In the late 90's, real options was one of the hottest buzzwords in management science. Companies were in
a hurry to "do" real options, often hiring consultants or even whole departments in a race to use the
most advanced decision-making methods. The authors of DPL played a leading role in the real options
revolution, pioneering the practical, decision tree based approach to valuing real options.
Today, in the post-Enron world, real options is regarded with far
more skepticism, and is sometimes associated with the excesses of the bubble period. As the dust
settles, analysts are finding that real options techniques, judiciously applied, are a very good way
to understand and quantify the value of management flexibility. Real options analysis is now often
integrated with decision and risk analysis to form a more complete picture -- one with an upside and a
downside -- of the value and risk in an investment.
What is a real option?|
An option is the right, but not the obligation, do something (usually buy or sell some
asset) at some time in the future after learning about uncertainty. The source of uncertainty is called
the underlying. A real option is an option where the underlying is not traded in
the financial markets.
In business areas with a high degree of uncertainty, management, knowingly or not, creates and
exercises real options on a regular basis. For example, R&D creates real options: management has the right,
but not the obligation, to commercialize the results. In a decision tree, real options are represented by
"downstream" decisions -- that is, decision nodes which follow one or more chance nodes. The picture at
the right shows a simple real option modelled in DPL.
How can real options be quantified?|
Techniques for quantifying real options fall broadly into two families. The first family involves
the use of methods from financial option pricing theory, such as the Black Scholes equation and
binomial lattices. These techniqes are appropriate for real options which are very similar to
financial options. For example, the real option to acquire a controlling stake in a non-traded
company that owns a gold mine might be very similar to a call option on a gold contract. The
second family uses methods that account for multiple and varied sources of uncertainty, such
as decision tree analysis and Monte Carlo simulation. These techniques are appropriate for more
"real world" problems where the uncertainties don't follow well-defined mathematical processes.
Examples include early-stage companies, R&D projects and capital investments in politically unstable
countries. This article gives tips for using DPL to implement the second family of techniques.
Simplistic real options models handle uncertainty in large chunks, with a single chance node
representing all the uncertainty in a given value driver. For events that play out over time, such
as sales over the lifecycle of a product, a more realistic formulation breaks the uncertainty into
several parts, allowing finer grained calculations of option value. A representation of
the way uncertainty is resolved over time is called a learning model. The tree at the
far right shows a two-period learning model on the size of a market. We see the rate of
adoption in the first year,
then make a decision about increasing production capacity, then resolve the longer-term uncertainty
in ultimate market size.
Some of the simplest learning models use Bayesian relationships. For example, a market test may
give an indication of whether a product will be a hit, but some hot products may have done poorly
in similar market tests. With DPL, you can assess the relationship in either order (e.g., the input probabilities
in the picture below show that the outcome of the market test is influenced by the actual market results -- this relationship may have been
ascertained from past data or experience). However, the actual order of events is: decide to
conduct a market test, observe its results and then decide to launch or not. With DPL, you can reverse the
uncertainties in the tree and DPL will "flip" the probabilities as necessary. The picture below shows the input
probabilities and Policy Tree for this market test example.
Link to the value of information|
If you've studied decision analysis, you may be thinking that this talk of learning models sounds
an awful lot like the value of information. Mathematically, the value of information and the value
of optionality often work out to be the same problem. Just as the value of imperfect information is
more subtle and often more relevant than the value of perfect information, a learning model can give
a more realistic estimate of option value than a simpler, all-or-nothing formulation of uncertainty.
Calculating the value of optionality in DPL|
Calculating the value of optionality in DPL is simple. First, build a tree which includes the
option as a downstream decision and run a decision analysis. Next, temporarily disable the option
using Branch Control, and run another decision analysis. The difference in expected value between
the two runs is the option value. For an introduction to the Branch Control feature, see the
"Timing and Structure" tutorial in the User Manual. In the picture at left, the Production Capacity
decision is controlled to High, disabling the optionality.
Where does the option value come from?|
Showing how an option creates value is as important as calculating its value. Decision-makers
need to see what the option means in business terms -- when is it exercised, what is the cash flow
impact -- if they are to include it in their view. DPL provides several outputs which can shed light
on the sources of option value. For example, in the risk profile graph shown at right, we can see
how the exit option (red line) cuts downside risk.
Other key questions about the option, such as when and how often it is exercised, can be answered
by DPL's Policy Tree and Policy Summary outputs. See the Market Entry
case study for an illustration of these outputs.
The importance of speed|
Real options models are often computationally intensive. A decision tree with learning can have
many more nodes than one which considers uncertainty only in one time period. Moreover, real world
problems often come with huge cash flow spreadsheets.
While these models are tough, DPL's proprietary algorithms and special performance features can
make them surprisingly tractable. Keeping model runtimes short allows for more (human) iteration,
leading to new insights and a higher level of comfort with the results.
Click here for suggestions on how to
speed up large models.
Real options techniques can bring insight to the analysis of investments with a high degree of
uncertainty. DPL provides a rich set of features for dealing with complex, real world real options
problems -- from building the model to making sense of the results. DPL's unique combination of
power, flexibility and transparency has made it the professional's choice for real options
analysis. To find out more about how DPL and real options analysis can contribute to your
organization, contact Syncopation.
Download a demo |