Analysis of Real Options in Hydropower Construction Projects

-- A Case Study in China

by

Tao Wang

B.S. Electrical Engineering

Fudan University, Shanghai, China, 1996

M.S. Economics
Fudan University, Shanghai, China, 1999

Submitted to the Engineering Systems Division

in Partial Fulfillment of the Requirements for the Degree of

Master of Science in Technology and Policy

at the

Massachusetts Institute of Technology

August 2003

2003 Massachusetts Institute of Technology

All rights reserved.

Signature of Author……………………………………………………………………………………………………...
Technology and Policy Program, Engineering Systems Division
May 13th, 2003

Certified by……………………………………………………………………………………………………………….
Richard de Neufville
Professor of Engineering Systems and of Civil and Environmental Engineering
Thesis Supervisor

Accepted by………………………………………………………………………………………………………………
Dava J. Newman
Associate Professor of Aeronautics and Astronautics and Engineering Systems
Director, Technology and Policy Program
 2

Analysis of Real Options in Hydropower Construction Projects

-- A Case Study in China

by

Tao Wang

Submitted to the Engineering Systems Division

On Aug 15th, 2003 in Partial Fulfillment of the
Requirements for the Degree of Master of Science in
Technology and Policy

Abstract

This thesis investigates the application of real options to the design of major civil works.
Specifically, it examines several methods to value the options, to determine which are
more suitable in the design context.

The thesis applies NPV, NPV with simulation, and binomial options pricing model to study
a case on Yalongjiang River basin development. More specifically, a deferral option of
Project A is studied. When doing the real options analysis, the thesis compares the
usage of the NPV of project and the electricity price as the underlying and finds that the
electricity price is a more appropriate underlying for the options analysis.

On the pros side, the study confirms that options analysis overcomes the inadequacy of
NPV analysis in uncertain environments, at least partly. Options analysis can evaluate
the flexibility intrinsic or built into projects facing uncertain environments. Moreover,
options analysis develops contingency strategy in uncertain environments.

On the cons side, traditional options analysis requires the geometric Brownian motion
assumption of the uncertainties and relatively high quality of data. The applicability of
 3

options analysis depends on the type of uncertainty of the environment and the data
availability. The deferral option in Project A fits in the category where options analysis is
applicable.

The thesis concludes that real options analysis using the electricity price as underlying is
an appropriate method for valuing the deferral options of Project A and similar
hydropower projects.

Thesis Supervisor : Richard de Neufville

Title: Professor of Engineering Systems and of Civil and Environmental Engineering
 4

Acknowledgements

My deepest thanks to Professor de Neufville for his guidance and help. I remember
clearly when he helped me to find the exciting topic to apply real options in river basin
development when I almost lost my confidence in studying and researching at MIT. In
Chinese culture, a teacher has the same status and receives the same respect as father.
My father is a professor. The two professors, my father and Prof. de Neufville, have
directed me through the most important points of my life and career development!

My special thanks to my friends in China. They helped me gather important information,

taught me many real world practice and knowledge, and facilitated my life in China.
Without their generous help, this thesis would not have been possible. And I do not know
how I can fully thank all the friends and reward them.

My sincere thanks to my family. My parents have supported me unwaveringly during my

study in the US. As the only child of the family and when my mother is sick, all their
understanding and encouragement helped me overcome various difficulties.
 5

Table of Contents:
Abstract __________________________________________________________________ 2
Acknowledgements _________________________________________________________ 4
Table of Contents: __________________________________________________________ 5
List of Figures: _____________________________________________________________ 9
List of Tables: ____________________________________________________________ 11

Chapter 1 Introduction _________________________________________________________ 12

1.1. General Methods and Tools _______________________________________________ 12

1.2. Theme and Structure of the Thesis _________________________________________ 13

Chapter 2 Literature Review_____________________________________________________ 14

2.1. Real options ____________________________________________________________ 14

2.1.1. Development of the methodology ________________________________________ 14
2.1.2. Application of the methodology__________________________________________ 16
2.1.3. Real Options applied in Energy and Natural Resources _______________________ 22

2.2. Water Resource Planning with regard to Facilities Design _____________________ 23

2.2.1. Historical development ________________________________________________ 23
2.2.2. Other important references______________________________________________ 26

2.3. Conclusions from the literature search ______________________________________ 27

Chapter 3 Options _____________________________________________________________ 29

3.1. Financial Options _______________________________________________________ 29

3.2. Cornerstones for Options Valuation ________________________________________ 31

3.2.1. No Arbitrage_________________________________________________________ 32
3.2.2. Brownian motion and Weiner Processes ___________________________________ 35

3.3. Options Valuation Tools__________________________________________________ 41

3.3.1. The Black-Scholes Model:______________________________________________ 42
3.3.2. Valuation by Simulation _______________________________________________ 46
 6

3.3.3. Binomial Real Options Valuation ________________________________________ 48

3.4. Risk-neutral Valuation ___________________________________________________ 51

3.5. Options Valuation and Decision Tree Analysis _______________________________ 52

3.6. Real Options ___________________________________________________________ 57

3.6.1. What is Real Options Method? __________________________________________ 57
3.6.2. Comparison of Real Options Method and Traditional NPV Method______________ 59
3.6.3. Types of Real Options _________________________________________________ 61
3.6.4. Some Implications of Real Options Method ________________________________ 63
3.6.5. Framework of Real Options Valuation ____________________________________ 67
3.6.6. Real Options versus Financial Options ____________________________________ 68

Chapter 4 Yalongjiang River Basin _______________________________________________ 70

4.1. Introduction of Sichuan __________________________________________________ 70

4.2. Sichuan power market ___________________________________________________ 73

4.3. Introduction to Yalongjiang River _________________________________________ 75

4.4. Dams built and proposed _________________________________________________ 77

4.4.1. Ertan _______________________________________________________________ 77
4.4.2. Project A ___________________________________________________________ 78
4.4.3. Project B____________________________________________________________ 79
4.4.4. Project C____________________________________________________________ 80

Chapter 5 Valuing Project A ____________________________________________________ 82

5.1. Traditional NPV ________________________________________________________ 82

5.2. Traditional Internal Rate of Return (IRR)___________________________________ 84

5.3. NPV with Simulation ____________________________________________________ 84

5.3.1. Technical Sub-Model __________________________________________________ 87
5.3.2. Market Sub-Model ____________________________________________________ 88
5.3.3. Simulation __________________________________________________________ 92

5.4. IRR with simulation _____________________________________________________ 96

 7

5.5. Real Options Analysis____________________________________________________ 97

5.5.1. A General Binomial Tree Framework for ROA______________________________ 97
5.5.2. ROA with electricity price as underlying___________________________________ 99
5.5.3. ROA with NPV as underlying __________________________________________ 111

5.6. Decision Tree Analysis __________________________________________________ 113

5.6.1. Scenarios on Condition Branches: _______________________________________ 114
5.6.2. Payoffs ____________________________________________________________ 116
5.6.3. Result _____________________________________________________________ 116
5.6.4. Shortcomings of Such an Analysis_______________________________________ 118

5.7. Summary of Results ____________________________________________________ 119

Chapter 6 Policy Implications – Options Thinking in Practice ________________________ 121

6.1. Making Real Impact of the Options Thinking _______________________________ 123

6.1.1. Intrinsic Difficulties in Promoting Real Options methodology in Developing Countries
_______________________________________________________________________ 124
6.1.2. My Strategy in making real impact of options thinking “in” hydropower projects on
Yalongjiang River ________________________________________________________ 126
6.1.3. Organizational Wisdom to Advocate New Thinking _________________________ 133

6.2. Uncertain Reality and Appropriate Method ________________________________ 138

Chapter 7 Conclusions ________________________________________________________ 141

7.1. ROA with electricity pricing as underlying is appropriate for Project A _________ 141

7.2. Comparison of Methods _________________________________________________ 143

7.3. Future Work __________________________________________________________ 145

7.3.1. Identifying Options __________________________________________________ 145
7.3.2. Modeling __________________________________________________________ 146
7.3.3. Obtaining Insights ___________________________________________________ 148

Appendix I Ito’s Lemma ______________________________________________________ 149

Appendix II China Steps Up Reform of Electric Power System ________________________ 150

 8

References __________________________________________________________________ 151

 9

List of Figures:

Figure 3-1 Payoff Diagram for a Stock Call Option ___________________________________ 31

Figure 3-2 Stock Price Movement in Numerical Example ______________________________ 33
Figure 3-3 One Path of Brownian Motion___________________________________________ 37
Figure 3-4 Two Hundred Paths of Brownian Motion __________________________________ 37
Figure 3-5 Lognormal Density ___________________________________________________ 40
Figure 3-6 Distribution of Stock Price for Valuation by Simulation_______________________ 47
Figure 3-7 Distribution of Exercise of Option for Valuation by Simulation _________________ 47
Figure 3-8 Stock and Option Price in a One-step Binomial Tree _________________________ 48
Figure 3-9 A Two-Step Binomial Tree______________________________________________ 50
Figure 3-10 Decision Tree for Options Valuation_____________________________________ 53
Figure 3-11: Applicability of Real Options Method ___________________________________ 66
Figure 3-12 Framework of Real Options Method _____________________________________ 67
Figure 4-1: Sichuan in China ____________________________________________________ 70
Figure 4-2: Sichuan Energy Composition ___________________________________________ 72
Figure 4-3 Trend of Electricity Power Supply and Demand _____________________________ 74
Figure 4-4: Map of the Yalongjiang River Basin _____________________________________ 76
Figure 4-5: Picture of Ertan Dam _________________________________________________ 77
Figure 4-6: Project B Tunnel Parameters___________________________________________ 80
Figure 5-1 Engineering Systems Model with Uncertainties _____________________________ 86
Figure 5-2 Yalongjiang River Basin Model with Uncertainties __________________________ 86
Figure 5-3 Payoff Diagram of Longing a Stock and Shorting a call_______________________ 87
Figure 5-4 Power Generation and Waterflow ________________________________________ 88
Figure 5-5 Assumed Power Supply and Demand Curve in Sichuan _______________________ 89
Figure 5-6 Distribution of NPV for Project A ________________________________________ 94
Figure 5-7 Distribution of IRR for Project A ________________________________________ 96
Figure 5-8 Analysis of a scenario on the event tree ___________________________________ 98
Figure 5-9 Capital Market Line__________________________________________________ 102
Figure 5-10 Electricity Price Movement ___________________________________________ 106
Figure 5-11 Binomial Tree for ROA with electricity price as underlying __________________ 108
 10

Figure 5-12 Option Value Contour _______________________________________________ 110

Figure 5-13 Binomial Tree for ROA with NPV as underlying___________________________ 113
Figure 5-14 Basic Structure of the Decision Tree ____________________________________ 114
Figure 5-15 Decision Tree Analysis ______________________________________________ 117
Figure 6-1 Distribution of NPV without options _____________________________________ 122
Figure 6-2 Distribution of NPV with options _______________________________________ 122
Figure 6-3 Contribution of my work ______________________________________________ 132
Figure 7-1 A binomial Tree _____________________________________________________ 147
 11

List of Tables:

Table 2-1: Different Valuation Approaches with Examples _____________________________ 19

Table 3-1 Option Valuation by Decision Tree Results _________________________________ 56
Table 3-2 Revenue Estimate for Technology Development ______________________________ 60
Table 3-3 NPV Valuation of Technology Development_________________________________ 60
Table 3-4 Approximate Options Valuation of Technology Development ___________________ 60
Table 3-5 Types of Real Options __________________________________________________ 62
Table 4-1: World Hydropower Capacity ____________________________________________ 72
Table 4-2 Sichuan Electric Power Generation Composition ____________________________ 73
Table 4-3: Project A Design Alternatives ___________________________________________ 78
Table 4-4: Project A Adding Capacity to Ertan ______________________________________ 78
Table 4-5: Streamflow for Project A _______________________________________________ 79
Table 4-6: Project C Design Alternatives ___________________________________________ 81
Table 4-7: Project C Adding Capacity to Downstream Stations__________________________ 81
Table 4-8: Streamflow for Project C _______________________________________________ 81
Table 5-1 Project A NPV Analysis_________________________________________________ 83
Table 5-2 Sensitivity Analysis of Option Value ______________________________________ 109
Table 5-3 Using Binomial Tree to Get Probability for Decision Tree ____________________ 115
Table 5-4 Summary of Results ___________________________________________________ 119
Table 6-1 Uncertain Reality and Appropriate Method ________________________________ 140
Table 7-1 Comparison of the two ROA methods studied in this Thesis____________________ 142
Table 7-2 Comparison between Methods __________________________________________ 144
 12

Chapter 1 Introduction

The future is inherently unknown, but the unknown is not unmanageable. It is fascinating
to explore designing large systems with full awareness of uncertainties. And no
systematic work has been done yet, at least in the area of river basin development. This
research develops a preliminary methodology for incorporating options thinking into the
valuation of river basin development projects, and compares the methodology with other
methodologies such as NPV and NPV with simulation.

1.1. General Methods and Tools

The real options approach is based on the seminal work of Black, Scholes, and Merton
that won the Nobel Prize in 1997. Modern Finance theory provides abundant
understanding of uncertainties and how to deal with them. It also provides some very
useful techniques that engineers can learn to better "optimize" designs in an uncertain
world.

Binomial tree and risk-neutral valuation should be introduced to engineers. Cox, Ross,
and Rubinstein developed the Binomial Tree Model in 1979. The model represents how
the value of an asset evolves where the value is monitored under successive short
periods of time. In each short period it is assumed that only two value movements, up or
down, are possible. Risk-neutral valuation, assuming the world is risk neutral, gives the
correct value for all worlds, not just in a risk-neutral world.

Modern computer technology removes most of the previous computing capacity problems
for system designers. People are able to establish huge optimization models on personal
 13

computers. But are they fully utilizing the computing capacity they did not possess before?
Taking into account uncertainties systematically and productively in systems design has
been made possible by the modern computer technology. The key to the success of this
study is that, based on traditional optimization models, a simulation model with options
thinking can be developed.

1.2. Theme and Structure of the Thesis

This thesis studies the Project A of Yalongjiang river basin development to determine the
most appropriate valuation method for this project given the various uncertainties. The
valuation methods studied include NPV, NPV with simulation, IRR, IRR with simulation,
decision tree analysis, and real options (using electricity price or project value as
underlying).

The structure of the thesis is as follows: Chapter 2 reviews literature on real options and
water resource planning. Chapter 3 introduces options, both financial and real. Its focus
is on the options valuation models and their key assumptions. Chapter 4 introduces the
background of Yalongjiang River basin development. Chapter 5 values Project A using
NPV, NPV with simulation, IRR, IRR with simulation, decision tree analysis, ROA (real
options analysis) with electricity pricing as underlying, and ROA with NPV as underlying.
After comparing the methods, it concludes that the ROA with electricity pricing as
underlying is most appropriate for Project A valuation. Chapter 6 explores some policy
implications of the study. The organizational wisdom to make real impact of real options
and the relationship between uncertainty reality and appropriate methodology are the
implications. Chapter 7 concludes this thesis and discusses future work.
 14

Chapter 2 Literature Review

This study integrates two threads of research. One is real options research, and the
other is water resource planning with regard to facilities design. This chapter reviews the
literature of these two threads.

2.1. Real options

Myers [1987] was one of the first to acknowledge that there are inherent limitations with
standard discounted cash flow (DCF) approaches when it comes to valuing investment
with significant operating or strategic options. He suggested that options pricing holds
the best promise for valuing such investments.

2.1.1. Development of the methodology

Dixit and Pindyck [1994] stressed the important characteristic of irreversibility of most
investment decisions, and the ongoing uncertainty of the environment in which these
decisions are made. In so doing, they recognized the option value of waiting for better
(but never complete) information. The focus of their book was on understanding
investment behavior of firms, and developing the implications of this theory for industry
dynamics and government policy.

Trigeorgis [1996] brought together previously scattered knowledge about real options.
Comprehensively, he reviewed techniques of capital budgeting and detailed an approach
based on the pricing of options that provides a means of quantifying flexibility. He also
 15

dealt with options interaction, the valuation of multiple options that are common in
projects involving real options, and the valuation of the impact of competitive interactions.
The methodology in this book was theoretical, and helped to shape more practical real
options valuation techniques later on.

Besides theoretical development, applications of real options are growing fast in business
strategy, corporate finance, market valuation, contract valuation, security analysis,
portfolio management, risk management, to engineering design. Real options
methodology is applied in industries from natural resources development, real estate,
R&D, information technologies, pharmaceutical, manufacturing, venture capital,
government regulation, shipping, environmental pollution and global warming, to
infrastructure.

Amram and Kulatilaka [1999] wrote an introductory book on real options, including
financial options and applications of real options. But it does not provide a detailed
practical methodology to evaluate real options. It gives readers an idea how widely real
options can be applied.

The beginning of twenty-first century sees a boom in the publication of books on real
options with more focus on applications. Rogers [2002] describes his framework and
insights in applying real options gained as a consultant with PricewaterhouseCoopers.
Mun [2003] provides a qualitative and quantitative description of real options and multiple
business cases and real-life scenarios. Brach [2003] explores how to apply real option
valuation techniques on a regular basis from the view of a corporate practitioner. Howell
et al. [2001] and Boer [2002] aim more at illuminating non-numerate readers of real
options.

Although options thinking has been successfully applied to many areas, the application of
real options in engineering has been slow, especially regarding building flexibility into the
physical systems themselves. de Neufville [2002] suggested distinguishing between real
options “on projects” and “in projects”. The real options “on projects” concern the project
 16

but not system design, for example, the options imbedded in bidding for opening a mine.
The real options “in projects” concern the design elements of system and require detailed
understanding of the system, for example, options for repositioning of system of
communication satellites.

Over around 20-year development of the methodology, thousands of articles have been
published on this topic. Alexander Triantis maintains a portal with over 700 articles on
real options at http://www.rhsmith.umd.edu/finance/atriantis/RealOptionsportal.html.
Another good information center of real options is made by Marco Dias at http://www.puc-
rio.br/marco.ind/main.html#contents.

2.1.2. Application of the methodology

This thesis values options imbedded in river basin development. The literature review,
thus, focuses more on the application of the valuation methodology, rather than theory.

