CORPORATE FINANCE

Thomas E. Copeland and Philip T. Keenan

A lot, if uncertainty is high

But discounting cashflows is the wrong way to

calculate it

Instead, use options theory to value management’s

flexibility to act in the future

38 THE McKINSEY QUARTERLY 1998 NUMBER 2

 AVE YOU EVER BEEN INVOLVED in a capital investment decision where

H the net present value calculations proved negative, but the manage-
ment team decided to go ahead anyway? Or been confronted with a
positive NPV project where your intuition warned you not to proceed? Oƒten,
it is not your intuition that is wrong, but your time-honored NPV decision-
making tools.

But there is another way. Managers can use a diƒferent tool: real option value.
When a situation involves great uncertainty and managers need flexibility
to respond, ROV comes into its own. If the decision you face involves low
uncertainty, or you have no scope to change course when you acquire new
information later on, then NPV works fine. If not, you will want to know
more about what real options are and how to value them.

Below, we compare the main decision-making tools and show why traditional
techniques such as NPV, economic profit (EP),* and decision trees are
incomplete, oƒten misleading, and sometimes dead wrong. We also look at
how real options have been used in several practical situations, drawing on
simple examples for illustrative purposes rather than going into the mechanics
of valuing complicated real options.†

Real options began to be properly understood in 1973, when Fischer Black,

Myron Scholes, and Robert Merton devised rigorous “arbitrage-free”
solutions to value them. Applications have proliferated, particularly in
securities markets, where the theory held up remarkably well when tested
against actual prices. However, from our point of view 25 years later, the
assumptions of the Black–Scholes model seem somewhat restrictive when
applied to real options.

≠ Defined as the return on invested capital minus the weighted average cost of capital, multiplied
by the invested capital; sometimes known as economic value added.
≤ For a detailed account of this sort, see L. Trigeorgis, Real Options: Managerial flexibility and
strategy in resource allocation, MIT Press, Cambridge, Mass., 1996.
The authors wish to acknowledge the contributions of Sam Blyakher, Cem Inal, Max Michaels,
Yiannos Pierides, and Dan Rosner.

Tom Copeland is a former principal in McKinsey’s New York oƒfice and Phil Keenan is a
consultant in the Cleveland oƒfice. Copyright © 1998 McKinsey & Company. All rights reserved.

THE McKINSEY QUARTERLY 1998 NUMBER 2 39

HOW MUCH IS FLEXIBILITY WORTH?

Put simply, “arbitrage-free” means that securities with exactly the same
risk/return profiles should be identically priced. If you can describe the
payouts on one risky security and then build a portfolio of other securities
with exactly the same payout, the price of both must be the same. If the prices
were not identical, arbitrage, or buying the underpriced security and selling
the other, would be possible. This simple idea is at the heart of option pricing.

As yet, option pricing has not been much used in the evaluation of corpo-
rate investments, for three reasons: the idea is relatively new, the mathe-
matics are complex, making the results hard to grasp intuitively, and the
original techniques required the source of uncertainty to be a traded world
commodity such as oil, natural gas, or gold. Recent McKinsey research
has helped to overcome some of these obstacles. We now know how to
value several situations involving real options without using complicated
mathematics.

What are real options?

Skip this section if you are already familiar with options. It simply describes
what real options are and how to recognize them in a managerial rather than
securities setting.
Options are rights without obligations
An option is the right, but not the obligation, to buy (or sell) an asset at some
point within a predetermined period of time for a predetermined price.

The first account of a real option is found in the writings of Aristotle. He tells
of how Thales the Melesian, a sophist philosopher, divined from some tea
leaves that there would be a bountiful olive harvest in six months’ time.
Having a little money, he approached the owners of some olive presses and
bought the right to rent their presses at the usual rate. When a record harvest
duly arrived and the growers were clamoring for pressing capacity, he rented
the presses to them at above the market rate, paid the normal rate to their
owners, and kept the diƒference for himself – proving for all time that sophism
is not only an honorable profession, but a profitable one too.

What is the real option in this story? First of all, Thales purchased the right,
but not the obligation, to rent the presses. (He purchased a call option, the
right to buy or rent. The opposite is a put option, the right to sell.) Had the
harvest been poor, he would have chosen not to rent, and lost only his original
small investment, the price of the option.

