(2.7) More details are available in Hoek and Elliot (2006).
Real options valuation with changing volatility 13
Simulation via binomial lattice is the simplest discrete t t + dt
.
process. Here, the value of underlying (random) asset takes u St
u
on at time t the value St; at the end of discrete interval (i.e. p
u
after passing stepping time) it can either jump up to St+dt or
d St
down to St+dt with some transition (risk-neutral) probability.
If the process can take on only positive values (typical for d u p = 1 - p .
d St
most financial variables), we work with geometric version,
where it holds for upward jump,
Figure 2 One period binomial lattice (geometric process).
u
S = u · S (3.1)
t+dt t t t + dt and downward jump, S + u pu t
d
= ·
St+dt d St (3.2)
and where u(d) are up and down factors. For these two fac-
rfdt
tors it holds following: u ≥ 1; 0 < d ≤ 1 and d < e < u. The d u
p = 1 - p
upward and downward factors u and d and transition (risk- St - d
neutral) probabilities are set uniquely in order to determine
Figure 3 One period binomial lattice (arithmetic process).
the evolution of underlying asset. Due to the fact that
expected return of any asset is assumed to be risk-free, the t t + dt t + 2dt
rf·dt
expected value at the end of discrete step equals St · e . 2 .
u St
It follows that the expected value of the underlying asset .
u St
can be written as,
·
rf dt u u . .
= · = · · + − · · E(St+dt) St e p u St (1 p ) d St (3.3) St d u St .
d St
and after some rearrangements, · 2 .
rf dt u u d St
e = p · u + (1 − p ) · d (3.4)
u Figure 4 Two-period recombining binomial lattice (geometric
From (3.4) the risk-neutral probability of upward jump p is process). given as,
rf ·dt
e − d
pu = (3.5) In situation, that the simulated process can take on both
−
u d positive and negative values, we work with arithmetic ver-
u d 2
= + = −
sion, where St+dt St u and St+dt St d, see Fig. 3.
and for downward jump must follows,
d = u
−
p (1 p ). (3.6) Simulation via binomial lattice with volatility
change
The variance of the underlying asset between two sub-
2
sequent discrete nodes at time t and t + dt is dt. And
If we assume the constant volatility over the period (T − t),
because the variance of random variable is generally given
2 2 2 the up and downward factors given according to (3.10) and
as (S) = E(S ) − [E(S)] , it is possible to write
(3.11) are constant throughout the whole lattice model and
2
2 u 2 u 2 u u due to (3.9) it holds u · d = d · u = 1, which follows in the result