Valuation approaches

Options valuation is arbitrage-enforced pricing. Sometimes, however, people use

decision tree analysis (DTA) to illustrate the idea of options and approximate the value of
flexibility for real projects.

Arbitrage-enforced Real Options Valuation

As Baxter and Rennie [1996] explain, if there is arbitrage, it will enforce a price for the
options. This price depends neither on the expected value nor on the particular
distribution of the underlying. There are three categories of arbitrage-enforced valuation
 17

methods: the partial differential equation (PDE) such as the Black-Scholes formula, the
dynamic programming such as the binomial tree, and simulation.

The PDE method is the standard and widely used in academic discussion because of the
mathematical insights of the method. McDonald and Siegel [1986] studied the value of
waiting to invest by PDEs. Siegel, Smith and Paddock [1987] valued offshore petroleum
leases using PDE equations. Pindyck [1993] established PDEs to model project cost
uncertainty, both technical and as regards input cost. Grenadier and Weiss [1997]
studied the options pricing for investment in technological innovations.

The binomial tree method is based on a simple representation of the evolution of the
underlying asset’s value. It is a powerful yet flexible method to value real options.
Luenberger [1998] showed examples using binomial trees to value a real investment
opportunity in a gold mine. Cox, Ross, and Rubinstein [1979] developed this widely
adopted method. Copeland and Antikarov [2001] elaborated how to use binomial trees to
value real projects and proved this method, equivalent to PDE solution, is easy to use
without losing the insights of the PDE model.

With development of computer technology, big simulation programs can be constructed

to value options that are very difficult to value by establishing/solving equations or
building up binomial trees. In the 1980s, Merck began to use simulation to value its R&D
real options [Nichols, 1994]. Tufano and Moel [2000] showed an elegant way to use
Crystal Ball to simulate the value of real options inherent in a bidding case for mining.
Juan et al. [2002] suggested a simulation methodology to calculate multiple interacting
American options on a harbor investment problem.

Expected value decision tree analysis

Strictly, decision tree analysis (DTA) is an expected value approach that does not yield a
theoretically correct options value, unless the risk-adjusted discount rates and actual
probabilities are used. To find risk-adjusted discount rate for each branch of the decision
 18

tree is difficult, if possible. However, people use DTA to illustrate the idea of real options
and approximate the value of flexibility.

Faulkner [1996] showed how DTA could do “options thinking” valuation. Though this
method does not provide a “correct” options value, it approximates the value, and more
importantly, provides insights into options thinking.

Ramirez [2002] compared discounted cash flow methodology, decision tree analysis, and
arbitrage-enforced real options. She examined their theoretical
advantages/disadvantages, the assumptions, and information required. She also
determined the consequences of the application of each approach on the nature of
infrastructure projects.. Although the real options methodology is theoretically superior in
the pricing of flexibility, its implementation requires information usually not available for
infrastructure assets. This makes the results of the analysis imprecise and complicates
the process of identifying an optimal strategy.

Hybrid Model

Hybrid real options valuation combines the best features of decision tree analysis and
real options analysis. Neely and de Neufville [2001] developed a hybrid real options
valuation model for risky product development projects. The traditional valuation
methods for risky product development are inadequate to recognize the value of flexibility
while the real options method meets with difficulties to obtain the data necessary for a
standard real options valuation. Their hybrid method analyzes project risks with real
options analysis and market risks with decision tree analysis.

Summary

Table 2-1 illustrates five major types of valuation approaches with some examples:
 19

Table 2-1: Different Valuation Approaches with Examples

Arbitrage-enforced Real
Options Valuation
Examples DTA Hybrid Model
PDE Binomial Simulation
Tree
Automobile R&D Management
X
(Neely and de Neufville, 2001)
Bogota Water Supply Expansion
X X X
(Ramirez, 2002)
Merck (Nichols, 1994) X X
Kodak (Faulker, 1996) X

Borison [2003] described, contrasted, and criticized the major proposed analytic
approaches for applying real options. He observed relative strengths and weaknesses of
the approaches and recommended on which ones to use in what circumstance. He
thought the integrated approach (the hybrid model) is based on the most accurate and
consistent theoretical and empirical foundation.

Several important issues regarding real options valuation

Underlying

Financial options are based on underlying assets such as stocks, stock indices, foreign
currencies, debt instruments, commodities, and futures contracts. They are traded in
markets. Despite the fact that real options are not traded on markets, Mason and Merton
[1985], and Kasanen and Trigeorgis [1993] maintained that real options may be valued
similarly to financial options. The existence of a traded portfolio that has the same risk
 20

characteristics (i.e., is perfectly correlated) as a non-traded real asset is sufficient for real
options valuation. Kulatilaka [1993] used the relative price of oil over gas to value the
flexibility of a dual-fuel industrial steam boiler. Luenberger [1998] showed an example
using the gold price as the underlying assets to value a real investment opportunity in a
gold mine. Similarly, as shown in this thesis, it is possible to use energy price as the
underlying asset to value a hydropower project under the assumption of a complete
energy market.

However, in many cases, it is hard to find a priced portfolio whose cash payouts are
perfectly correlated with those of the project, or in other words, to find market-priced
underlying assets.

Is it possible to relax the definition of underlying to an agent that determines the value of
a project, not necessarily market-traded? Copeland and Antikarov [2001] developed the
assumption of “market asset disclaimer” and used the NPV of the underlying project as
underlying assets to build event trees to value real options. “Instead of searching in
financial market,” they recommended, “that you use the present value of the project itself,
without flexibility, as the underlying risky asset—the twin security.” (pp. 94) This method
has the key disadvantages that it makes it impossible to identify the optimal strategy and
blurs the exercise condition, because the NPV of a project is not readily market
observable. How to find an appropriate underlying is an interesting question.

Volatility

Volatility is a measure of uncertainty, a key input of the options valuation. How to find the
volatility is one of the key difficulties of application of real options if there is no market-
traded underlying. Luehrman [1998] described three approaches: an educated guess,
historical data, and simulation. Copeland and Antikarov [2001] suggested, by estimating
first the stochastic properties of variables that drive volatility, using Monte Carlo
simulation to estimate it.
 21

The estimate of volatility is often one of the weakest points of a real options valuation,
while the valuation is usually sensitive to the volatility. Because the volatility is the
essence of a lot of information, it is theoretically impossible to estimate it for some real
options valuation simply due to lack of data. Sometimes, therefore, the insights provided
by a real options analysis are more important than a specific quantitative result.

Compound options and parallel options

Most real options are not well-defined simple options. They can be compound or parallel.
They are often options on options (compound options) and the interactions between
options are significant. So the methodology for valuing compound options is very
important for the applicability of real options methodology in the real world. Parallel
options are different options built on the same project, where those options interact. For
example, several possible applications of a new technology or several possible target
markets of a new product. Oueslati [1999] described three parallel options for fuel cell
development as automotive applications, stationary power, and portable power.

Geske [1979] developed approaches to the valuation of compound options. Trigeorgis

[1993] focused on the nature of the interactions of real options. The combined value of a
collection of options usually differs from the sum of their separate values. The
incremental value of an additional option, in the presence of other options, is generally
less than its value in isolation, and declines as more options are present. Oueslati [1999]
explored the evaluation of compound and parallel real options in Ford’s investment in fuel
cell technology.

With all the developments in the application of the real options, the author is confident in
applying the methodology on the river basin problem in this thesis. However, a lot of
problems still await solutions. Without a market-observable underlying, the parameters
used for the valuation are based on model that must be subjective, to a certain extent.
So the model risks are not negligible in the method presented in this thesis.
 22

Real Options applied in Energy and Natural Resources

The real options concept has been successfully applied in the energy industry. Siegel,
Smith, and Paddock [1987] valued offshore petroleum leases using options, and provided
empirical evidence that options values are better than actual DCF-based bids. Since
then, research on real options on energy has been a hot topic. Dias [2002] gave a
comprehensive overview of real options in petroleum. Miltersen [1997] presented
methods to value natural resource investment with stochastic convenience yield and
interest rates. Cortazar and Casassus [1997] suggested a compound option model for
evaluating multistage natural resource investment. Cherian, Patel, and Khripko [2000]
studied the optimal extraction of nonrenewable resources when costs accumulate.
Goldberg and Read [2000] found that a simple modification to the Black-Scholes model
provides better estimates of prices for electricity options. Their modification combines the
lognormal distribution with a spike distribution to describe the electricity dynamics.

Pindyck [1993] studied the uncertain cost of investment in nuclear power plants. He
derived a decision rule for irreversible investments subject to technical uncertainty and
input uncertainty. The rule is to invest if the expected cost of completing the project is
below a critical number. The critical expected cost to completion depends on the type
and level of uncertainty. Pindyck’s work focused on finance issues of the project, the
engineering model was apart from his interest.

Koekebakker and Sodal [2002] developed an equilibrium-based real options model of an

operating electricity production unit whose supply is given by a stochastic mean-reverting
process. Hlouskova et al. [2002] implemented a real options model for the unit
commitment problem of a single turbine in a liberalized market. Price uncertainty was
captured by a mean-reverting process with jumps and time-varying means to account for
seasonality. Rocha, Moreira, and David [2002] studied the competitiveness of
thermopower generation in Brazil under current regulations and used real options to
assess how to motivate private investment in thermopower.
 23

2.2. Water Resource Planning with regard to

Facilities Design

Water resource planning with regard to facilities design is an area that came into maturity
by 1980, though the prevailing methodology does not consider the design issues in the
full context of the changing and uncertain world.

2.2.1. Historical development

After economic analysis was brought into water resource planning studies, research on
water resource planning with regard to facilities design can be divided into three phases:
mathematical programming, multiobjective analysis, and risk recognition.

Mathematical programming

Maass et al. [1962] summarized the contributions of Harvard Water Program (1955 –
1960). They introduced the most advanced techniques at that time: such as linear
programming, mathematical synthesis of streamflow sequences, and computer
simulation. This study laid foundation for future development of water resource planning.

Hufschmidt and Fiering [1966] described a river basin computer simulation model
thoroughly. Before computers became generally available, the simulation models were
physical models scaled to maintain static or dynamic similitude. The simulation model of
Hufschmidt and Fiering dealt with a large number of randomly selected designs, included
generation of long hydrologic sequences, and measured outcomes in economic terms.
 24

Their model was much more advanced than the then prevailing physical simulations.
They tested 20 randomly selected designs for 250 years (3000 months) of simulated
operations. The 3 designs with the highest net benefits were subjected to further
analysis using single-factor and marginal analysis methods. The model was written in
FORTRAN. The computer used was the IBM 7094 that had 32,768 directly addressable
memory (whereas modern computers have memory measured in Gigabytes!) It took
about 7.5 minutes for one single operation of simulation for 250 years.

Jacoby and Loucks [1974] developed an approach to the analysis of complex water
resource systems using both optimization and simulation, not as competing techniques
but as interacting and supplementing partners for each other. They used preliminary
screening models to select several alternative design configurations; then they simulated
the preferred designs using the same annual benefit, loss, and cost functions. They
estimated the expected benefits for each state for each 5-year period from 1970 to 2010.

Major and Lenton [1979] extended this work in a study of the Rio Colorado river basin
development in Argentina. Their system of models consisted of a screening model, a
simulation model and a sequencing model. The screening model is a mixed-integer
programming model with about 900 decision variables, including 8 0-1 integer variables,
and about 600 constraints. Objectives were incorporated either into the objective
function or as constraints on the system. Multiobjective criteria underlay the whole
formation of the model. The simulation model evaluated the most promising
configurations from the screening model in terms of net benefits and hydrological
reliability. The runs from the simulation model improved the configurations from the
screening model. The simulation model was operated with 50 years of seasonal (4-
month) flows. The sequencing model scheduled a candidate configuration optimally in 4
future time periods, taking into account benefits over time, budget constraints, constraints
on the number of farmers available, and project interrelationships. The mixed-integer
programming sequencing model had about 60 continuous variables, 120 integer
variables, and 110 constraints depending on the exact configuration being modeled.
 25

Loucks, Stedinger, and Haith [1981] summarized the art of water resource planning till
then, such as evaluation of time streams of benefits and costs, plan formulation, objective
functions and constraint equations, Lagrange multipliers, Dynamic Programming, Linear
Programming, Simulation, probability and distribution of random events, stochastic
processes, and planning under uncertainty.

After the 1980s, developments in water resource planning theory were less concerned
with the design of water resource facilities, partly because the water resource systems
were almost all developed in the US and partly because the art had matured. Entering
the 1990s, powerful computers and computer programs such as GAMS improved the
performance of previous methodologies and enabled the solution of much bigger
problems.

Multiobjective analysis

There are many different objectives for water resource planning, such as national or
regional income maximization, income redistribution, environmental quality, social well-
being, national security, self-sufficiency, regional growth and stability, and preservation of
natural areas. Some objectives can be easily expressed in monetary terms, while some
can not. Some (or all) objectives are conflicting and non-commensurable. Multiobjective
analysis does not yield a single optimal solution, but identifies the production possibility
frontier and trade-offs among objectives.

Cohon and Marks [1974] discussed an application of multiobjective theory to the analysis
of development of river basin systems. Major [1974] provided a case of the application of
Multiobjective analysis to the redesign of the Big Walnut Dam and reservoir in Indiana.
Major and Lenton [1979] demonstrated the application of mathematical modeling and
multiobjective investment criteria to river basin development.
 26

Risk recognition

The above-mentioned studies transformed technical parameters into expected total

annual net efficiency benefits (or the utility for human and society) and maximized the net
efficiency benefits (or utility) to obtain the “optimal design”. They carefully considered
technical uncertainties, such as that of waterflow, and used dynamic models. However,
none of them took into account uncertainties in the human and social sphere. Ignoring
the human and social uncertainties, the methodology cannot reach the “optimal design”
(if such designs exist) by simply recognizing the technical uncertainties.

Recent studies on water resource planning are more explicitly taking human, social,
environmental uncertainties into account. Morimoto and Hope [2001, 2002] applied
probabilistic cost-benefit analysis to hydroelectric projects in Sri Lanka, Malaysia, Nepal,
and Turkey. Their results were in the form of distributions of NPV. These studies
recognized the uncertainties from human, social, and environmental perspectives. But
they did not take into account the value of options, or the flexibility. Decision-makers do
not passively succumb to fate and they will respond to the uncertain environment.

2.2.2. Other important references

Based on Manne [1961], Hreisson [1990] dealt with the problem of obtaining the “optimal
design” of hydroelectric power systems regarding sizing and sequencing. The conclusion
was to make the current marginal value of each new project equal to the discounted
weighted average of the long-term marginal unit cost of all future projects. He further
investigated economies of scale and optimal selection of hydroelectric projects. The
tradeoff between large and small projects was studied by weighting the lost sales during
the period of excess capacity against the benefit of using larger projects due to the
economies of scale. All his studies were based on a deterministic view. If uncertainties
 27

regarding the demand and supply are high, the rules Hreisson developed may be
misleading.

Aberdein [1994] illustrated the case of excess electricity on the South Africa
interconnected grid resulted from the mismatch between planned capacity and actual
demand. She stressed the importance of incorporating risks into power station
investment decisions.

Some papers available on the website of World Commission on Dams

(http://www.dams.org) are very helpful. For example, Fuggle and Smith [2000] prepared a
report on dams in water and energy resource development in China. It provides
important background information on China’s dam building program, financing, and policy
development. Clarke [2000] reported the findings of a global dam survey covering 52
countries and 125 large dams. This report provides important information for the
uncertain factors to build the real options model in this thesis, such as the project
schedule performance data, actual-to-planned hydropower energy out, and many others.

2.3. Conclusions from the literature search

Although there is increased interest in real options, research has not expanded its
influence into the physical engineering design, where uncertainty and flexibility are key in
many circumstances. On the other hand, water resource planning studies focus on the
technical sides, including various technical uncertainties, but they have not incorporated
the important risks from human and social sides. Anyway, any engineering system is
built to serve human and human society. Real options methodology offers the advanced
accounts of some of the very important risks on human and social sides.
 28

It would be exciting to look into building real options into the engineering design of water
facilities themselves, and develop some general methodology to build flexibility into
engineering systems. Throughout the literature research, no prior systematic research
on this topic is found. This master’s thesis is a first step of this research. A first simple
deferral option on a river basin development will be studied.
 29

Chapter 3 Options

According to Hull [1999], Stock options “were first traded on an organized exchange in
1973”. The land-mark Black-Scholes model that won Nobel Prize in 1997. The model
was initially developed for financial options in 1973 by Black and Scholes and later
modified by Merton. Gradually, options methodology and thinking has been extended to
broader areas in finance and non-finance. Its insights into uncertainty and flexibility
enhance the ability of human beings to deal with forever-changing environments.

3.1. Financial Options

There are two basic types of options: calls and puts. A call option gives the holder the
right to buy an underlying asset for a specified exercise price within or at a specified time.
A put option gives the holder the right to sell the underlying under similar circumstances.
Expiration date is also called maturity. Exercise price is also called strike price.

Financial options are also categorized by the time when they can be exercised.
American options can be exercised any time up to the expiration date. European options
can exercised only on the expiration date.

Compound options are options on options. There are four main types of compound
options: a call on a call, a put on a call, a call on a put, and a put on a put. Many real
options have the compound options features.

The underlying assets for financial options include stocks, stock indices, foreign
currencies, debt instruments, commodities, and futures contracts. Besides the real
 30

options, this thesis is only discussing the financial options built on underlying assets of
stocks, or stock options.

Example of a stock call option:

John buys one European stock call option contract on Lucent stock with a strike price of
$1.50. Suppose the current price of Lucent is $1.30, the expiration date is in three
months. Because the option is European, John can exercise the option only on the
expiration date. If the stock price on the expiration date is less than $1.50, John will
choose not to exercise. If the stock price on the expiration date is greater than $1.50,
John will choose to exercise. For instance, if the stock price on the expiration date is
$1.45, John will not exercise the option, he can buy a share of stock directly on the
market for $1.45 less than the exercise of $1.50. If the stock price on the expiration date
is $1.60, John will exercise the option and earn $0.10 because he can immediately sell
the stock that he buys for $1.50.