Thales contracted for a predetermined rental price that in option pricing

terminology is called the exercise price. If the market price is higher than the
exercise price, the call option is said to be “in the money,” and Thales would

40 THE McKINSEY QUARTERLY 1998 NUMBER 2

 HOW MUCH IS FLEXIBILITY WORTH?

exercise it. If the market price is lower than the exercise price, then the call is
“out of the money,” and would not be exercised.

The underlying source of uncertainty in the story was the size of the olive
harvest, which aƒfected the market rental value of the presses. As the value of
the underlying variable increases, so does the value of the option. In other
words, the greater the harvest of olives to be pressed, the more valuable
Thales’ option to rent the presses will be.

The value of the option also increases with the level of uncertainty of the
underlying variable. The logic is straightforward. If there is no uncertainty
over the size of the olive harvest, which is known to be normal, then the
market rental value of the presses will also be normal, and Thales’ option
will be worthless. But if the size of the harvest is uncertain, there is a chance
that his option will finish in the money. The greater the uncertainty, the
higher the probability that the option will finish in the money, and the more
valuable the option.

So far we have mentioned three of the five variables that aƒfect the value of
the option. It increases with the value of the underlying variable and with its
uncertainty, and it decreases as the exercise price goes up. The fourth variable
is the time to maturity of the option. Thales purchased his option six months
before the harvest, but it would have been even more valuable two months
earlier, because uncertainty increases with time.

To see why, suppose that Thales has agreed to pay 10 drachmas per hour to
rent the presses, and the market rental price is also 10 drachmas. With only
one second to go before his option expires, it has no value. But with a month
to go, there is a good chance that the market value will rise above 10 drachmas
and the option will finish in the money. Therefore, the longer the time to
maturity, the more valuable an option is.

Finally, the value of the option increases with the time value of money, the
risk-free rate of interest. This is because the present value of the exercise cost
falls as interest rates rise.
Real options can be easy to overlook
One of the problems in learning how to use real options is that we oƒten
don’t know how to recognize them in real-life managerial settings. Here are
two examples.

In the 1960s, life insurance companies were vying to sign up baby boomers for
whole life policies. A feature oƒten included in the policies was the right to
borrow against the cash value of the policy at a fixed rate of interest, say
8 percent. At the time, with interest rates of 3 to 4 percent, this feature didn’t
seem important. But the insurance companies were unwittingly giving away

THE McKINSEY QUARTERLY 1998 NUMBER 2 41

 HOW MUCH IS FLEXIBILITY WORTH?

a lifelong call option on borrowing that could be exercised at a fixed interest

rate of 8 percent.

That option proved extremely valuable when interest rates soared to double
digits in the early 1980s. Suddenly, the baby boomers were able to borrow at
8 percent and invest at 12 percent, while the life companies had to borrow
at rates higher than 8 percent in order to honor their contracts. Many were
threatened with insolvency simply because they had no idea of the value of
the options they gave away.

The second example concerns a manufacturer of jet engines. In this highly

competitive industry, the secret is to get your engines onto the wings of
aircraƒt; that done, you have locked up the profits from a 30-year stream of
spare parts. What the manufacturer’s financial oƒficers did was to buy aircraƒt
and lease them, with their own engines on the wings, to airlines. They also
extended lease cancellation options that gave the airlines the right to cancel
delivery of the aircraƒt at any time before delivery and a year aƒter, for only a
small penalty payment.

The financial oƒficers wondered how much these cancellation options were
worth. Analysis revealed that they were worth on average 83 percent of the
value of the engine on narrow-body aircraƒt, and 19 percent on wide-body
aircraƒt (Exhibit 1). The finan-
Exhibit 1

The value of cancellation options cial oƒficers were horrified.

What should they do?
Example: Jet engine manufacturer
Percent of engine price
Wide-body Narrow-body
aircraft aircraft Their new understanding of
Pre-delivery cancellation
option
16 real options showed them
58

Post-delivery 3 25 that the cancellation option

cancellation option
was most valuable to airlines
Total 19 83
that experienced high vari-
ability in demand. They stop-
ped oƒfering lease cancellation options to these airlines. A year or so later,
passenger revenue miles fell steeply throughout the industry. Thanks to its
change of policy, the company saved tens of millions of dollars.
Options or bets?
Once you understand what real options are, you begin to realize that they are
embedded in a whole range of management decisions. Options are everywhere.
But it is important to know the diƒference between a bet and an option.

A situation where there are real options involves uncertainty about two things.
One is the future; the other is the ability of management to respond to what
it learns as the picture gradually becomes clearer. If management cannot
respond in a material manner to new developments, the situation represents
a bet, not an option.

42 THE McKINSEY QUARTERLY 1998 NUMBER 2

 HOW MUCH IS FLEXIBILITY WORTH?

Suppose a company must decide whether to invest $100 million in con-

structing a new factory in the face of uncertainty about future demand for the
product it will produce. If there are no follow-up investments – if it is an “all
or nothing” decision – then management faces a bet, not an option. It can
put up the money, roll the die, and win or lose.

But suppose the $100 million investment can be recast as $10 million spent
now on a pilot project and $110 million invested in a year’s time to build the
factory if it still seems like a good idea. Under this scheme, paying the $10
million immediately gives management the option to proceed with the $110
million investment a year from now, provided the pilot produces the right
results. If it fails, management has no obligation to proceed with the project.

Even though building the factory costs more this way – $110 million in present
value at a 10 percent cost of capital, rather than $100 million – it may make
good economic sense. In particular, if the uncertainty surrounding demand
for the product is high, so that the correct decision may well be not to go
ahead, the option may be worth much more than the cost of creating it.

How real options capture the value of flexibility

Real options capture the value of managerial flexibility in a way that
net present value analysis does not. Consider the example described above.
Our aim is to illustrate concepts, not describe the methodology, so we will
deliberately simplify the calculation.

Suppose there is a 50 percent chance that aƒter investing $100 million in the
new factory, management will be rewarded with strong sales for many years.
Revenues exceed costs, and the factory produces an operating income stream
with a present value of $150 million.

On the other hand, suppose there is a 50 percent chance that demand is poor
and the present value of the operating income stream is only $10 million.

A traditional NPV analysis of this bet would put the expected present value
of the future operations of the factory at $80 million (the average of $150
million and $10 million). This is not enough to oƒfset the upfront investment
of $100 million, and the project has an NPV of minus $20 million. It will
almost certainly be thrown out.

But rather than invest the whole $100 million up front, what if management
elects to buy the option to expand, at the price of $10 million for the pilot?

There is a 50 percent chance that the pilot succeeds; management responds by

building the factory (for $110 million) and reaping profits of $150 million, as

THE McKINSEY QUARTERLY 1998 NUMBER 2 43

HOW MUCH IS FLEXIBILITY WORTH?

before. However, both building the factory and obtaining the profits are delayed
a year because of the pilot, so we discount them both back one year at 10 percent
to a $100 million investment and a $135 million profit, or a net $35 million gain.

There is also a 50 percent chance that the pilot fails, in which case man-
agement halts the project with no further outlay.

The overall value of the project (ignoring for simplicity’s sake subtle points
about investor risk sensitivity and the degree of correlation between the
project and the market) is thus $7.5 million (the average of $35 million and
zero, minus the upfront $10 million). Management should indeed proceed
with the pilot.

What we have so far is a classic decision tree analysis. So how does real option
valuation diƒfer from decision tree analysis, and which is best?

Decision trees and real option valuation are closely related: if you can
implement the first, it is not much work to implement the second. Decision
tree analysis involves building a tree representing all possible situations and
the decisions management can make in response to them. To value a decision
tree, one calculates expected cashflows based on their objective probability
and then discounts them at some chosen rate – usually the weighted average
cost of capital.

Option valuation diƒfers from decision tree analysis in calculating values in

accordance with the “no arbitrage” principle, or law of one price. This states
that two diƒferent investment opportunities that produce the same (equally
uncertain) payoƒfs must be worth the same amount; otherwise, arbitrageurs
would buy the undervalued investment and sell the overvalued investment,
making a risk-free profit in the process.