Key Property of an Option

The holder of an option has the right to exercise the option, but no obligation to exercise
the option. The key property of an option is the asymmetry of the payoff, an option holder
can avoid downside risks and limit the loss to the price of getting the option, while she
can take advantage of the upside risks and the possible gain is unlimited. See Figure 3-1
for the example of a stock option.
 31

Payoff

Stock Price
Striking Price (Exercise Price)

Figure 3-1 Payoff Diagram for a Stock Call Option

For the above stock call option in Figure 3-1, if the current stock price is lower than the
strike price, people would not exercise it, the loss is limited to the price to get the option;
if the current stock price is higher than the striking price, people would exercise it, and the
payoff is the current stock price minus the strike price. There is no upper bound of the
payoff and a lower bound of the payoff, so asymmetry exhibits.

3.2. Cornerstones for Options Valuation

The value of an option is not straightforward, and it is an interesting question that how to
value an option objectively. The cornerstones for the modern stock options valuation
models are two assumptions: no arbitrage and Brownian motion of stock price.
 32

3.2.1. No Arbitrage

Arbitrage involves getting profit by simultaneously entering into transactions in 2 or more

markets. See the following example: Considering a stock that is traded on both the New
York Stock Exchange and the London Stock Exchange. If the stock price is $17.7 in New
York and ₤10 in London when the exchange rate is $1.8000 per pound. An arbitrager
could simultaneously buy 1000 shares of the stock in New York and sell them in London
to obtain a risk-free profit of

1000 × [($1.8 × 10) − $17.7]

or $300. Arbitrage opportunities such as the one just described cannot last for long. As
arbitrageurs buy the stock in New York, the forces of supply and demand will cause the
stock price to rise. Similarly, as arbitrageurs sell the stock in London, the forces of supply
and demand will cause the stock price to drop. Very quickly, the two prices will be
equivalent at the current exchange rate. Indeed, the existence of profit hungry
arbitrageurs makes it unlikely that a major price disparity could ever exist in the first place.

If no arbitrage opportunity exists, a portfolio of the stock and the stock option can be set
up in such a way that there is no uncertainty about the value of the portfolio. Because
the portfolio has no risk, the return earned on it must equal the risk-free interest rate.

A riskless portfolio consisting of a position in the option and a position in the underlying
stock is created. In the absence of arbitrage opportunities, the return from the portfolio
must be risk-free interest rate. The reason why a riskless portfolio can be created is the
stock price and the option price are both affected by the same courses of uncertainty:
stock price changes. In a short period of time, when an appropriate portfolio is
established, the gain or loss from the stock option always offset by the loss or gain from
the option position so that the value of the portfolio is known with certainty at the end of
the short period of time to earn risk-free rate of interest. For that short period of time, the
 33

price of a call option is perfectly positively correlated with the price of the underlying stock,
and the price of a put option is perfectly negatively correlated with the underlying stock.

For a simple example: A stock price is currently $10, and it is known that at the end of
period of y months the stock price will be either $11 or $9. There is a European call
option to buy the stock for $10.5 at the end of y months. This option will have one of the
two values at the end of the six months. If the stock price turns out to be $11, the value
of the option will be $0.5; if the stock option turns out to be $9, the value of the option will
be 0. The situation is illustrated in Figure 3-2:

Stock price = $11

Option value = $0.5
Stock Price = $10

Stock price = $9
Option value = $0

Figure 3-2 Stock Price Movement in Numerical Example

Consider a portfolio consisting of a long position1 in x shares of the stock and a short
position2 in 1 call option. How to calculate the value of x that makes the portfolio riskless.
If the stock price moves up from $10 to $11, the value of the share is 11x and the value
of the option is 0.5, so that the total value of the portfolio is 11x – 0.5. If the stock price
moves down from $10 to $9, the value of the shares is 9x and the value of the option is 0,
so that the total value of the portfolio is 9x. The portfolio is riskless if the value of x is
chosen so that the final value of the portfolio is the same for both cases. This means:

1
A long position is to buy the underlying asset on a certain specified future date for a certain
specified price.
2
A short position is to sell the underlying asset on a certain specified future date for a certain
specified price.
 34

11x − 0.5 = 9 x
or
x = 0.25

If the stock price moves up to $11, the value of the portfolio is

11 × 0.25 − 0.5 = 2.25

If the stock price moves down to $8, the value of the portfolio is

9 × 0.25 = 2.25

Regardless of whether the stock price moves up or down, the value of the portfolio is
always $2.25 after y months.

Riskless portfolio must, in the absence of arbitrage opportunities, earn the risk-free rate
of interest of r. It follows that the value of the portfolio today must be present value of
2.25, or

ry
−
2.25e 12

The value of the stock price is known to be $10. Suppose the option price is f. The value
of the portfolio is

rf
−
10 × 0.25 − f = 2.25e 12

or
rf
−
f = 2.5 − 2.25e 12
 35

3.2.2. Brownian motion and Weiner Processes

The standard model for stock prices is a geometric Brownian motion with constant
volatility. Standard Brownian motion is one of the most important basic notions of
stochastic processes, and in particular, is the basis of modern options theory.

To develop a sound theory of option pricing, one should describe the stock price
evolution using a dynamic model with a reasonable agreement with reality. The exact
formulation of the model for stock price evolution was a subject of debates for over a
century.

Brownian motion originally refers to the random motion observed under microscope of a
pollen immersed in water. Albert Einstein pointed out that this motion is caused by
random bombardment of heat-excited water molecules on the pollen. More precisely,
each of his steps (in both x- and y- directions) is an independent normal random variable.

Albert Einstein developed the notions of Brownian motion in the beginning of the 20th
century. In 1905 he defended his Ph.D. thesis on the subject of the separation of two
large particles experiencing random hits from surrounding small molecules. For this work
he received the Nobel Prize (Ironically, he did not receive the Nobel Price for the Theory
of Relativity). Although he himself considered his work not particularly important, this
work laid the ground for the theoretical understandings and beginnings of stochastic
processes altogether. Further contributors to the subject were Markov, Uhlenbeck,
Khintchine, Weiner, Smoluchowski, Ito, and Stratonovich. It was only in the 1960s that
the theory of Brownian motion was applied to modeling stock prices.

Stock prices are influenced by astronomical independent random factors together. Each
factor is trivial in the total influence. This kind of random variables, stock prices in this
case, usually follows normal distribution approximately. The reason why a normal
distribution is not used to describe stock prices is because stock prices cannot be
 36

negative. Lognormal distribution describes the change rate of stock prices (expressed
using continuously compounding) to be normal. The change of an stock price can be
negative, which means the effective market is decreasing while it is still positive.

Consider the following discrete construction. Let Zt0 be the position of a particle at time t0.
Let at time t0 + t the position of the particle be Zt0 + ∆Z, where the increments are
related:

∆Z = ε ∆t

where ε denotes a random sample from a standard normal distribution (mean 0 and
standard deviation 1).

Zt = Zt + ε n ⋅ ∆t
n n −1

Compounding n such increments, one can get for a finitely large interval of time T=n∆ t:

n
Z (t 0 + T ) = Z t 0 + ∑ ε i ∆t
i =1

Here ε i are all independent samplings from a standard normal distribution. Considering

the limit of ∆t → 0 + , it may be shown that the resulting process converges to a limit,
which is called standard Brownian motion, and is also referred to as a Wiener process.
 37

Brownian Motion

2
Position

0
0 50 100 150
-2

-4

-6
Time

Figure 3-3 One Path of Brownian Motion

Figure 3-3 exhibits a single path of a standard Brownian motion with initial condition X0 =
0.

Figure 3-4 Two Hundred Paths of Brownian Motion

Figure 3-4, in turn, shows two hundred paths of the standard Brownian motion.

The basic Wiener process, dz, has a drift rate (i.e. average change per unit of time) of
zero and a variance of 1.0. The drift rate of zero means that the expected value of z at
 38

any future time is equal to its current value. The variance rate of 1.0 means that the
variance of the change in z in a time interval of length T equals T.

A generalized Wiener process for a variable x can be defined in terms of dz as follows:

dx = adt + bdz Equation 3-1

where a and b are constants, dz is the basic Wiener process. The adt term implies that x
has an expected drift rate of a per unit of time. If without the bdz term, in a period of time
of length T, x increases by an amount of aT. The bdz term can be regarded as adding
noise or variability to the path followed by x. The amount of this noise or variability is b
times a basic Wiener process. A basic Wiener process has a standard deviation of 1.0.
It follows that b times a Wiener process has a standard deviation of b.

In a small time interval ∆t , the change in the value of x, ∆x , is

∆x = a∆t + bε ∆t

Since ε is a random number drawing from a standard normal distribution. Thus, ∆x has
a normal distribution with

Mean of ∆x = a∆t

Standard deviation of ∆x = b ∆t

Variance of ∆x = b 2 ∆t
 39

Stock Price Process Model

It is usually assumed that asset prices follow Geometric Brownian Motion where the
logarithm of the underlying variable follows generalized Wiener process.

If the price of a non dividend paying stock, S, follows geometric Brownian motion:

dS = µSdt + σSdz Equation 3-2

where S is the stock price, µ is the expected return on the stock, and σ is the volatility of
the stock price. The volatility of a stock price can be defined as the standard deviation of
the return provided by the stock in one year when the return is expressed using
continuous compounding. The volatility is also the standard deviation of the natural
logarithm of the stock price at the end of one year.

Then, using Ito’s lemma (see Appendix I) to get:

σ2
d ln S = ( µ − )dt + σdz Equation 3-3
2

From this equation, the variable lnS follows a generalized Wiener process, the change in
lnS between 0 and t is normally distributed, so that S has a lognormal distribution.

Lognormal Properties of stock price

In general, a lognormal distribution probability density function is as follows:

 40

− (ln x − µ ) 2

1 1
f ( x) = ⋅ ⋅e 2σ 2 t
Equation 3-4
x σ 2πt

but the following will show a more intuitive explanation of the lognormal distribution of the
stock prices. See Figure 3-5 for the shape of lognormal density function.

The lognormal distribution of price means the logarithm of the price has a normal
distribution. To illustrate, if a stock is priced at $100 per share and prices have a normal
distribution, the distribution of prices is the familiar bell-shaped curve centered at $100,
but if the prices have a lognormal distribution, then it is the logarithm of the price which
has a bell-shaped distribution about ln(100) = 4.605. The logarithm of the prices is
equally likely to be 5.298 or 3.912, i.e., 4.605 ± 0.693 corresponding to prices of $200 and
$50, respectively. If the lognormal probability density curve is plotted as a function of
price rather than as a function of the logarithm of price, the curve will appear positively
skewed with tails more nearly depicting the observed behavior of stock prices.

Lognormal Density

0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0 1 2 3 4 5 6 7 8

Figure 3-5 Lognormal Density

 41

Lognormality arises from the process of return compounding, in other words, the
lognormal property of stock prices apply when the return rate earned on a stock between
time 0 and t is continuously compounded. It is important to distinguish the continuously
compounded rate of return and the annualized return with no compounding as

1 St − S0
( )
t S0

3.3. Options Valuation Tools

The first model to calculate options value is the Black-Scholes formula, which is
sometimes deemed arcane. Interests in option pricing, however, has picked up in recent
years as more powerful computers can aid very sophisticated model building. With
simulation methods available easily as Excel add-ins or more professional alternatives
such as @Risk or Crystal Ball, people are able to do a hundred thousand simulations
easily and get the payoff distribution as well as the value of real options 1 . Besides,
binomial model proves very successful in option pricing and decision analysis is
sometimes another approach to value options approximately2.

1
Simulation generates values of uncertain variables according to the probability distribution of the
variables, uses those values as inputs, and gets the output. With a great number of repetitions,
the probability distribution of the output is established.
2
The valuation by decision analysis is not real options valuation strictly because it does not use
the risk-neutral valuation (refer to 3.4. ).
 42

Five inputs are needed for a options valuation (if considering the simplest situation when
there is no dividend):

- strike price
- risk-free interest rate
- time to expiration
- current stock price
- uncertainty (volatility is the measurement)

Among the five inputs of an options model, the first four inputs are relatively easier to get,
while the last one uncertainty, which is estimated by volatility in most cases. with a lot of
historical data available on the stock market, it is relatively trivial to get σ for a stock
option. However, for a real option, lack of historical data is a common problem except a
few specific industries, such as pharmaceutical industry. Because of the lack of historical
data, it is very hard to justify the choice of volatility. This is one of the practical difficulties
facing real options valuation method.

3.3.1. The Black-Scholes Model:

The Black-Scholes-Merton analysis is based on no-arbitrage condition.

The stock price process is the one as we developed in the last section as Equation 3-2:

dS = µSdt + σSdz

Suppose f(S,t) is the price of a call option, which is some function of the stock price of S
and time of t. Hence from Ito’s Lemma (see Appendix I):
 43

∂f ∂f 1 ∂ 2 f 2 2 ∂f
df = ( µS + + σ S )dt + σSdz Equation 3-5
∂S ∂t 2 ∂S 2
∂S

The Wiener processes dz underlying f and S are the same. It follows that by choosing a
portfolio of the stock and the stock option, the Wiener process can be eliminated. The
appropriate portfolio is short one call option and long an amount ∂f ∂S of shares.

Define ∏ as the value of the portfolio. By definition

∂f
∏ = − f + ∂S S Equation 3-6

Note the portfolio is riskless only for an infinitesimally short period of time. As S and t
change, ∂f ∂S also changes. To keep the portfolio riskless, it is necessary to constantly
change the composition of the portfolio.

Because the discrete version of equations Equation 3-5 and Equation 3-2 are

∂f ∂f 1 ∂ 2 f 2 2 ∂f
∆f = ( µS + + σ S )∆t + σS∆z
∂S ∂t 2 ∂S 2
∂S

and

∆S = µS∆t + σS∆z

Then the change ∆ ∏ in the time interval ∆t is given by

 44

∂f
∆ ∏ = − ∆f + ∆S
∂S
∂f ∂f 1 ∂ 2 f 2 2 ∂f ∂f ∂f
= − ( µS + + σ S )∆t − σS∆z + µS∆t + σS∆z Equation 3-7
∂S ∂t 2 ∂S 2
∂S ∂S ∂S
∂f 1 ∂ f 2 2
2
= (− − σ S )∆t
∂S 2 ∂S 2

The equation does not involve ∆z , the portfolio must be riskless during time ∆t under the
assumption of no arbitrage. It follows that

∆ ∏ = r ∏ ∆t Equation 3-8

where r is the risk-free interest rate. Substituting from equations Equation 3-6 and
Equation 3-7, this becomes

∂f 1 ∂ 2 f 2 2 ∂f
( + σ S )∆t = r ( f − S )∆t
∂t 2 ∂S 2
∂S

So that

∂f ∂f 1 2 2 ∂ 2 f
+ rS + σ S = rf Equation 3-9
∂t ∂S 2 ∂S 2

Equation 3-9 is the Black-Scholes-Merton differential equation. Its

solution depends on the boundary conditions used. In case of a European call option,
the key boundary condition is

f = max[S − X ,0] when t = T

 45

Solving the differential equation subject to the boundary conditions to get the Black-
Scholes formulas for the prices at time zero of a European call option on a non-dividend-
paying stock:

c = S 0 N (d1 ) − Xe − rT N (d 2 ) Equation 3-10

where

ln( S 0 / X ) + (r + σ 2 / 2)T
d1 =
σ T
ln( S 0 / X ) + (r − σ 2 / 2)T
d2 = = d1 − σ T
σ T

and N(x) is the cumulative probability distribution function for a variable that that is
normally distributed with a mean of zero and a standard deviation of 1.0.

In case of a European put option, the key boundary condition is

f = max[ X − S ,0] when t = T

solving the differential equation to get the Black-Scholes formula for the prices at time
zero of a European put option on a non-dividend-paying stock

p = Xe − rT N (−d 2 ) − S 0 N (− d1 )

After going through all the derivation of the Black-Scholes formula, the two most
important assumptions under the options pricing are stressed again here:
 46

• no arbitrage
• geometric Brownian motion

It is key to understand what these two assumption’s role in real options valuation. And
thus understand the applicability for real options in different scenarios.

3.3.2. Valuation by Simulation

Simulation models roll out a huge number, typically thousands, of possible paths of the
evolution of the value of the underlying from today to the final day studied. With options
decision rule imbedded in each of the path, the average of the value on the final day is
discounted back to today to obtain the options value.

For example, the current price of a stock is $20, volatility is 30% per year, risk-free rate is
5% per year. A stock European call option is built on such a stock with time to expiration
of 3 months and strike price of $22.

Simulating the stock price 3 months later to get the distribution of the stock price as of
Figure 3-6. With the European option imbedded, the distribution of price less than $22 is
cut because people won’t exercise it in such case. See Figure 3-7. The average of the
chunk of the distribution greater than $22 minus $22 is the future value of the option. The
future value of the option is then discounted to get the option value.

One of the advantages of the simulation model is that it can handle path-dependent
options, in which the value of options depends not only on the value of the underlying, but
also on the particular path followed by the underlying.
 47

10,000 Trials Frequency Chart 90 Outliers

.024 237

.018 177.7

.012 118.5

.006 59.25

.000 0
16.19 18.17 20.15 22.12 24.10

Figure 3-6 Distribution of Stock Price for Valuation by Simulation

10,000 Trials Frequency Chart 90 Outliers

.024 237

.018 177.7

.012 118.5

.006 59.25

.000 0
16.19 18.17 20.15 22.12 24.10

Figure 3-7 Distribution of Exercise of Option for Valuation by Simulation

With the fast development of computer hardware and software technologies, the
simulation model has been more and more powerful and easy-to-use. A normal laptop
can run thousands of simulations in seconds, and new software program such as Crystal
Ball makes the simulation method accessible to everybody.

Find an example of options valuation by simulation in section 3.5.