The option approach can be interpreted in the decision tree context as

modifying the discount rate to reflect the actual riskiness of the cashflows.
A call option, for instance, corresponds to a leveraged position in the under-
lying asset, and is therefore by definition riskier than the asset.* As a result, the
appropriate discount rate is considerably higher than the weighted average
cost of capital. Moreover, it changes throughout the decision tree depending
on how far the option is in or out of the money.

Decision tree methodology gives no guidance on how to choose the discount

rate or adjust it for risk or leverage. Traditional decision tree analysis using the

≠ Imagine a given stock price is $20 and the exercise price of the call option is $15. The call option
will be worth roughly the diƒference between the two, namely $5. If the stock price goes down by
$1, a 5 percent change, the option value will fall by $1, a 20 percent change. Therefore the option
is riskier than the stock.

44 THE McKINSEY QUARTERLY 1998 NUMBER 2

 HOW MUCH IS FLEXIBILITY WORTH?

weighted average cost of Exhibit 2

Real options valuation and decision tree analysis

capital as the discount rate
can thus lead to false results. Valuation of sample options on stocks, $

European call option* American put option†

To compare decision trees ROV 48 27

and options on a level play- DTA 116 11

ing field, we applied the two DTA overvalues the DTA undervalues the
option by 140% option by 58%
approaches to the pricing of Correct DTA would use Correct DTA would use
financial options – a call and discount rate varying discount rate varying
from 232% to 21% from –83% to –13%
a put on a stock – so that throughout the tree throughout the tree
questions of investor risk pref- * 2 years to expiry, 10% volatility, current price $20, exercise price $24, risk-free rate 5%,
WACC 10%, 18 time step decision tree
erences and how to quantify † As above except exercise price now $19

uncertainty and managerial

flexibility would not cloud the comparison. Although in some situations the
approaches give similar results, decision trees can be wrong by a factor of
two, as Exhibit 2 shows.

Why traditional tools are inadequate

Managerial corporate finance theory has been struggling for years with the
question of how to evaluate investments under uncertainty. The “obvious”
approach of comparing the costs and benefits of an investment is actually
highly complex, since costs are incurred today with certainty while benefits
are uncertain and reaped in the future. In practice, managers use a range of
methodologies including earnings per share or per share growth, economic
profit, decision trees, discounted cashflow, and real options.

Until recently, discounted cashflow and economic profit were the most
popular approaches. The DCF method forecasts future cashflows, discounts
them at a risk-adjusted rate (the weighted average cost of capital), and
subtracts the current investment cost to estimate the net present value of a
project. Projects with positive NPV are said to create value and are accepted;
negative NPV projects are thrown out.

Advocates of this approach point out that it fulfills important criteria: it is

cashflow based, risk adjusted, and multi-period or forward looking. However,
as we saw earlier, it does not Exhibit 3

capture flexibility. Exhibit 3 Key criteria for decision-making tools

compares the methodologies Cashflow Risk Multi- Captures
using these criteria. based adjusted period flexibility
Real option value
NPV/DCF
DCF techniques were ori-
Decision trees
ginally developed in order
Economic profit
to value investments such
Earnings growth
as stocks and bonds, and

THE McKINSEY QUARTERLY 1998 NUMBER 2 45

 HOW MUCH IS FLEXIBILITY WORTH?

assume that companies hold investments passively. They overlook manage-

ment’s flexibility to alter the course of a project in response to changing
market conditions. In eƒfect, they assume that management makes an
irrevocable decision based on its view of the future, and then does not
deviate from its plan no matter how things actually shape up. The life of the
project is assumed to be fixed, and the possibility of abandoning it in the
face of adverse circumstances or, conversely, expanding it in response to
unanticipated demand is not even considered. Such rigid assumptions
are rather like planning a drive from New York to Los Angeles with only
fragments of a map, and then sticking firmly to your original route even as
you see highway signs and find out what traƒfic and road conditions are like.*

Both real options and decision trees capture the mechanics of flexibility.
However, only options adjust for risk.

Horses for courses

Real option valuation is most important in situations of high uncertainty where
management can respond flexibly to new information, and where the project
value without flexibility is near breakeven. (If the NPV is very high, the project
will go full steam ahead, and flexibility is unlikely to be exercised. And if the
NPV is strongly negative, no amount of flexibility will help.) Optionality is of
greatest value for the tough decisions – the close calls where the traditional NPV
is close to zero (Exhibit 4).

Consider two investments: a new brewery and a pharmaceutical R&D program.

The brewery is a one-oƒf investment in a fairly stable environment in which
demand can be forecast reasonably confidently. If the brewery’s operating
Exhibit 4 margins are high, its NPV will be high. The
When managerial flexibility is valuable only alternative to going ahead is deferral,
Uncertainty which is unlikely since there is little uncer-
Likelihood of receiving
new information tainty about when new capacity will be needed
Low High or what the operating margins will be. DCF
Room for
Moderate
High flexibility
High
flexibility
methods will work well in this setting because
managerial value value their implicit assumptions are valid.
flexibility
Ability to Low Moderate
respond Low flexibility flexibility
value value
The pharmaceutical R&D program is another
Flexibility value greatest given
1. High uncertainty about the future (very likely
matter. Investment is needed at several stages,
to receive new information over time) with substantial outlays on basic research,
2. Much room for managerial flexibility (allows
management to respond appropriately to this developmental testing, clinical testing, and
new information) product rollout. At each stage, management
3. NPV without flexibility is near zero (if a project
is neither obviously good nor obviously bad, can choose to abandon the project, defer it, go
flexibility to change course is more likely to
be used and therefore more valuable) ahead as planned, or spend more to accelerate
Under these conditions, the difference between
ROV and other decision tools is substantial
≠ We thank Sam Blyakher for this example.

46 THE McKINSEY QUARTERLY 1998 NUMBER 2

 HOW MUCH IS FLEXIBILITY WORTH?

it. Flexibility is high, as is uncertainty, and the Exhibit 5

Profitability in PC assembly
value of the project without flexibility may be
marginal. In this setting, most of the assump- Percent, 1995–96 Estimated spread
(ROIC–WACC)*
tions of DCF do not hold. Option valuation Gateway 30
will give much better results. Acer –2
Compaq –4
In 1995–96, the PC assembly business was in Apple –7
turmoil. Gateway was one of the few players Packard Bell –11
making money. Exhibit 5 shows the estimated * For consumer portion of assembly businesses, estimated
from publicly available figures for the consolidated
spread between return on invested capital companies

(ROIC) and weighted average cost of capital

(WACC) for a number of major players. Amid tremendous uncertainty about
how the industry would shake out, managers had to decide whether to exit or
stay as their businesses bled cash. The standard decision tools suggested they
should exit immediately. Those businesses with gross operating margins of
11 percent or below have negative DCF and EP values (Exhibit 6).
Exhibit 6

Different methods, different values

Example: PC assembly
$ billion, 1997
Gross One-year DCF ROV
operating EP
margin
15% –0.05 2.62 2.98
13% –0.07 1.02 1.71
11% –0.09 –0.59 0.79
9% –0.11 –1.79 0.36

Suppose we focus on a business with an operating margin of 11 percent. Its

EP is minus $90 million, and DCF minus $590 million. Yet its real option
value is $790 million. Why the huge diƒference?

DCF overlooks flexibility. In particular, it ignores the possibility of exiting

and reentering the industry. If the costs of doing so were trivial, the best
strategy would be to exit immediately and reenter if conditions improve. But
exit and reentry costs are high. Managers will thus decide to stay in an
industry despite a negative DCF because there is tremendous uncertainty
about what operating margins will be like in the future, and because exit and
reentry costs are high. ROV not only captures the value of flexibility, but also
indicates how long a company should stay in a business before exiting, and
when to reenter.

Classifying real options

Individual real options can be classified into growth options (scaling up, switching
up, or scoping up a project), deferral/learning options, and abandonment options
(scaling down, switching down, or scoping down a project) (Exhibit 7).

THE McKINSEY QUARTERLY 1998 NUMBER 2 47

 HOW MUCH IS FLEXIBILITY WORTH?