 48

3.3.3. Binomial Real Options Valuation

Considering a stock whose price is initially S 0 and an option on the stock whose current

price is f . Suppose that the option lasts for time T . During the life of the option, the

stock price can either move up from S 0 to a new level, S 0 u , or down from S 0 to a new

level S 0 d . The proportional increase in the stock price when there is an up movement is

u − 1 ; the proportional decrease when there is a down movement is 1 − d . If the stock

price moves up to S 0 u , the payoff from the option is assumed to be f u ; if the stock price

moves down to S 0 d , the payoff from the option is assumed to be f d . Figure 3-8

illustrates the situation:

Su
S fu
f
Sd
fd

Figure 3-8 Stock and Option Price in a One-step Binomial Tree

Considering a portfolio consisting of a long position of x shares and a short position in

one option. If there is an up movement in the stock price, the value of the portfolio at the
end of the life of the option is

S 0 ux − f u

If there is a down movement in the stock price, the value becomes

S 0 dx − f d
 49

The two are equal when

S 0 ux − f u = S 0 dx − f d

or

fu − fd
x= Equation 3-11
S 0u − S 0 d

In this case, the portfolio is riskless. Due to the no arbitrage condition, the portfolio must
earn risk-free interest rate. x is the ratio of the change in the option price to the change in
the stock price.

The present value of the portfolio is

( S 0 ux − f u )e − rT

where r is the risk-free interest rate. The cost of setting the portfolio is

S0 x − f

It follows that

S 0 x − f = ( S 0 ux − f u )e − rT
or

f = S 0 x − ( S 0 ux − f u )e − rT
 50

Substituting x from Equation 3-11, this equation reduces to

f = e − rT [ pf u + (1 − p ) f d ] Equation 3-12

where

e rT − d
pu = Equation 3-13
u−d

One way to match volatility with u and d is

u = eσ ∆t
Equation 3-14

d = e −σ ∆t
Equation 3-15

One more step can be added to the binomial tree as Figure 3-9.

S uu

Su f uu

S fu S ud
f
Sd f ud

fd
S dd

f dd

Figure 3-9 A Two-Step Binomial Tree

 51

Repeated application of Equation 3-12 gives

f u = e − rT [ pf uu + (1 − p) f ud ]

f d = e − rT [ pf ud + (1 − p ) f dd ]

And finally get:

f = e − rT [ pf u + (1 − p ) f d ]

3.4. Risk-neutral Valuation

Refer to Figure 3-8, the expect stock price at time T, E(ST), is given by

E ( S T ) = pu S 0 u + (1 − pu ) S 0 d
or

E ( S T ) = pu S 0 (u − d ) + S 0 d

Substituting from Equation 3-13, this reduces to

E ( S T ) = S 0 e rT Equation 3-16

showing the expected growth of the stock price is the risk-free rate. Setting the
probability of the up movement equal to pu is to assume that the expected return on the
stock is the risk-free rate.
 52

In a risk-neutral world, all individuals are expected value maximizers and require no
compensation for risk, and the expected return on all securities is the risk-free rate.
Equation 3-12 shows that the value of option is its expected payoff in a risk-neutral world
discounted at the risk-free rate. The risk-neutral valuation principle states that it is valid
to assume the world is risk neutral when pricing options. The result is correct for all
worlds, not only in the risk-neutral world.

3.5. Options Valuation and Decision Tree

Analysis

Options analysis is continuous, but decision tree is discrete. Normally, the decision tree
analysis will not give the correct value for options because it is not a risk neutral analysis.
Decision tree analysis does not refer to arbitrage-enforced price, and use the actual
probabilities of the price movement of the underlying assets. If using risk-neutral
evaluation and simulation, however, the decision tree analysis will give the exactly same
answer as Black-Scholes.

To a certain extent, what the options theory offers is an understanding of the stock prices
that is lognormally distributed. From today's stock price and the volatility of the stock price,
the distribution of a stock price on a future date can be derived. With the understanding
of the stock price distribution, decision tree analysis can be applied.

An Example of Option Valuation

If there is a stock, the stock price S0 is $20 now, the volatility σ is 30%, risk-free interest
rate r is 5%. A call option is built on this stock, the strike price X is $22, and the time to
maturity T is 1 year.
 53

Applying two methods to get the value of this call option.

Method 1: Black-Scholes Formula

The Black-Scholes Formula is as Equation 3-10

c = S 0 N (d1 ) − Xe − rT N (d 2 )

where

ln(S 0 / X ) + (r + σ 2 / 2)T
d1 =
σ T
ln(S 0 / X ) + (r − σ 2 / 2)T
d2 = = d1 − σ T
σ T

and N(x) is the cumulative probability distribution function for a variable that that is
normally distributed with a mean of zero and a standard deviation of 1.0.

Substitute the actual value of S0, σ , r , X, and T into the formula, and the value of this
call option is $1.994.

Method 2: Decision Tree Analysis using Monte Carlo Simulation

The basic structure of the tree is as Figure 3-10:

Figure 3-10 Decision Tree for Options Valuation

 54

The rectangle is a decision point at the expiration day, at which there are two possible
decisions: exercise the option, and the value of the option is the stock price then minus
the strike price of $22; do not exercise the option, the value of the option is 0 in this case.

Now the key of the above decision tree analysis is the stock price. The stock price is a
stochastic process. It has a specific distribution on the expiration day that can be derived
from the current stock price, volatility, risk-free rate, and time to expiration. The
assumptions needed for that deduction is the cornerstone for the modern finance theory,
i.e. the geometric Brownian motion of the stock price and the no-arbitrage assumption.
These two assumptions also lead to the Black-Scholes formula.

The two assumptions lead to that the stock price on the expiration day must follow a
lognormal distribution with a expected value as Equation 3-16

E ( S T ) = S 0 e rT

where T is time to expiration and r is the risk-free rate.

Substitute the actual value of S0, r, and T into Equation 3-16 to get the distribution of the
stock price on the expiration day after a year:

E ( S T ) = $21.025

Two parameters, i.e. µ and σ, are needed to specify the lognormal distribution of the
stock price on the expiration day. Please refer to Equation 3-4. In this example, σ is 0.3.
µ is the expected value of the annual return expressed using continuous compounding,
and

µ ≠ ln[ E ( S )] ,
 55

but

µ = E[ln(S / S 0 )] = ln{E ( S )] − σ 2 / 2 Equation 3-17

or

µ = ln(21) − (0.3) 2 / 2 = 3.001 .

With the value of µ and σ, the distribution of the stock price on the expiration day is
specified. The last thing important is that the option is to expire a year later, but the value
of the option as of today needs to be calculated. So the expected value obtained by the
decision tree needs to be discounted.

Finally, Monte Carlo simulation is applied to get the value of the option c. The software
used is Crystal Ball. The relative precision of the simulation of the mean is set to be 1%.
It means that the software will stop simulation after the mean of the simulated results is
within ±1% range of true expected mean.

The software simulates 249,500 times before it stops and reaches the relative precision
of 1%. Please see the output from Crystal Ball as Table 3-1:

The result of Table 3-1 shows the true expected value or the value of the call option
should be in the range $2.00082 ± 1% x $2.00082, or ($1.981, $2.021). The result from
the Black-Scholes formula is $1.994, exactly in the range. This test shows that the
expected value of the tree (after discounted) is the call option value and it is the same as
the result from Black-Scholes formula.
 56

Table 3-1 Option Valuation by Decision Tree Results

Statistic Value Precision

Trials 249,500
Mean 2.00082 1.00%
Median 0.00000
Mode 0.00000
Standard Deviation 3.87972 0.88%
Variance 15.05222
Skewness 2.72981
Kurtosis 12.53695
Coeff. of Variability 1.93906
Range Minimum 0.00000
Range Maximum 50.71999
Range Width 50.71999
Mean Std. Error 0.00777

The precision of the simulation can be set better, even though the mean of the simulated
results fluctuating around the expected value. The Crystal Ball software student version
used can only have two decimals to specify the distribution, which means 3.00 have to be
used as the µ instead of 3.001. This error is unnecessary and can be eliminated by
better programming.

This example shows an interesting result that options can be valued by decision tree
analysis with simulation. In some sense, modern finance theory helps people to get the
distribution of the stock price at any future day with the observable parameters of the
current stock price, the risk-free rate, and the volatility of the stock price.
 57

3.6. Real Options

“The classic way to value businesses is to compute the discounted present value of their
future cash payouts. Not good enough, says Michael Mauboussin, the chief U.S.
investment strategist at Credit Suisse First Boston. You should also throw in something
for the company's ‘real options’”. [Schoenberger, 2000]

An article published on McKinsey Quarterly argued, “Real Options are especially valuable
for projects that involve both a high level of uncertainty and opportunities to dispel it as
new information becomes available”. [Leslie and Michaels, 1997]

MIT professor Stewart Myers [1984] first coined the term of “real options”:

“Strategic planning needs finance. Present value calculations are needed as a check on
strategic analysis and vice versa. However, standard discounted cash flow techniques
will tend to understate the option value attached to growing profitable lines of business.
Corporate finance theory requires extension to deal with real options.” (pp. 136)

3.6.1. What is Real Options Method?

An opportunity is like a call option because the company has the right, not the obligation,
to invest in a project. It is possible to find a call option sufficiently similar to the
opportunity. The value of the option would tell us something about the value of the
opportunity. Although most projects are unique and the likelihood to find a similar option
on the market is low, people can reliably find a similar option by constructing one.

Before the formalization of the options theory, people intuitively know the benefit of
options, such as the ancient Chinese proverbs “ a cunning rabbit has three caves” and
“never put all the eggs in one basket”. With the development of option theory, people can
 58

now estimate the value of opportunities more precisely, which enabling them to compare
the value of an option with its cost. A more sensible decision can be reached, and people
shouldn’t spend more for an option than it’s worth.

Example 1: Petroleum chemical company

Paraphrasing an example from Amram and Kulatilaka [1999], a petroleum Chemical

company might begin to invest in new capacity, but is worried about the size of the
market opportunity and whether the manufacturing process could meet the government
regulations regarding environment protection. Traditional Net Present Value (NPV)
analysis suggested that the project should not be pursued. A real options analysis,
however, valued the exit option held by the company – the option that the company could
walk away if there were bad news about the market or the government regulation.
Although there would be a loss of initial investment if the project were cancelled,
including the abandonment option, the project value increased and the company began
to construct new capacity.

Example 2: Oil exploring

Paraphrasing an example from Leslie and Michaels [1997], a North Sea oil company
accumulated a portfolio of license blocks – five-years rights to explore and produce oil
and gas. The development was unsuccessful and left it with unwanted blocks that were
consuming cash. The company decided to sell the blocks initially. During the divestment
program, it was suggested, however, instead of calculating what the block would be
worth if the company started developing them immediately, the company should value its
opportunity as an option to develop if, at sometime in the future, recoverable reserves
could be increased through new technologies. A simple financial model was developed
to show how to price the blocks at their option value over 5 years, incorporating
uncertainty about the size of the reserve, the oil prices, and room for flexible response to
the outcome. The managers reevaluated the company’s portfolio, and instead of letting
 59

blocks go, they held on to those with high option value and to sold the rest at the revised
values.

3.6.2. Comparison of Real Options Method and Traditional NPV

Method

Often, although the NPV proves to be negative, the management team decides to go
ahead anyway; or the NPV is positive, but intuition warns people not to proceed. It is not
the intuition is wrong, but the time-honored NPV decision-making tool. As a practical
matter, many managers seem to understand there is something wrong with the simple
NPV rules, i.e., there is a value to waiting for more information and the this value is not
reflected in the standard NPV calculation.

Traditional NPV valuation tools ignore an important value of a project - the value of
flexibility. The traditional NPV method assumes, if an investment is irreversible, the
investment is now-or-never, or in other words, if the company does not make the
investment now, it will lose the opportunity forever. The traditional NPV method does not
take into account an important reality: business decision in many industries and situations
can be implemented flexibly through deferral, abandonment, expansion, or in a series of
stages that in effect constitute real options.

Please see the following example based on Prof. De Neufville’s class notes [2002] for
MIT engineering school-wide elective, Engineering Systems Analysis for Design:

Suppose a project can be started for $100, and $1100 more will be required to complete.
We must decide whether or not to continue after observing the initial result. And the
commercial feasibility is decided by the initial result and the market condition then. Our
final objective is to license the technology to a bidder who offers the highest price. The
revenue estimate is shown in Table 3-2.
 60

Table 3-2 Revenue Estimate for Technology Development

Revenue Chance
License for $2000 50%
License for $100 50%

Assuming the discount rate is 10%, the question is: do we fund the project?

Table 3-3 shows traditional discounted cash flow (DCF) and net present value (NPV)
valuation:

Table 3-3 NPV Valuation of Technology Development

Year 0 Year 1 Year 3

Initial cost -$100
Development -$1100
License revenues 0.5x$2000 + 0.5x$100
Present Value -$100 -$1000 $868

The traditional NPV valuation is -$232, so the project should be rejected.

But if we employ Real Options thinking, we understand that we have the option to
develop only if $2000 license is expected. Now the analysis is as Table 3-4:

Table 3-4 Approximate Options Valuation of Technology Development

Year 0 Year 1 Year 2

Initial cost -$100
Development 0.5×$1100
License revenues 0.5×$2000 + 0.5×$0
Present Value -$100 -$500 $826
 61

And the new NPV is $226, so we should accept the project1. The thinking of option is
always natural and intuitive for managers even without the formal option valuation tools.
However, we were not able to valuate options strictly and rigidly before we have the
option valuation models. Now, with those option valuation tools, the option thinking can
be applied from the state of qualitative intuition to the state of quantitative rigidity.

In addition, there is another key difference between Real Options valuation and NPV.
NPV needs an appropriate discount rate to bring the future cash flows back into present
dollars, while real options models are attractive because they eliminate the need to
resolve this issue. Black Scholes Formula (Equation 3-10) shows that options pricing
does not require a discount rate. The question about an appropriate discount rate has a
lot of debates, and there is no consensus or natural way to get an appropriate discount
rate for a project. Real Options approach can circumvent the pain of discount rate choice.

3.6.3. Types of Real Options

Some options occur naturally (e.g., to defer, contract, shut down or abandon), while
others may be planned and built-in with extra cost (e.g. to expand growth options, to
default when investment is staged sequentially, or to switch between alternative inputs or
outputs). Table 2 describes briefly the most common categories of real options.

1
Note this $226 is not the options value. It is only an approximation of the options value because
it is not a risk-neutral valuation.
 62

Table 3-5 Types of Real Options

Category Description Important In

Option to defer Management holds a lease on (or an All natural resource extraction industries;
option to buy) valuable land or resources. real estate development; farming; paper
It can wait (x years) to see if output prices products
justify constructing a building or plant, or
developing a field.
Time to build option Staging investment as a series of outlays All R&D intensive industries, especially
(staged investment) creates the option to abandon the pharmaceuticals; long-development
enterprise in midstream if new information capital-intensive projects, e.g., large-scale
is unfavorable. Each stage can be viewed construction or energy-generating plants;
as an option on the value of subsequent start-up ventures
stages, and valued as a compound option.
Scaling Option (e.g., to If market conditions are more favorable Natural resource industries such as mine
expand; to contract; to shut than expected, the firm can expand the operations; facilities planning and
down or restart) scale of production or accelerate resource construction in cyclical industries; fashion
utilization. Conversely, if conditions are apparel; consumer goods; commercial real
less favorable than expected, it can reduce estate.
the scale of operations. In extreme cases,
production may halt or start up again.
Option to abandon If market conditions decline severely, Capital intensive industries, such as
management can abandon current airlines and railroads; financial services;
operations permanently and realize the new product introductions in uncertain
resale value of capital equipment and other markets.
assets in secondhand markets.
Option to switch (e.g., If price or demand change, management Output shifts:
outputs or inputs) can change the output mix of the facility Any good sought in small batches or
(“product flexibility”). Alternatively, the subject to volatile demand, e.g., consumer
same outputs can be produced using electronics; toys; specialty paper; machine
different types of inputs (“process parts; autos; Input shifts:
flexibility”) All feedstock-dependent facilities, e.g., oil;
electric power; chemicals; crop switching;
sourcing
 63

Growth option As early investment (e.g., R&D, lease on All infrastructure-based or strategic
undeveloped land or oil reserves, strategic industries, especially high-tech, R&D, or
acquisition, information industries with multiple product
network/infrastructure) is a prerequisite or generations or applications (e.g.
link in a chain or interrelated projects, computers, pharmaceuticals);
opening up future growth opportunities multinational operations; strategic
(e.g., new generation product or process, acquisitions.
oil reserves, access to new market,
strengthening of core capabilities). Like
interproject compound options.
Multiple interacting Real-life projects often involve a Real-life projects in most industries

options “collection” of various options, both discussed above.

upward-potential enhancing calls and
downward-protection put options present in
combination. Their combined option value
may differ from the sum of separate option
values, i.e., they interact. They may also
interact with financial flexibility options.
(Source: Lenos Trigeorgis, 1993. Real Options and Interactions with Financial Flexibility. Financial Management.
Autumn.)

3.6.4. Some Implications of Real Options Method

Valuation of projects incorporating flexibility

Applying real options thinking, people can actively manage risks and uncertainties, not
passively perceive the value of options vaguely as before. People can systematically
identify and establish options into a project, increasing the value of the project,
appreciating the value of the project wholly, and taking advantage of upside risks while
avoiding downside risks.

Appreciating that a project is like a financial call option can help people recognize the
crucial role that uncertainty plays in the investment decisions. With a financial call
option, the more volatile the price of the stock in which the option is established, the more
 64

valuable is the option and the greater incentive to keep the option open. This is true
because of the asymmetry in the option - the higher the price rises, the higher the payoff
is; however, if the stock price falls, one can lose only what for the price of the option.

The same goes for project investment decisions. The greater the uncertainty of a project,
the greater the value of the opportunity and the greater incentive to wait and keep the
opportunity alive rather that exercise it immediately. Of course, the traditional NPV
method also considers the uncertainty by way of the choice of discount rate. But in the
real options thinking, uncertainty is far more important and fundamental.

In addition to understanding the role of uncertainty, the real option thinking helps
companies to think systematically and actively to obtain options by their technological
knowledge, reputation, managerial resources, market position, and possible scale.
People need to understand options and get opportunities in hand first.

With some data, real options approach can add a quantitative rigor to the valuation of the
flexibility. Flexibility comes with cost. Using quantitative real options valuation, people
can maximize the net value of an option, i.e., the value of an option minus cost, given a
certain amount of investment budget. With binomial and simulation valuation, moreover,
people can get the possibility distribution of a project’s payoffs with/without options. In
this way, people can have a more holistic understanding of the project than only given a
single prediction of payoffs as in the traditional NPV method.

Investment with Options Thinking

In an uncertain world, strategic investment can be analyzed from a real option

perspective:
 65

Irreversible investments

Irreversible investment requires more careful analysis because, once the investment
takes place, the investment cannot be recouped without a significant loss of value. With
the real options analysis, it is understood that irreversible investments, for most of the
cases, should be delayed until a significant amount of the uncertainty is resolved, or the
investments should be broken into stages.

Flexibility investments

Flexibility investment builds options into the initial design. Flexible design allows a
production line to be easily switched across products. The option to switch is part of the
capital investment.

Insurance investments

Insurance investments reduce exposure to uncertainty. Investment in excess capacity

ensures against if demand surges, but with a cost or “insurance premium”. Decision-
makers using real options approach are able to value the flexibility and check to see
whether the value exceeds the cost.

Platform investments

Platform investments create valuable follow-on contingent investment opportunities.

Using Real Options approach, managers can create a portfolio of projects, maximizing
the value of the portfolio, balancing the portfolio with high-risk-high-return and low-risk-
low-return projects, and align the projects tightly with the corporate strategy.
 66

Growth investments

Growth investments are made to obtain information that is otherwise unavailable. For
example, oil exploration is a growth investment because it generates geological
information.

Value of Real options Method in Different Situations

Real options valuation is important in situations with high uncertainty and people have
many options when new information received. If the uncertainty is low and the available
practical options are few, the Real Options approach will not add much insight into
traditional NPV method. This is because the flexibility value is near zero. Please see
Figure2:

Value of flexibility Appropriate strategy Value of real options approach

Uncertainty Uncertainty Uncertainty
Low High Low High Low High

Moderate High
Case Flexible
High flexibility flexibility High High Medium High
Number Number specific strategy Number
value value
of of of
available available available
feasible feasible feasible
Low Moderate
Options Options Focused Case Options
Low flexibility flexibility Low Low Low Medium
strategy specific
value value

Figure 3-11: Applicability of Real Options Method

For the case of high uncertainty and a big room for options, the flexible strategy and the
real options approach are most valuable. And the Real Options approach will provide a
much better result than NPV method.
 67

3.6.5. Framework of Real Options Valuation

Before using real options to evaluate a project, people first need to understand clearly
what decision to be made and check if it is advantageous to use this approach over
traditional NPV method. If so, the valuation can be divided into 5 steps, as shown in
Figure 3-12:

Step 1: find out the most important

uncertainties
Step 3: analyze available real
Step 2: approximate the probability options
distribution of the uncertainties

Step 4: valuate real options

Step 5: develop real options

mind-set

Figure 3-12 Framework of Real Options Method

The first step, most important drivers and uncertainties of the project should be found out.
Usually uncertainties include market risk (such as the market demand, price of the
product, economic cycle), technical risk (such as if the project can be finished on time, if
the project can achieve its technical objectives).

The second step, an approximate probability distribution should be assigned to each

uncertainty. In many cases, a lognormal distribution is used for a market risk. If there
are other project-specific risks associated with the project, their probability distributions
should be studied case by case.
 68

The third step, the most important options should be identified. Possible options practical
to the project studied can be identified with reference to Table 3-5 of the types of real
options.

The fourth step, appropriate method among Black-Scholes formula, binomial model, and
simulation is identified and applied to obtain the value of the options.

The fifth step, by comparing the value of the options and cost to obtain options, a set of
strategies and decisions can be reached.

Meanwhile, the mind-set regarding flexibility available and different is established.

Valuators need to be careful of the false precision of the value of an option, because the
value is established on many approximations and assumptions. This is why a sensitivity
analysis is sometimes needed. Nevertheless, the mind-set to value the flexibility is one
of the major gains of this project.

3.6.6. Real Options versus Financial Options

To use the methodology originally developed for financial options, there should be an
appropriate underlying. For the most talked-about stock options, the underlying is the
stock price. In real options study, however, a generalized concept of underlying is
needed. An underlying is the agent that determines the value of a project or an
investment. An underlying can be assets, but it also can be other agents such as market
size or utility price. To use Black-Scholes or binomial model, the underlying should
follow the Geometric Brownian motion like stock prices.

If the underlying is not following a geometric Brownian motion, the options thinking can
still be applied. The reason is that the key for options thinking has not necessarily to
resemble financial options. The essence is the right not the obligation for a property or
 69

project. If appropriate distribution for the underlying can be found, people can use
mathematical deduction to get the valuation (like Black-Scholes formula to get option
value based on the assumption of geometric Brownian motion), or can just use Monte
Carlo simulation to get the option value.

There are a number of difficulties to apply real options to the fields other than financial
area where the options theory was originally developed. The key problem is that no
market exists for the object studied, so the most powerful characteristics of financial
options theory is partly gone. For financial options theory, the power originates from the
efficient market where the stock price contains all the information. While there is no
market, it is very hard to find out these information.

Often, the real options approach is hard to be applied because of no justifiable σ. One of
the ways to circumvent this problem is to use hybrid real options method [Neely and de
Neufville, 2001] that uses decision analysis for the part of analysis where historical data
is not sufficient, e.g., the R&D stage for a new product, and real option analysis, e.g., the
marketing of a new product.
 70

Chapter 4 Yalongjiang River Basin

Chinese central government and Sichuan local government are planning a number of big
projects on the Yalongjiang River in the near future, total investments near $10 billion.

4.1. Introduction of Sichuan

Yalongjiang River is located in Sichuan Province, the West of China. See Figure 4-1.

Figure 4-1: Sichuan in China

 71

Sichuan has an area of 485,000km2. The population is 83,000,000 (figure as of

November, 2000. Source: Sichuan Provincial Statistics Bureau.). Sichuan locates
in the upstream of Yangtze River. Sichuan has a rich cultivable land resource, ranking
fifth among the provinces of China. However, the per capita land area is low because of
the province’s large population, which has been one of its major challenges. The province
is rich in biological resources, with more than 10,000 kinds of utilizable plants, and ranks
second in China in terms of its flora and fauna resources. Sichuan is an important forest
area in China, and it has rich grassland and animal resources. Sichuan is the home of
panda. One hundred and twenty-three kinds of mineral resources have been discovered.
It is also rich in water transport channels, coal, natural gas, and biological forms of
energy. In addition, the natural and man-made landscape in Sichuan has also contributed
to the exploitation and development of the local tourism resources.

Sichuan tops China in hydropower resources. Sichuan is close to Tibet and its west part
features a very high altitude that leads to tremendous hydropower resources in this
province. According to the data from Sichuan Hydrology and Hydropower Institute,
theoretically, total hydropower resource in Sichuan is 143,000,000kW, or 27.4% of all that
of China; the technically exploitable total hydropower is 103,460,000kW, or 29.0% of all
that of China. Among all technically exploitable hydropower resources in Sichuan, only
10.6% has been exploited (as of 2002). Given China’s gross theoretical capability and
technically exploitable capacity are ranked number one in the world (see Table 4-1:
World Hydropower Capacity). The great potential of hydropower in Sichuan is obvious.

The hydropower resources in Sichuan mainly reside in Jinshajiang River, Yalongjiang

River, and Daduhe River in the West. These rivers have technically feasible capacity of
82,700,000kW, or 80% of all that of Sichuan. The scale of the step hydroelectric stations
on mainstream of the three rivers is mostly over 1,000,000KW. It is China's largest base
of hydropower exploitation and transmission from west to east China.
 72

Table 4-1: World Hydropower Capacity

World Hydropower Capacity (TWh/yr)

Gross theoretical Technically exploitable
Indonesia 2 147 402
India 2 638 660
Russian Federation 2 800 1 670
Brazil 3 040 1 488
United States of America 4 485 529
China 5 920 1 920
(Source: World Energy Council, 1999)

The features of energy resources in Sichuan is few coal, no petroleum, lack of natural
gas, and very abundant of hydropower. The composition of the energy resources in
Sichuan is as Figure 4-2.

Sichuan Energy Composition

0.75%
23.27%
Hydro
Coal
Natural Gas
75.98%

(Source: Sichuan Hydrology and Hydropower Institute, 2002)

Figure 4-2: Sichuan Energy Composition

 73

4.2. Sichuan power market

By the end of 2000, the total power generating capacity in Sichuan was 17,000 mW. The
amount of power generated in 2000 was 55.638Twh and grew 13.0% more than 1999.
The composition of the capacity and power generated is as Table 4-2.

Table 4-2 Sichuan Electric Power Generation Composition

Total % Hydro power % Thermal power

Capacity 17,000 mW 64.4% 35.6%
Power Generated 55.64 TWh 66.3% 33.7%
(Source: Sichuan Hydrology and Hydropower Institute, 2002)

Interestingly, in 1996, it was expected that the amount of electricity power

generated would be increasing at a rate of 6 –7 % per year till 2020. However, the actual
rate was almost double this figure till 2001. The total demand of the electricity power in
Sichuan was 52.123 TWh, grew 11.37% more than 1999 (Source: Sichuan Hydrology
and Hydropower Institute). The GDP growth rate of Sichuan was 5.6% in 1999, 9.1% in
2000, and 8.9% in 2001, it is expected to grow at a rate of 8% in 2002 (Source: Sichuan
Bureau of Statistics). So the growth of power generating capacity is faster than the
growth of the demand. (Source: Sichuan Hydrology and Hydropower Institute, 2002)
Figure 4-3 illustrates the trend of electricity power supply and demand in Sichuan.
 74

Trend of Electricity Suppliy and Demand in Sichuan

70
65 Actual power
60 generated
Forecasted Power
TWh

55
generated
50 Power demand in
45 Sichuan

40
1998 1999 2000 2001
Year

(Source: Sichuan Hydrology and Hydropower Institute, 2002)

Figure 4-3 Trend of Electricity Power Supply and Demand

It has already been over-supply of electricity in the region, as the World Bank financed
Ertan Hydropower plant is losing money because they cannot sell all the power that can
be generated. Though the demand of electricity is growing faster than expected, too
much new power generation capacity is built, especially many small thermal power plants
built by local governments. Since the growth of electricity power generating capability is
faster than the local economic growth, the oversupply is expected to continue. So the
current stress of the strategy of energy development in Sichuan is to send power out of
the province to the economically more developed east part of China, or so-called “West-
to-east power transfer”. However, this project involves huge amount of investment to
establish the grid.

Though, from a holistic view of China energy development, the hydropower resources in
Sichuan are more environmentally friendly and cheaper than thermal power burning coal
in the eastern economically more developed part. Although the hydro resources in
Sichuan are likely to be developed one day, the key question is how long this will happen
according to many factors, such as the grid build-up, the conflicts of interests of different
 75

regions, and other political and economical problems. Especially China’s economy is
under transformation to market economy, in other words, not a established market
economy now. So even if the region develop low cost and high quality electricity, it does
not necessarily mean that the region is able to sell it and pay back the loans, as it is
happening to the first Dam on Yalongjiang River, the Ertan Hydropower plant.

4.3. Introduction to Yalongjiang River

Yalongjiang River is the biggest tributary of the Jinshajiang River, which is a section of
the upstream of Yangtze River. It originates from Yushu County in Qinhai Province,
enters Sichuan near Kayishi. It turns 180 degrees around Mount Jinping, forming the
famous 150 km “great bend-over”. It enters Jinshajiang River near Panzhihua. The total
length of the river is 1570 km, watershed area 128,400 km2. The Yalongjiang River Basin
is located between 26º32’ ~ 34º20’N and 96º52’ ~ 102º48’. The average length of the
river basin is 950 km, and its average width is 137 km. The river drops quickly, from the
origin at an elevation of 5400m to the mouth at an elevation of 980m, total drop is 4420m.
Figure 4-4 provides a map of Yalongjiang River Basin.

The mainstream from Project C to the river mouth has a theoretical gross capacity of
18,140,000 kW, equals to 82.4% of the total of Yalongjiang River. To date, the only
hydropower plant has been built is Ertan. There are a lot more to do on the further
development on this section of the Yalongjiang River.
 76

(Source: Sichuan Hydrology and Hydropower Institute, 2002)

Figure 4-4: Map of the Yalongjiang River Basin

 77

4.4. Dams built and proposed

There are 4 major projects built or under serious scrutiny for further development. The
four projects are Ertan (built), Project A, Project B, and Project C as described below.
These hydro projects are mainly for energy production, with a secondary purpose of flood
control for Yangtze River. Because Yalongjiang River is located in a remote area with
few human activities where the economy is primitive and natural conditions are not
suitable for agriculture, the projects have virtually taken no consideration of the irrigation,
drinking, and recreational purposes.

4.4.1. Ertan

Ertan comprises a 240m concrete arch dam and Asia's largest underground powerhouse,
which is 280m long by 25.5m wide and 65m high. The construction began on 1991.
Loans for the project came from World Bank and China Development Bank. The project
was completed in 2000. It added a capacity of 3,300MW with annual energy output of 17
TkWh to the Sichuan and Chongqing grids. It eliminated power shortages in Sichuan for
the first time in 27 years, and also enhanced service quality and reliability. See Figure 4-5
for some pictures of the Ertan Dam.

(Source: http://www.power-technology.com/projects/ertan/)

Figure 4-5: Picture of Ertan Dam

 78

Ertan hydroelectric power station is the biggest power station China built in the 20th
century. It is an "engineering" success. However, it failed to pay back World Bank loans
for 2 years after it was put into use. When people designed the dam, few had expected
the current energy market in Sichuan and thought about taking any " insurance" measure.

4.4.2. Project A

Project A has already passed the feasibility study period, and people are doing final
design on the project. The hydropower station is to be located between Santan and
Shoupagou, around the beginning part the huge bend-over of the river around Mount
Jinping. The considered design options are described in Table 4-3:

Table 4-3: Project A Design Alternatives

Alternative 1 Alternative 2 Alternative 3 Alternative 4

Normal Water Storage Level m 1870 1880 1890 1900
Dead Water Level m 1790 1800 1810 1820
Reservoir Capacity billion m3 6.97 7.76 8.62 9.64
3
Adjustable Capacity billion m 4.52 4.91 5.32 5.75
Installed Capacity mW 3060 3240 3420 3600
(Source: Sichuan Hydrology and Hydropower Institute)

All the alternatives will add power-generating capacity of the Ertan Hydroelectric Plant as
Table 4-4.

Table 4-4: Project A Adding Capacity to Ertan

Alternative 1 Alternative 2 Alternative 3 Alternative 4

Increased Ertan Dry Season
mW 381 415 447 477
Power Generating Capacity
Increased Ertan Yearly Power
TkWh 0.94 1.05 1.13 1.22
Generation

(Source: Sichuan Hydrology and Hydropower Institute)

 79

The streamflow data for Project A is provided in Table 4-5.

Table 4-5: Streamflow for Project A

Unit: m3/s Qp
Average
p = 10% p = 50% p = 90%
Yearly 1200 1510 1180 904
Dry season (Dec - May) 437 505 435 371
(Source: Sichuan Hydrology and Hydropower Institute)

4.4.3. Project B

As the map of Yalongjiang River Basin as Figure 4-4, Yalongjiang River has a very big
turn-around at the Mount Jinping area. Project A is built at the beginning of the turn-
around. The Project B project will dig an 18 km long tunnel that bypasses 123 km of the
river. Shallow gradient river sections are bypassed and approximately 300m head can
be captured.

The choices for the installed capacity are 1,500 MW, 1,600 MW, and 1,700 MW. The
choices for tunnel water speed are 4.0m/s, 4.5m/s, and 5.0m/s. The tunnel diameter can
be calculated from the tunnel water speed. The choices for tunnel/turbine combination
are 3 tunnels 3 turbines, 2 tunnels 4 turbines, and 2 tunnels 6 turbines. The important
engineering parameter choices can be shown in Figure 4-6:
 80

Figure 4-6: Project B Tunnel Parameters

Project B would have 2 stages, each stage has an investment of 3.691 Billion RMB, has
an average power generating capacity of 11.578 TWh/yr, 2 tunnels and 4 turbines, each
turbine has an installed capacity of 400MW

4.4.4. Project C

Project C hydropower station would be located 2 km downstream to the connection of

Yalongjiang River mainstream and its tributary Xianshuihe River. Because Project C is in
a remote area, so the cost of flooding of the reservoir is virtually negligible.

The considered design options see Table 4-6.

Project C hydropower station has adjustable capability over year. The establishment of
Project C will add considerable power generating capacity to downstream stations. See
Table 4-7: Project C Adding Capacity to Downstream Stations.

The streamflow data for Project C is provided in Table 4-8.

 81

Table 4-6: Project C Design Alternatives

Alternative 1 Alternative 2
Normal Water Storage Level m 2840 2880
Dead Water Level m 2760 2800
3
Reservoir Capacity billion m 7.68 12.03
3
Adjustable Capacity billion m 5.28 7.49
Installed Capacity mW 2500 3000
(Source: Sichuan Hydrology and Hydropower Institute)

Table 4-7: Project C Adding Capacity to Downstream Stations

Alternative 1 Alternative 2
Increased Dry Season Power
mW 3109 2841
Generating Capacity
Increased Yearly Power
TkWh 8.906 10.642
Generation

(Source: Sichuan Hydrology and Hydropower Institute)

Table 4-8: Streamflow for Project C

Unit: m3/s Qp
Average
p = 10% p = 50% p = 90%
Yearly 657 829 649 495
Dry season (Dec - May) 283 342 281 227
(Source: Sichuan Hydrology and Hydropower Institute)

The study on the Yalongjiang River Basin will be divided into 2 steps: the first step is to
value Project A using real options method with consideration of the market uncertainties;
the second step is to study the River Basin as a whole using real options method with
consideration of both market and technical uncertainties.
 82

Chapter 5 Valuing Project A

The key part of this chapter applies binomial options pricing model to study a case on
Yalongjiang River basin development, more specifically, a deferral option of Project A. To
study the proc/cons and applicability of real options method in the valuation for Project A,
the NPV, IRR, NPV with simulation, IRR with simulation, and decision tree options
valuation are compared with the real options method.