Exhibit 7
The seven basic real options
7S framework: Grow, defer, or quit
can also occur in combina-
Scale
Early entrants can scale up later
through cost-effective sequential
tions, as compound options.
up investments as market grows A company that invests in an
Speedy commitment to first R&D project, say, may be
Invest/ Switch generation of product or technology
grow up gives company preferential position buying both the option to
to switch to next generation
commercialize the resulting
Investments in proprietary
Scope assets in one industry enables product and the option to
up company to enter another
industry cost-effectively engage in subsequent R&D
projects to develop future
Defer/ Study/ Delay investment until more
learn start information or skill is acquired generations of related prod-
ucts. These subsequent R&D
Shrink or shut down a project part
Scale
way through if new information projects themselves contain
down
changes the expected payoffs options for commercializa-
Disinvest/ Switch Switch to more cost-effective and tion and further develop-
flexible assets as new information
shrink down
is obtained ment, leading to a type of
compound option called a
Limit the scope of (or abandon)
Scope
down operations when there is no further growth staircase.
potential in a business opportunity

Real options can also depend

on more than one source of uncertainty. The value of an option to
commercialize an R&D project, for instance, depends on at least two:
technological uncertainty (will the scientists succeed in inventing the new
product?), and market uncertainty (what will the demand for this new
product be?). Options that depend on multiple sources of uncertainty are
oƒten called rainbow options.
A deferral option (natural resource development)
Real options played an important role in a decision made by one coal-mining
company. It needed to work out how much to bid for the lease of a plot of
land that could be developed into a coal mine. Using the current price of coal
and extrapolating its growth into the future, and forecasting extraction costs,
taxes, and the estimated quantity of coal in the mine, the company calculated
the NPV of the cashflows involved in developing the mine and selling the
coal, and concluded that the lease was worth $59 million – not very much.

However, the company knew that the price of coal could fluctuate sub-
stantially. As its current price was close to the project’s breakeven point, the
revenue projections were highly sensitive to future price changes.

The company realized that acquiring the lease would give it an option: to
defer opening the mine until such time as the price of coal rose far enough to
make the project’s economics reasonable. This option turned out to be worth
$57 million – almost as much as opening the mine immediately. When the
value of the option was factored in, the lease was actually worth $116 million.

48 THE McKINSEY QUARTERLY 1998 NUMBER 2

 HOW MUCH IS FLEXIBILITY WORTH?

The company successfully bid $72 million, waited until the price of coal rose,
and made a tidy profit from the mine.
Learning options (natural resource development and R&D)
Learning options arise when a company is able both to spend money to speed
up its acquisition of important information (for example, to reduce tech-
nological uncertainty in R&D or to learn more about the quantity of
resources in the ground in exploration and development projects) and to use
what it has learned about the market demand for the project output to modify
future investment decisions. It must balance the value of the option to act
on the information learned against the cost of acquiring that information.

Take the company considering a site for a mine. It must weigh the value of
deferring development of the mine against the value of the information it
could gain about the quantity of ore in the mine as a result of full or partial
development. This is an example of a rainbow option. The sources of
uncertainty are the price of coal and the quantity of coal in the mine.

Deferral is valuable because of the uncertainty surrounding the price of the

coal the mine will eventually sell. If the economics are presently near
breakeven, waiting gives management the chance to react to price shiƒts.
However, partial development is also valuable because it reduces uncertainty
about the quantity of coal in the mine, while preserving management’s ability
to adjust future investment according to what is learned. Partial development
thus represents a learning option that is in conflict with the deferral option;
the company cannot exercise both.

Multi-stage R&D projects generally contain a series of embedded options

based on technological and market uncertainty. They too are learning options.
Undertaking an R&D project gives management the right, but not the
obligation, to commercialize a product if and when the R&D eƒfort is
successful and the economics of producing and marketing the product are
attractive. Although an R&D project viewed in isolation may have a negative
NPV, the option to commercialize the result is oƒten extremely valuable –
enough to determine that the project be undertaken anyway.

Making irreversible investment decisions in the face of uncertainty is risky.

Being able to change a decision as new information becomes available helps
reduce the risk. Traditional decision-making tools such as NPV, EVA, and
earnings per share neglect the value of such flexibility. Real options, on the
other hand, provide a theoretically sound tool for valuing management’s
strategic scope. Recent advances in theory have made ROV techniques
applicable to a multitude of real-world situations. At the same time, advances
in technology have enabled option pricing capability to move out of Wall
Street and into mainstream corporations.

THE McKINSEY QUARTERLY 1998 NUMBER 2 49