5.1. Traditional NPV

Project A is scheduled to start generating power in the mid of the 11th year, and to
produce at full capacity in the 13th year. It is projected that the installed capacity will be
1,258 MW in the 11th year, 2516 MW in the 12th, and 3564 MW in the 13th. According to
the historical data, an average waterflow per year can be derived. The production is to
be 5.94 TWh in the 11th year, 11.88 TWh in the 12th, and 16.83 TWh a year thereafter.
Based on the guidelines China is now using to calculate energy price, the price for the
electricity is 0.25 RMB/KWh. The revenue is the annual production times the price, and
the time frame used is 60 years with no residual value. 60 years is the expected life of
the project, and any variation of the life to 70 years or 50 years will not affect the result a
lot. It is not a big effort to study a time frame as long as 60 years with an Excel
spreadsheet. The investment is taken to be 14.6 Billion RMB as in year 0. The actual
investment happens each year during the construction. Discounting the investment each
year back to year 0 using the interest rate of 5.72% and adding all of them up, we get the
equivalence of total investment as of 14.6 Billion RMB in year 0.

The NPV analysis is as Table 5-1:

 83

Table 5-1 Project A NPV Analysis

Year 0 1 … 10 11 12 13 … 60
Quantity (in TW h) 0 0 0 5.94 11.88 16.83 16.83 16.83
Price (in RMB/KW h) 0.25 0.25 0.25 0.25 0.25 0.25 0.25 0.25
Revenue (in B RMB) 0 0 0 1.485 2.97 4.2075 4.2075 4.2075
Investment (in B RMB) 14.6
Cash Flow -14.6 0 0 0 1.485 2.97 4.2075 4.2075 4.2075

The NPV is 7.02 Billion RMB using a discount rate of 8%. With such an analysis,
because the NPV is greater than 0, Project A should be carried out immediately. All the
analysis is a point estimate of value and the corresponding strategy is a go or no-go
decision without a third choice, such as “wait and see”.

Note in the above net present value analysis, there are three very important assumptions:

1, the power plant will produce to full capacity, and all the electricity produced can
be sold.
2, the price is fixed and will not change in 60 years.
3, here an 8% discount rate is used, the same as the normal discount rate used in
the China Development Bank.

These three assumptions are hard to defend:

1. How can a plant always produce at full capacity and sell all the electricity
produced for a period of 60 years? It is not only a simplified case, but also the
most optimistic case and apparently not realistic. For example, Ertan only
produced at half of its capacity for its first two years. Also, all the quantity of
electricity produced is based on the average annual waterflow; the actual
waterflow fluctuates every year.
2, it seems unreasonable to assume that the price for electricity does not change
over 60 years. Even in the case the power supply is allocated by government, the
 84

price is decided by some government plans, it is still unimaginable that the price
would not change.
3, the choice of discount rate is always a problem when evaluating public projects,
there is no concrete logic or proof for a specific discount rate chosen, and leaves
this topic to a game of politics. Here the discount rate applied is 8%, but the
evidence why it should be 8% was not and cannot be provided.

Although there are apparently unrealistic assumptions behind the NPV method, the
method is still ubiquitous. The reason is the computation difficulty of more refined models,
and the NPV is much better than nothing. However, with the development of computer
technology, more refined models can be easily set up and calculated.

5.2. Traditional Internal Rate of Return (IRR)

The Internal Rate of Return (IRR) for Project A can also be calculated. It is 10.05%. IRR
provides a better measure than NPV because it avoids the problem of choice of discount
rate, and makes it possible to compare the project with the expected return or capital cost
to decide what to do. Applying the traditional IRR valuation, the conclusion is that as long
as the capital cost is less than 10.05%, the best strategy is to build the dam immediately.
Like the traditional NPV method, there are only go or no-go decisions, no third choice.

5.3. NPV with Simulation

In the Project A cash flow simulation model, the key uncertainties affecting quantity and
price are decomposed and analyzed, thereby establishing the function between the cash
flow and the uncertainties. Then the uncertain factors are simulated to get thousands of
 85

realizations of the cash flows over 60 years. For each realization over the 60 years, a
NPV or IRR is calculated. With the samples of thousands of NPV, the expected NPV can
be calculated.

A generalized framework of engineering systems analysis with focus on uncertainty can

be seen as Figure 5-1:

Engineering design needs to take into account both the technical uncertainty and the
market uncertainty. People often design engineering systems assuming the technical
parameters are deterministic, which is often not the case. For example, if a reservoir and
hydropower plant design is based on the waterflow data of a typical year, the design
would miss an important element, i.e. the uncertainty of waterflow and thus the uncertain
amount of electricity generated. Though market uncertainty is more recognized, ironically,
the market analysis is still dominated by deterministic NPV projections.

Another tendency is that, for a big engineering system, engineering features are taken
care of by engineers, and marketing side is taken care of by marketing people. Although
a great lot of literature emphasizes the importance of integrating engineering and
marketing, the technical and marketing sides of an engineering system are taken care of
by two functions with totally different kind of people and thinking. Consequently, the
engineering systems are analyzed separately. Engineering people optimize the system
given technical constraints, while marketing people optimize sales given market
conditions. The result is that the two parts are near optimization given constraints for the
specific part respectively, but the whole system is suboptimal. So it is important to take
the engineering system optimization into more holistic settings, considering technical and
marketing in an integrated model.

In Figure 5-1, the engineering systems model with uncertainties can be divided into two
sub-models: the first one is a technical sub-model, which analyzes the technical
constraints; the second one is a market sub-model, which analyzes the market
 86

uncertainties. The key difference of this model compared to many previous models is
that it explicitly integrates the technical and market sub-models.

Applying the above framework specifically to the problem of the Yalongjiang River basin
development, the model in Figure 5-1 can be redrawn as Figure 5-2:

Technical Submodel Market Submodel

Technical Market
Uncertainty Uncertainty
Demand Price

Engineering Utilization Utilization

Benefit
Design (Technical) (Market)

Cost

Figure 5-1 Engineering Systems Model with Uncertainties

Technical Submodel Market Submodel

Waterflow Market
Uncertainty Uncertainty
Demand Price

Installed Utilization Utilization

Revenue
Capacity (Technical) (Market)

Cost

Figure 5-2 Yalongjiang River Basin Model with Uncertainties

 87

5.3.1. Technical Sub-Model

The main purpose of the Yalongjiang River basin development is power generation, so
installed capacity is one of the most important engineering parameters. The uncertain
waterflow each year makes the average utilization of the capacity less than 100%. In the
years with a lot of water, water has to be wasted; while in dry years, the capacity cannot
be fully realized.

To some extent, this situation is like holding a stock and being short a call option on that
stock. If at the expiration day, the stock price is greater the exercise price, then the
option will be exercised, and the payoff is the exercise price; if at the expiration day, the
stock price is less than the exercise price, the option will not be exercised, the payoff is
the price of the stock then. The payoff chart is as Figure 5-3.

The relationship of power generated and waterflow is similar, as in Figure 5-4.

Figure 5-3 Payoff Diagram of Longing a Stock and Shorting a call

 88

Power Generated

Installed Capacity

Waterflow

Figure 5-4 Power Generation and Waterflow

Given the waterflow distribution, the distribution of the utilization of a specific installed
capacity can be obtained. The maximum utilization is 100%, which means the maximum
amount of electricity produced is the installed capacity, additional water has to be wasted;
the utilization can be less than 100%, when it is dry season or dry year, there is not
enough water for the hydropower plant to run whole capacity. With the distribution of the
utilization, the actual power generating capacity can be simulated. As a first
approximation in this thesis, the technical utilization is taken to be full, or 100%.

5.3.2. Market Sub-Model

The revenue from selling electricity is determined by the amount sold and its price. The
amount sold is determined by the demand and supply of the market. The demand of
electricity is highly correlated with the GDP growth, and is taken in this thesis to grow at
the same speed as GDP. The supply of electricity can be divided into thermal, hydro,
and nuclear power. A key assumption in the first simple model is that the market for
energy is complete, which means that the market always first buys the cheapest power, if
the cheapest power is used up, then it turns to the next cheapest power, and so on.
Because there is no nuclear power in Sichuan, the types of power we are considering are
hydro and thermal. The marginal cost of producing hydropower is cheaper than that of
 89

thermal, so the market uses hydropower first. After no more hydro capacity available, the
market turns to the thermal power, as in Figure 5-5:

Price

Demand

Supply
Thermal

Hydro

Quantity

Figure 5-5 Assumed Power Supply and Demand Curve in Sichuan

Uncertainties

Demand

It is assumed that the demand of electricity is highly correlated with the local GDP growth.
In the model, the demand is growing at the GDP growth rate beginning from the demand
for 2000, i.e., 52.123 TkWh. In order to model the GDP growth stochastically, a simple
mean reverting model is used. Its mathematical expression is as Equation 5-1:

dr = η (r − r )dt + σdz Equation 5-1

 90

where r is the GDP growth rate, η is the reversion parameter signifying the strength of

the reverting process, r is the long-run mean of GDP growth rate, σ is the volatility, and
dz is a basic Wiener process with a drift rate of zero and a variance of 1.0.

This mean reverting model points out the trend that when a country is developing at a
high speed, the probability that it grows even faster is smaller than the chance that the
growth rate decreases; when a country’s economy is stagnating and slower than the
long-run mean, the chance of regaining growth is bigger than growing even slower. For
the long run, the growth of an economy tends to converge to an average, while the n+1
year’s growth rate is related to the growth rate for year n. This is a simple yet acceptable
model for the first model for the Yalongjiang river basin development.

This thesis uses some plausible numbers for the parameters for a first try. As a next step,
the history of economy growth of Japan, Korea, or other countries may be studied to help
getting such parameters, based on the assumption that the current China economy will
be similar to some other countries when their economies took off.

Supply

This thesis assumes the power plants compete in a complete market. Because
hydropower is cheaper than thermal, so the market will first use hydropower, and then
use thermal for the amount of power needed more than that can be supplied by hydro. If
the hydropower supply is greater than the total market demands, each hydropower
company is producing

its hydro capacity X (total demand / total hydro capacity)

Total exploitable hydropower resources in Sichuan are 290 TkWh, assuming all
exploitable resources will be developed in 30 years. The growth of hydropower supply is
staged. Each time a planned new hydropower station is put into use, the total supply is
 91

increased. The growth begins on the basis of the power production in 2000 of 39.61
TkWh.

Price

The price of electricity is a highly uncertain in China right now. This is because China is
going to reform its energy market to allow free competition. Currently, the energy in
China is still mainly under the old planned system, and exhibits a lot of inefficiency. For
example, the Ertan Hydro Power failed to sell even half of its capacity for the first year
after it was put in use. This put Ertan in a very awkward position. Because of their huge
loans for building the project, one year’s additional interest of the principal when they just
begin to operate seems too big a burden for them to recove. The World Bank ranked the
project unsatisfactory because their financial difficulties, though all the other aspects of
the project are very good, from engineering to environmental to reallocation of residents.
The key reason of this impasse is because local electricity authorities refused to use the
hydropower from Ertan though it is much cheaper. They used electricity generated by
many small thermal power plants whose price is much higher. The thermal plants also
cause serious environmental problems because they burn coal. The reason is simply
because the local electrical power authorities have direct or indirect investments in those
newly built thermal power plants. They also have to repay loans.

Given the highly complex economical, social and political surroundings in China, it is hard
to predict the price of power after the energy market reforms. Current price data are all
but unusable because they are highly distorted, and there will be a system discontinuity
after the reforms that make everything different. The fact that the marginal cost of
hydropower is low makes the price of the electricity even harder to predict.

The parameters for analysis are based on expert estimates. I talked with two Chinese
experts on China energy market in the China Development Bank to get their low, most
 92

likely, and high estimate of the electricity price for 3 years later1. The experts reached the
high price estimate of 0.315RMB/kWh and low 0.18RMB/kWh both with 95% confidence.
The experts estimate the electricity price will drop 0.33% per year.

The two experts in the China Development Bank believe that the major energy market
reforms are done by 2007. Please refer to a report on China energy market on the
China official media, People’s Daily, as Appendix II. It is assumed that the energy price
will then be mostly subject to the inflation factor and more predictable. In the model
established, all terms are in real money, and inflation does not need to be considered
explicitly.

5.3.3. Simulation

Having established the model, it is possible to simulate the NPV using a software
package called Crystal Ball. The procedure is to:

• identify the basic factors that influencing a specific outcome,

• determine their distributions,
• draw a set of samples for each of the factors according to its distribution,
• get the value of the outcome that we are interested in with each set of samples,
• repeat this a great number of times to get correspondingly number of samples for
the outcome of interest (in this case, the NPV). The distribution of the samples
converges to the actual distribution with probability one.

The process is called Monte-Carlo simulation.

1
The purpose of this thesis is to study the methodology of real options, not to offer real decision
support. So the accuracy of the parameters is not explored further. However, for particular
applications, several ways to estimate the parameters need to be used concurrently to crosscheck
the validity of the estimates.
 93

For example, if O is a function of random variables X1 and X2, or

O = f (X1, X 2 ) Equation 5-2

If X1 follows a normal distribution and X2 follows a triangular distribution, a set of samples

for X1 and X2, say X11 and X21, is randomly drawn according to the distribution of the two
random variables, i.e., normal and triangle distribution. Then X11 and X21 are put into the
function to get

O1 = f ( X 11 , X 21 )

Repeat this process to draw another set of samples for X1 and X2 to calculate O2. Repeat
it a large number, say 10,000, times. Then the distribution of O1, O2, …, O10000 is
obtained. Mathematically, it converges to the true distribution of O with probability one,
and different statistics on these 10000 values can be obtained to see how close it is to
the true distribution.

In this hydropower case, a model is established to describe the uncertainties (with certain
distributions) and the function between the uncertain variables and the subject of interest
(NPV in this case). Then the model is simulated for a great number of times, say 2000. It
is like that the project is implemented for 2000 times, 2000 NPVs corresponding to the
2000 times are obtained, and thus the distribution of the NPV is obtained. Now the
expected value of the NPV is easy to calculate. For Project A, Monte-Carlo simulation is
applied to the spreadsheet with the tool of Crystal Ball. The distribution is as Figure 5-6.
 94

Forecast: NPV I
2,000 Trials Frequency Chart
.02 53

.02 39.7

.01 26.

.00 13.2
Mean = RMB 0.98B
.00 0
RMB –8.52B RMB –2.84B RMB 2.84B RMB 8.52B RMB 14.19B

Figure 5-6 Distribution of NPV for Project A

The expected value of the NPV is 0.98 Billion RMB. It is distinct from the traditional NPV
result of 7.02 B RMB. This is a notable result, and signifying some important limitation of
the traditional NPV method:

1. The logic of traditional NPV method obtains the expected values of the important
factors and plugs them into Equation 5-2 to get

O = f ( E ( X 1 ), E ( X 2 )) .

Specifically, in the NPV calculation above, X1 is the quantity of electricity sold, and X2 is
the price with which the electricity to be sold, and O is the NPV. E(X1) is taken to be the
capacity of the hydropower plant, or 16.83 TWh/year1, and E[X2] is the expected price for
which the electricity is sold, taken to be 0.25 RMB in the calculation for traditional NPV.
With E(X1) and E[X2], the value of traditional NPV is calculated to be 7.02 B RMB.

1
During the first two years when the plant just put into use, the capacity is assumed to be 1/3 of
the total capacity for the first year, and 2/3 of the total capacity for the second year.
 95

In most cases, however,

E[ f ( X )] ≠ f [ E ( X )] .

The only case that the equality is guaranteed is when the f(X) is a linear function of all
uncertainties. Savage [2000] had a discussion on this issue what he called “flaw of
averages”, which states “plans based on the assumption that average conditions will
occur are usually wrong”.

2. What makes things even worse is that it is usually hard to identify the expected value
of uncertainties. People tend to take values that are easy to get as the expected values,
partly because the expected values are hard to find out. For example, in the traditional
NPV calculation above, the value plugged in to calculate the NPV is not even the
expected quantity of electricity produced, but the maximum of quantity can be sold (the
capacity of the power plant, which is an upper limit of the quantity that the amount of
electricity can be produced, and any real produciton of electricity can only be less or
equal than this amount). It is the most convenient value to get, but it is not the right value.
Though the price of the electricity is actually the expected value, the above calculation for
traditional NPV sues the formula:

O = f ( Max( X 1 ), E ( X 2 ))

So the value is seriously distorted, and too much optimistic. As found using the
simulation technology, the expected NPV for project A is only 0.98 B RMA, while the
traditional NPV calculation renders a value as high as 7.02 B RMB. This is not an
uncommon practice in the application of NPV methodology.
 96

5.4. IRR with simulation

The distribution of the IRR can also be obtained using the Project A cash flow simulation
model. Because IRR is better than NPV in the sense that it avoids the tough problem of
choice of discount rate, the distribution of the IRR can also deliver a lot of information
useful for decision-making. See the result of the IRR simulation as Figure 5-7:

Forecast: IRR I
2,000 Trials Frequency Chart 23 Outliers
.027 54

.020 40.5

.014 27

.007 13.5

Mean = 0.088
.000 0
0.049 0.070 0.090 0.111 0.131

Figure 5-7 Distribution of IRR for Project A

Notice that the average IRR presented in Figure 5-7 as 8.8%. When it is interpreted as
the expected IRR, there is complexity underneath this issue because of the nonlinearity
of IRR. The interpretation of the IRR simulation result should be based on the figure as a
whole, such as the range of the IRR, and average, and both tails.

This figure will render much more information than traditional IRR. And the same
principle of “flaw of averages” as NPV applies to IRR calculation. In general:
 97

IRR( E[ X 1 ], E[ X 2 ]) ≠ E[ IRR( X 1 , X 2 )]

Where X1 and X2 are random variables.

The traditional IRR is 10.05%, while the average of simulated IRR is 8.8%. One of the
reasons for this is that the quantity of electricity sold used for the calculation of traditional
IRR is the maximum value, not the expected value as identified in section 5.2. .

5.5. Real Options Analysis

This section compares two possible variants of real options analysis (ROA). The first
uses the electricity price as underlying and the second uses the NPV of the project as the
underlying.

5.5.1. A General Binomial Tree Framework for ROA

The general framework is the same for ROA with NPV or electricity price as underlying.
This section introduces the general binomial tree framework for ROA, and the following
sections will study the real options value with either underlying.

Event tree

Firstly, a binomial tree needs to be built based on some underlying describing the market
uncertainty. After building the binomial tree, a series of scenarios are developed with
probabilities, and what happens in each scenario can be analyzed. So a binomial tree
can be thought as an event tree in essence. To get such an event tree describing the
 98

market uncertainty, one way is to lay out the scenarios of different NPVs, another way is
to present the evolution of the electricity price.

Analysis of each scenario

Given the scenarios of underlying evolution, the expected payoff of each option can be
established. As Figure 5-8, a project with option 1, option 2 … option n1 is valued. Given
the underlying value M, payoffs for each of those options can be calculated. The hold
value is calculated by discounted the expected value of next phase of the binomial tree.

M
Payoff for option 1
Payoff for option 2
Payoff for option 3
…
Hold value

Figure 5-8 Analysis of a scenario on the event tree

In the case of analyzing the deferral option for Project A, there is only one option. Only
the exercise value and hold value are entered into Figure 5-8.

Roll back

After establishing the options values for each scenario, the tree can be rolled back to get
the value of the option. The value for each scenario is

Max (payoff for option 1, payoff for option 2, …, payoff for option n, the hold value)

1
If n > 1, it is a compound option.
 99

Beginning from rightmost scenarios, the up probability pu and down probability pd are
applied to roll back to get the hold value for any scenario. Note the discount rate used
must be risk adjusted. The way to risk adjust the discount rate is to simulate the IRR and
get the standard deviation of the IRR. Note that different discount rates need to be
applied to different cash flows with distinct risk properties.

After rolling back the leftmost scenario, not only the value of options but also a
contingency strategy can be developed.

5.5.2. ROA with electricity price as underlying

The first step for a ROA with electricity price as underlying for Project A was to estimate
the drift rate r and the volatility σ. For the options on stocks, the drift rate r is the risk-free
interest rate as shown in Section 3.4. Risk-neutral Valuation. If the underlying is
electricity price, however, the drift rate cannot use the risk free rate, it should be the
expected change of electricity price per period of time.

Estimation of drift rate r and volatility σ for electricity price

The key uncertainties under the valuation of the Project A are the market features of
demand and price. Assuming that the market is complete, people will first use all the
hydro capacity and then turn to thermal. Given that in Sichuan, total power consumption
was 55.638TWh in 2000. Among the total consumption, 66.3% is hydropower, so a first
simple assumption is that, in the future, the total hydropower supply will always less than
or equal to the total electricity demand of Sichuan given the growth scheme of supply of
hydropower described in section 5.3.2. Therefore, all the power produced by a hydro
plant will be able to find a buyer.
 100

Then price becomes the only market uncertainty in this analysis. Because China is
undergoing aggressive energy market reform, the price of electricity in the future is highly
uncertain. Projecting what happens in 3 years, the experts [China Development Bank,
2002] give the optimistic price estimate of 0.315 RMB/kWh and pessimistic
0.18RMB/kWh, both with 95% confidence. And the experts estimate the electricity price
drops 1%. Therefore, the drift rate r of electricity price is –0.33% per year.

Defining the volatility of electricity price σe can be defined as the standard deviation of
P1
ln( ) in one year, where P1 is the electricity at year 1 and P0 is the electricity price at
P0

year 0. The standard deviation for 3 years is 3σ e . And the 95%confidence means 2 σe
range. So

0.315
ln( )
0.25 × (1 − 1%)
3σ e =
2
or

σ e = 6.96%

Risk-adjusted discount rate

The underlying of this binomial model is electricity price, and the drift rate of –0.33% is
the actual rate not in a “risk-neutral world”, in other words, the expected growth rate is not
the risk-free interest rate. The scenarios established by the binomial tree are not in a
risk-neutral world, but the real world. Since the scenarios are in the real world, risk-
adjusted discount rate rather than risk-free rate applies. The discount rate adjusted to
the specific risk profile of the project needs to be determined.
 101

The key development that makes it possible to find out a risk adjusted discount rate for a
specific project is the power of computers to simulate complex systems more readily and
easily. After simulating, the measure of the risk of the project, or σ2, can be obtained.
Modern finance theory can help get the corresponding market expected rate of return for
efficient assets with such σ2, which should be the upper bound of risk adjusted discount
rate.

Previously, the choice of discount rate is always a controversy. The root of this problem
is that people did not have a way to decide the discount rate from the intrinsic
characteristics of a project, and had to turn to some extrinsic sources to decide the
discount rate that does not have a convincing logic for a specific project. For example,
US Office of Management and Budget has regulations regarding the discount rate should
be used. Actually, finance theory states the discount rate can be decided by three
parameters, that is rf, the risk-free rate, σ, the measure of risk, and pσ, the market price
for risk. The rf and pσ are readily observable in the financial market. Unlike stocks or
many other financial instruments, σ is hard to get for many public projects because no
corresponding stock is traded in the market, and the σ cannot be derived from other
stocks or projects because of the uniqueness of every project. However, the measure of
risk of a project can be obtained with simulation.

[Capital market line and CAPM]

According to Luenberger [1998], the capital market line as Figure 5-9 in shows the
relation between the expected rate of return and the risk of return for efficient assets or
portfolio of assets. M in Figure 5-9 represents the market portfolio, or the summation of
all assets on the market.
 102

rf

Figure 5-9 Capital Market Line

The relationship in Figure 5-9 can be described by a straight line in mathematical terms:

rM − r f
r = rf + σ Equation 5-3
σM

where rM and σM are the expected value and the standard deviation of the market rate of
return of the market portfolio, r and σ is the expected value and the standard deviation of
the market rate of return of an efficient asset that earns the highest possible expected
rate of return given a specific risk profile.

The slope of the capital market line is pσ = (rM – rf)/σM, and this value is frequently called
the price of risk. So the capital market line can be also written as:

r = r f + pσ ⋅ σ Equation 5-4

Note the capital market line does not show how the expected rate of return of an
individual asset relates to its individual risk, such as Project A in this thesis. The relation
is expressed by the capital asset pricing model (CAPM) as:
 103

ri = r f + β i (rM − r f ) Equation 5-5

σ iM
βi = Equation 5-6
σ M2

where ri is the expected return of asset i, and σiM is the covariance of the rate of returns
between asset i and the market portfolio.

The empirical problem for valuation of Project A is, however, that it is virtually impossible
to find out the σiM that describes the covariance of the rate of return of Project A and that
of China finance market. The CAPM model is not applicable in this case.

The capital budget line, however, defines the efficient boundary of the feasible set of the
expected rate of return and the standard deviation of rate of return for any individual
asset. This means that any point on the capital market line is the highest possible
expected rate of return of the given the standard deviation of the rate of return of the
assets or portfolios of assets.

So if the capital market line is used, the upper bound of the risk-adjusted discount rate
can be found. As an approximation, this thesis adopts the upper bound as the risk-
adjusted discount. Since the upper bound of the discount rate is used, the valuation is
lower than the actual value. It is good in the sense of being conservative, meanwhile.

[The standard deviation]

After building up Project A cash flow simulation model, it is assumed that the life of the
project is 60 years. The Internal Rate of Return (IRR) can be calculated from each path of
the simulated cash flows for 60 years. Assuming the value for the project at year 1 is V1,
and the invest is I, invested at year 0, CFi is the cash flow for year i, and the IRR is r. So
 104

60
CFi
−I +∑ =0
i =1 (1 + r ) i

while

60 CF j
V1 = CF1 + ∑
j =2 (1 + r ) j −1

so
V1
−I + =0
1+ r

which means

V1 = I (1 + r )

Applying Crystal Ball to the model of Project A, and simulated the IRR as Figure 5-7. The
standard deviation in Figure 5-7 is 0.0115. Notice because this standard deviation of the
IRR of Project A, or σv is averaged out for 60 years, so it is

σ v / 60 = 0.0115

so the standard deviation of IRR

σ v = 0.0115 × 60 = 0.089
 105

[The risk-adjusted discount rate]

What is the risk free rate in this analysis? Examining the China Treasury Bill as August
2002, the longest is 20 years with an annual rate of 4.26%. Because the life span of the
project is 60 years, longer than 20 years, risk free rate is a little bigger than 4.26%. This
thesis uses 4.5%. According to the website about China finance market at
www.genius.com.cn/data/, the expected value and standard deviation of the market rate
of return are approximately 18% and 30%. So the market price of risk

rm − r f 18% − 4.5%
pσ = = = 0.45
σm 30%

All the parameters chosen here are only for illustration. A better examination to
determine the parameters is needed when real decision is made based on this model.

Substituting rf, pσ, and σ into Equation 5-4, the discount rate used for Project A is

r = r f + pσ ⋅ σ v = 0.045 + 0.45 ⋅ 0.089 = 0.086

2

The risk-adjusted discount rate r represents the intrinsic risk profile of the project.

Binomial Tree

The time period of the tree is 12 years with per stage of 3 years. The decision of starting
the construction is likely to be made in the coming 10 years, though the life span of the
hydropower project considered is 60 years. If a stage of 1 year used, there will be 10
stages. 10 stages are too many for this thesis, whose main purpose is to illustrate the
method. So the stage is taken to be 3 years.
 106

With r and σ specified, according to Equation 3-13 to 3-15, the various parameters for per
stage are as follow:

3σ e
u=e = 1.128
1
d= = 0.886
u
er − d
pu = = 42.87%
u−d
p d = 1 − pu = 57.13%

Then the electricity price movement is established as the binomial tree in Figure 5-10:

Year 0 Year 3 Year 6 Year 9 Year 12

pu 0.40
pu 0.36 pd
pu 0.32 pd pu 0.32
pu 0.28 pd pu 0.28 pd
0.25 pu 0.25 pd pu 0.25
pd 0.22 pd pu 0.22 pd
0.20 pd pu 0.20
0.17 pd
0.15

Figure 5-10 Electricity Price Movement

Note that this event tree gives different scenarios of the evolution of the market. For each
scenario, the exercise value and deferral value can be calculated.

The exercise value is just the expected NPV of starting Project A given the specific price
in the scenario. Given the starting price of electricity in a specific scenario, the Project A
cash flow simulation model can be used to get the expected value in this scenario. For
example, for the scenario in which the electricity price is 0.22RMB/kWh, 0.22 is put as the
 107

starting price of electricity in the Project A cash flow simulation model, and the expected
NPV of the project is –0.68 B RMB. See Figure 5-11.

The deferral value is the expected value if the option is not exercised. The deferral value
is

( value in up state 3 years later) ⋅ p u + ( value in down state 3 years later) ⋅ p d

(1 + discount rate) 3

The risk-adjusted discount rate for Project A should be 8.6%.

The value of the project with deferral options at any point of time is the maximum of the
exercise value and the deferral value.

The binomial tree developed for this Project A deferral options is shown in Figure 5-11.

The value for the deferral option is 1.25 – 0.98 = 0.27 billion RMB. 1.25 billion RMB is
the value of Project A with deferral options. And 0.98 billion RMB is the value of the
project without deferral options, in other words, the value of the project if it has to begin
right now.

The tree in Figure 5-11 allows people to get important insights into the strategy of this
project. Given the high volatility of price due to the pending energy market reforms, the
best strategy is to wait until the price of electricity is mostly resolved. If the price is high
then beginning building the plant, if it is low then still wait.
 108

Year 0 Year 3 Year 6 Year 9 Year 12

Price 0.25 0.28 0.32 0.36 0.40
Exercise (B RMB) 0.98 3.24 5.41 8.04 10.85
Deferal 1.25 2.24 3.02 6.04 0.00
Value 1.25 3.24 5.41 8.04 10.85
Exercise? no yes yes yes yes

Price 0.22 0.25 0.28 0.32

Exercise (B RMB) -0.68 0.97 3.24 5.41
Deferal 0.37 0.90 2.24 0.00
Value 0.37 0.97 3.24 5.41
Exercise? no yes yes yes

Price 0.20 0.22 0.25

Exercise (B RMB) -2.09 -0.68 0.97
Deferal 0.11 0.32 0.00
Value 0.11 0.32 0.97
Exercise? no no yes

Price 0.17 0.20

Exercise (B RMB) -4.02 -2.09
Deferal 0.00 0.00
Value 0.00 0.00
Exercise? no no

Price 0.15
Exercise (B RMB) -5.23
Deferal 0.00
Value 0.00
Exercise? no

Figure 5-11 Binomial Tree for ROA with electricity price as underlying

Sensitivity Analysis

The option value of 0.27 Billion RMB is 27.6% of the value of the project that is taken to
be as the NPV with simulation of 0.98 Billion RMB. The option value should be
compared to the NPV with simulation rather than the cost of the project of 14.6 Billion
RMB, because the option value is based on the value of the project that is revenue net
cost. On each node of the binomial tree as Figure 5-11, the exercise value of the option
is the NPV with simulation, given the specific value of underlying electricity price.

The key element deciding the value of the option is the volatility. Besides the volatility of
6.96% as the guess of the experts, a sensitivity study on various values of volatility and
 109

drift date must be done. The result is shown in Table 5-2. The first value in each cell is
the option value, and the second value is the percentage of the option value to the total
investment.

Table 5-2 Sensitivity Analysis to Option Value

(Option Value in Billion RMB

percentage of total investment)
Volatility
Drift rate
6.96% 10% 15% 20% 25% 30%
-0.01/yr N/A* N/A* N/A* N/A* N/A* N/A*
0.27 0.45 0.97 1.55 2.13 2.59
-0.0033/yr
1.85% 3.08% 6.64% 10.62% 14.59% 17.74%
0.45 0.64 0.88 1.87
0.005/yr 0 0
3.08% 4.38% 6.03% 12.81%
0.24 0.59 0.90
0.01/yr 0 0 0
1.64% 4.04% 6.16%
0.02/yr 0 0 0 0 0 0
0.08/yr 0 0 0 0 0 0
15.74 15.93 16.29 17.98 20.20
0.09/yr N/A**
107.81% 109.11% 111.58% 123.15% 138.36%
60.19 63.86 64.96 66.46
0.10/yr N/A** N/A**
412.26% 437.40% 444.93% 455.21%
* because the NPV with simulation is less than 0, there is no option for waiting or not. The only
choice is to wait, not carry on the project. It is the area of “Option Value not Defined (II)” in Figure
5-12.
** for the binomial tree formula used, the volatility has to be greater than the drift rate. Otherwise,
no option value can be calculated. It is the area of “Option Value not Defined (I)” in Figure 5-12.

The option values given different combination of volatility and drift rate can be sketched
as in Figure 5-12 . For high drift rate and low volatility, the option value is not applicable
because the binomial model requires that volatility greater than drift rate. In the middle of
the diagram, the option value is 0. The option has a positive value both when the drift
 110

rate is big1 and small. An intuitive explanation is that an option has value either to avoid
downside risks when drift rate is low or to wait for even better chance when drift rate is
high. Option has value when situation is problematic or promising. The latter
observation is a little counterintuitive without an options analysis. Without an options
analysis, the decision was often to go immediately when it looks promising while
overlooking the chance that things could be better by waiting.

Figure 5-12 Option Value as Function of Volatility and Drift Rate

1
Although a drift rate as high as 8% per year is not reasonable for a hydropower case in this
thesis, the sensitivity to high drift rates is done, nevertheless. This is to illustrate a general insight
gained by options analysis.
 111

5.5.3. ROA with NPV as underlying

For the ROA with NPV as underlying, the underlying asset of the analysis is the value of
the project, or the NPV of the project without flexibility. As calculated above, the
expected value of the NPV with simulation is 0.98 B RMB. Because the investment is
14.6 B RMB, so the expected revenue of the project is 14.6 + 0.98 = 15.58 B RMB. The
underlying has an initial value of 15.58 B RMB, and the exercise price of the option is the
investment, or 14.6 B RMB.

The next step is the get the volatility. Remember the IRR for the project has been
simulated as Figure 5-7, and the standard deviation of the IRR is 1.15% (the mean is
8.8%). The IRR is the average IRR over a lifetime of the project of 60 years. So the

standard deviation of 1.15% is equal to σ . Therefore,

60

σ v = 8.9% 1

With σ obtained, the up factor and down factor for a 3-year period are as follow:

3σ v
u=e = 1.167
d = 1 / u = 0.857

Here the risk free rate rf is taken to be 13.5% because the time interval is 3 years, and
the risk free rate is 4.5% per year. Therefore, the risk neutral probabilities

1
Note the volatility should be the standard deviation of logarithm of IRR. However, since the
standard deviation of IRR is very close to the standard deviation of logarithm of IRR because both
are very small. So the volatility is approximately taken to be the standard deviation of IRR.
 112

r
e f −d
pu = = 0.928
u−d
p d = 1 − pu = 0.072

With the information about the initial value of 15.58 B (net revenue without flexibility), the
exercise price as of the investment of 14.6 B, the parameters of u, d, pu, and pd, the
binomial tree is established as Figure 5-13.

The project value with option is 2.83 B RMB, so the value of option is 2.83 – 0.98 = 1.85
B RMB.

Applying such ROA with NPV as underlying, though we get the value of the option, there
are several problems:

1, the underlying is the NPV of the project, and the exercise conditions is not clear
because the NPV of the project is not readily observable. The NPV itself at any point is
not determined. People can only get the expected value of NPV at a specific node. For a
financial option, at any time, the underlying asset of stock price is readily observable and
determined. If the exercise condition is not clear, then it is unknown whether the
hypothetical exercise in the binomial tree is practical. So the value of the option is
questionable if the options are not exercisable themselves.

2, the binomial tree as in Figure 5-13 assumes that on each node the underlying is a
determined and observable measure, such as stock or other commodity price. But the
NPV is not determined at any point of time. It has a distribution. In other words, it has
variance or risk, so just applying a risk free rate to discount is not enough (though it is
correct for the case where the underlying are determined, for example using stock price
as the underlying where, at the time people exercise the stock option, the stock price is
determined). Because systematically applying a small discount rate (risk-free rate) than
necessary, the valuation tends to be too big.
 113

Year 0 Year 3 Year 6 Year 9 Year 12

Project Value 15.58 18.18 21.21 24.75 28.88
Exercise (B RMB) 0.98 3.58 6.61 10.15 14.28
Deferal 2.83 4.33 6.17 9.39 0.00
Option Value 2.83 4.33 6.61 10.15 14.28
Exercise? no no yes yes yes

Project Value 13.35 15.58 18.18 21.21

Exercise (B RMB) -1.25 0.98 3.58 6.61
Deferal 1.75 2.72 4.24 0.00
Option Value 1.75 2.72 4.24 6.61
Exercise? no no no yes

Project Value 11.44 13.35 15.58

Exercise (B RMB) -3.16 -1.25 0.98
Deferal 0.39 0.62 0.00
Option Value 0.39 0.62 0.98
Exercise? no no yes

Project Value 9.81 11.44

Exercise (B RMB) -4.79 -3.16
Deferal 0.00 0.00
Option Value 0.00 0.00
Exercise? no no

Project Value 8.41

Exercise (B RMB) -6.19
Deferal 0.00
Option Value 0.00
Exercise? no

Figure 5-13 Binomial Tree for ROA with NPV as underlying

3, because the underlying is unobservable, so using the ROA with NPV as underlying will
not render a clear-cut strategy like the ROA with electricity price as underlying or decision
tree analysis do.

5.6. Decision Tree Analysis

Another way to obtain the value of the Project A is through decision tree analysis. The
analysis assumes as before that there is a decision every three years. For any
considered year, that is, years 0, 3, 6, and 9, there are three possible price schemes,
 114

high, medium, and low. For any year there is a decision, the structure of the tree is as
Figure 5-14:

Figure 5-14 Basic Structure of the Decision Tree

“Go” in the decision tree means to invest.

5.6.1. Scenarios on Condition Branches:

The probability and price and determined by a Binomial tree with a drift rate of –0.33%
per year and an electricity price volatility of 6.96% per year. The decision tree has three
branches for each node, corresponding price high, medium and low, while the binomial
tree extends two branches per period. To deal with this mismatch, the binomial tree used
to calculate parameters for the decision tree has a length of period of 1.5 year. This
binomial tree is different from the binomial trees in Figure 5-11 or Figure 5-13. The data
for years 0, 3, 6, and 9 are kept and the data for years 1.5, 4.5, and 7.5 are discarded.
 115

Therefore, evolution from year 0 to year 3 has three paths (branches) and suits the use of
the decision tree. For example, see Table 5-3:

Table 5-3 Using Binomial Tree to Get Probability for Decision Tree

0 year Prob. 1.5 years Prob. 3 years Prob.

0.25 1 0.27 0.449 0.30 0.202
0.23 0.551 0.25 0.495
0.21 0.303

For the price move form year 0, or present, to year 1.5, there are two possibilities, either
moves up by a factor of u, or moves down by a factor of d. where

u = e 1.5 ⋅σ = 1.089
1
d = = 0.919
u

The probability of moving up is

e −0.005 − d
pu =
u−d

-0.005 is the drift rate of electricity price per 1.5 year. So:

pu = 44.9%
p d = 1 − pu = 55.1%

Refer to Table 5-3, at year 3, the high branch has a price of u 2 ⋅ 0.25 = 0.30 with a

probability of pu2 = 20.2% ; the medium branch has a price of u ⋅ d ⋅ 0.25 = 0.25 with a
 116

probability of 2 p u p d = 49.5% ; the low branch has a price of d 2 ⋅ 0.25 = 0.21 with a

probability of p d2 = 30.3% .

Applying the same techniques, all the price scenarios in this tree and the associated
probability are obtained.

5.6.2. Payoffs

In this analysis, the Project A cash flow simulation model is used to get all the payoffs.
For each of the price scenario, the price is input into the model as the initial price at year
0 to get the expected NPV of the project under such a price scheme. For example, at
year 3, the high price branch is price of 0.30 RMB with a possibility of 20.2%, the price of
0.30 RMB is set as the initial price, and the expect NPV for this case is 4.29 B RMB. It
should be discounted to present value and use the present value as the payoff. For the
payoffs at year n, it needs to be discounted for n years. The choice of discount rate
problem presents again. It is hard to find solid grounds for any discount rate used. Here,
the discount rate of 8% is used, as in the traditional NPV analysis.

With the structure of the tree, scenarios of price and probability, and payoffs, the software
of TreeAge is used to build the tree as Figure 5-15 .

5.6.3. Result

Roll back the decision tree to get the value of the decision tree is 1.20 B RMB.
Compared the expected NPV with simulation (actually it is a case without flexibility) of
0.98 B RMB, the value of the deferral option is the difference between the value from
decision tree and value from NPV with simulation, or 0.22 B RMB.
 117

Final Decision

Figure 5-15 Decision Tree Analysis

 118

5.6.4. Shortcomings of Such an Analysis

1. The decision tree can be very bushy, for the tree we built that considering 9 years 4
periods, it has totally ((2 × 3 + 1) × 3 + 1) × 3 + 1 = 67 branches. If 12 years 5 periods are

analyzed, 67 × 3 + 1 = 202 branches are needed.

A tree of 202 branches needs a big effort to build and needs to be very careful otherwise
the establishment of such a tree will involve a big cost of debugging. One of the
problems with such a tree is that people do not have other mechanism to crosscheck if
the tree is built correctly; because there is no means to cross-check, people has to
examine each node and branch carefully. It is a tedious and tiring task, very easy to
make mistakes. Because the number of branches of a decision tree is increasing
exponentially, A bushy tree is one of the most salient problems of the decision tree
analysis method.

A binomial tree can analyze the same problem of 9 years 4 periods with 4 braches, or 12
years 5 periods with a tree of 5 braches. Binomial tree is a very smart tool to make many
scenarios recombined.

2. Another problem facing this decision tree analysis methodology is the problem of
choice of discount rate. And explained, the choice of 8% as discount rate here is not
easy to justify objectively.
3. The tree is so bushy that any sensitivity analysis is not simple, for example, if people
are going to get the sensitivity analysis for the discount rate. A big portion of the tree
needs to be redone, and it is not a trivial task. Even if with a tool like TreeAge that
people can link the tree to a spreadsheet, it is still a significant task to build this linkage
and do sensitivity analysis.
 119

5.7. Summary of Results

With all the above analyses using different methods, the following results for the Project
A are obtained as Table 5-4:

Table 5-4 Summary of Results

in Billion RMB, $1 = 8 RMB

ROA
NPV IRR Electricity
NPV IRR DTA NPV as
w/Simulation w/Simulation Price as
Underlying
Underlying
Value 7.02* 0.98* 10.05% 8.80% 1.20* 2.83 1.25
Option Value N/A N/A N/A N/A 0.22 1.85 0.27
Decision Go Go Go Go Wait Wait Wait
* the discount rate is chosen to be 8%

Using traditional NPV method and NPV with simulation, the project value without
flexibility is studied. As analyzed above, the NPV with simulation gives a more accurate
answer for the value of the project. Then decision tree analysis and real options analysis
(using NPV and electricity price as underlying, respectively) are used to obtain the value
of the project with flexibility (the deferral options). The difference between the project
value with flexibility and the value without the project flexibility is the deferral options
value. The base case for the value of the project without flexibility is the NPV with
simulation or 0.98 B RMB.

Among the three methodology valuating the flexibility: DTA result has the problem of
using an unjustified number of 8% as the discount rate, so it is less accurate than the real
options analysis. Between the two Real options models used, the ROA with NPV as
underlying has a very vague exercise condition; actually the options are not exercised in
 120

the way it describes because the NPV is not readily observable. Moreover, it
systematically overvalues the options because the risk-free discount rate is not enough
for an additional risk component because, at any time, the underlying NPV is not a
determined value.

The conclusion is that the ROA with electricity pricing as underlying gives the most
accurate valuation as 0.27 B RMB, and it gives a strategy to develop the project, a
contingency plan regarding what to do under different scenario. Given the high volatility
of price due to the pending energy market reforms, the best strategy is to wait until the
price of electricity is mostly resolved. If the price is high then beginning building the plant,
if it is low then still wait.
 121

Chapter 6 Policy Implications – Options

Thinking in Practice

Before studying how to apply options thinking in practice, we have to first make sure that
policy makers would use it. This means that options analysis has to make a substantial
improvement over previous analysis tools. Two substantial improvements must be
presented to and understood by policy makers before the efforts to apply options thinking.
The two improvements from options thinking are value of options (better mean) and/or
change of distribution (desirable distribution)

Better mean. It is often the case1, once options are imbedded in the systems planning
process, there is an improvement in the value of the project. The value is substantial
when uncertainty is higher, which is measured by a higher volatility. For the case of
Yalongjiang River studied in this thesis, when the volatility is 30% and drift rate is –0.33%
per year, the value of a deferral option is 17.74% of the total investment, or 2.59 Billion
RMB (refer to Table 5-2). According to Alvarado and Rajaraman (2000), the US electricity
price volatility has been 21.8%. It is reasonable that volatility in China is higher than that
in the US, given many uncertainties in the current China energy market, such as the
pending energy market reform. 30% is a more reasonable guess of volatility than 6.96%
provided by the experts in the China Development Bank. The value of option cannot be
neglected in the planning of the projects on Yalongjiang River.

Desirable distribution. With options on hand, the distribution of the value of a project is
changed. Some extreme values are avoided for the downside, and/or some more
favorable upside potential is obtained. Smaller downside risk is very important for many

1
Under circumstances, people may use an option decreasing the mean actually, e.g. insurance.
See following paragraph.
 122

policy makers. The more risk averse a policy maker, the more important a smaller
downside risk is. For some cases, even if the mean of the value of a project is smaller
than the another project, the project may be still be preferred by policy makers because
of a smaller downside risk. An options analysis will provide a plan with a smaller
downside and/or bigger upside.

Forecast: NPV without option

2,000 Trials Frequency Chart 29 Outliers
.03 61

.02 45.7

.01 30.

.00 15.2

.00 0
RMB –0.1B
Certainty is 52.35% from -Infinity to RMB-0.1B

Figure 6-1 Distribution of NPV without options (Drift = -0.33%/yr, σ = 6.96%)

Forecast: NPV with Option

2,000 Trials Frequency Chart 49 Outliers
.43 859

.32 644.

.21 429.

.10 214.

.00 0
RMB –0.1B
Certainty is 23.75% from -Infinity to RMB –0.1B

Figure 6-2 Distribution of NPV with options (Drift = -0.33%/yr, σ = 6.96%)

For example, Figure 6-1 shows the distribution of NPV of Project A without a deferral
option, the probability that the project will have an NPV less than 0.1B RMB is 52.35%.
 123

Figure 6-2 shows the distribution of NPV of Project A with the deferral option. It has a big
spike because many values concentrate around 0.1 B RMB for the case that the project
is abandoned at year 3. When comparing Figure 6-1 and Figure 6-2, notice the change
of scale for the y-axis. With the option, the probability that the project will have a NPV
less than –0.1B RMB is 23.75%. Much smaller. Although for the case studied here with
–0.33% drift rate and 6.96% volatility, the value of the option is only 0.27B RMB, or
1.85% of the total investment, the deferral option may help the project decrease the
chance of losing more than 0.1B RMB by 28.6%. This significant risk avoidance may
justify the use of the deferral option though the option value is negligible.

After convincing policy makers of the significance of options thinking, some application
tactics is very important in making real impact of the options thinking. Meanwhile, we
should be aware that options analysis is not a panacea, sometimes it works, sometime it
does not. So it is important to understand the uncertainty reality and choice of
appropriate methodology.

6.1. Making Real Impact of the Options Thinking

The author of this thesis spent 4 months in the China Development Bank in Beijing to
help the bank study the real options method and implement the real options thinking in
their project evaluation process. The China Development Bank is the major investor of
China public projects in energy, transportation, and other public areas where other
investors are unwilling or unable to invest.

To use options thinking in a developing country like China, there are more difficulties than
using the methodology in a developed country like the US. In the US, there are still many
problems need to be solved before real options method can be applied maturely and
widely. The major difficulties of the real options method in developing countries are how
 124

to make decision-makers understand the method and obtain the quality data necessary
for options valuation models.

The options methodology develops a proactive and analytical view of uncertainty and
flexibility. Despite the difficulties mentioned above, it has a good possibility to make real
impact in a developing country such as China. Besides the methodology itself, the
organizational tactics and skills to make people accept new ideas are also among the
keys to make real impact of the method.

This section discusses the two aspects of promoting the real options method in China,
and more generally, in developing countries. The first aspect is the intrinsic difficulties in
promoting real options method in developing countries, and the second is the
organizational wisdom to advocate new thinking.

6.1.1. Intrinsic Difficulties in Promoting Real Options

methodology in Developing Countries

First of all, it should be stressed the real options methods have big room to grow in
developing countries, despite the difficulties discussed in this section. In other words,
these difficulties do not make the method inapplicable in developing countries.

Understanding the method

The options theories and models always seem arcane to people, the partial differential
equations… the dynamic processes… people usually think options are the work of rocket
scientists.

In developing countries, what makes things even worse is that financial options are not
traded in local financial markets. This is because the trade of options is very easy to
 125

foster fraudulence in a weak legal system. For example, China once allowed the trade of
Chinese treasury note options. Soon after the beginning of the trade, a scandal broke
out and a broker’s firm lost more RMB 10 billion overnight. And after that, financial
options have not been allowed to trade on China financial markets. So Chinese people
do not know what options are intuitively like people in developed countries.

If a person does not establish an understanding of financial options, he/she will find it
very hard to develop the idea of what the real options are and to have confidence in the
method. For example, in the China Development Bank where I worked, a manager heard
that the real options method overvalues projects, then he had deep misunderstanding of
the method because the real options method was a black box for him. He could not
appreciate that the real option valuation will overvalue a project only when it is misused.

After all, a manager is not going use a method if he/she does not understand it.

Data

Options analysis needs a lot of historical data to do objective analysis. In developed

countries, the abundant historical data on financial market provides the power of options
analysis - there is little subjective element in the analysis, and the magic of market tells
all. However, in developing countries, there is no complete financial market, and
consequently, the data is incomplete. Even with some data, the financial market is
decided by too much government interference so the information is highly distorted.
China is such an example. So even if there is data, people must be very cautious when
using it. Also, because China has been undergoing tremendous reform for a long time,
there has been several major system changes in the past 10 years. The state changes
imply that the available data do not reflect the current situation.
 126

A way to circumvent the data problem is to use simulation to obtain important parameters
such as the volatility. In this way, the model risk is huge because the model includes of
subjective assumptions of the model-makers.

Although the difficulties in helping managers understand the methodology and the
availability of data problems, the real options method will be able to spread fast in
developing countries because of its insights into uncertainty and flexibility. The thinking
is invaluable, nevertheless.

6.1.2. My Strategy in making real impact of options thinking “in”

hydropower projects on Yalongjiang River

The first step to think about making real impact with options thinking is to understand the
institutions that have the most influence over the implementation of real options “in”
hydropower projects in Yalongjiang River Basin. To the best of my knowledge, the most
influential institutions are:

• Ertan Hydroelectric Development Corporation (EHDC)

• China Development Bank (CDB)
• Sichuan Hydrology and Hydropower Institute (SHHI)
• Sichuan Trust and Investment Corporation (STIC)
• Ministry of Water Resources (MWR)
• China State Power (CSP)

Ertan Hydroelectric Development Corporation (EHDC)

This corporation is now in charge of the development of Yalongjiang River Basin. It is the
entity that developed the first hydropower project, Ertan Hydropower Station, on
 127

Yalongjiang River. EHDC is eagerly pushing further projects on the river. It is natural for
any institution to seek more projects to enhance its control of resources and influence.
Moreover, the first project of Ertan has been losing money, and EDHC is in serious debt.
They would like to have further projects to alleviate their hard times.

There is competition between Yalongjiang River and Daduhe River, another tributary of
Yangtze River in Sichuan that also has good natural conditions to develop hydro projects.
Given the limited funding and other resources, a new project approved by the state
government on Daduhe River will greatly delay the development of Yalongjiang River
Basin.

EHDC is the most active player in pushing hydropower projects on the Yalongjiang River.
They would accept anything that is beneficial to the success of new projects on
Yalongjiang, in finance or in engineering. The idea of incorporating of flexibility into the
design of projects is definitely to their benefit.

China Development Bank (CDB)

The Bank is a major funding source for new projects in the Yalongjiang River Basin.
Since EHDC is already deep in the red, they have to get substantial support of loans from
CDB.

In the China loan market, CDB faces serious competition from other banks in China.
Currently, CDB is having difficulties in finding good customers to lend money. They have
money but lack good projects. Also, since the previous centralized system of energy and
hydrology in China, CDB and EHDC have very close ties. Many influential managers in
CDB have worked in EHDC before. A big percentage of senior managers in EDHC are
from CDB. And a number of managers in CDB are looking forward to the opportunity to
work as senior management in EHDC one day. They have personal interest in pushing
the projects. So a group of managers in CDB are enthusiastically pushing the projects.
 128

However, as they are mainly concerned about finance, engineering flexibility is apart from
their interest.

Given the current financial situation of EDHC and the astronomical amount of investment
of hydro projects, the risk is too big for CDB. So CDB is hesitating to invest more in
Yalongjiang River Basin. They are hoping other financial institutions will share the risk.
One of the possible institutions to share the risk with CDB is the Sichuan Trust and
Investment Corporation (STIC).

Sichuan Trust and Investment Corporation (STIC)

STIC is a local investment company that participated in the financing of the first dam on
Yalongjiang River, Ertan. Both CDB and EHDC want STIC to provide part of the funding
for the new projects. However, STIC are not in a good financial situation partly because
of the financial failure of Ertan for several years. STIC is not big enough as CDB to suffer
the loss. They are not enthusiastic on the projects, let alone interested in the engineering
aspect.

Sichuan Hydrology and Hydropower Institute (SHHI)

This is a research institute. They have been responsible for the hydrology research and
engineering design for the Yalongjiang River Basin development. They are genuinely
interested in any new methodologies. Furthermore, because it is likely that they have to
bid for projects in the future given the rapid reform in China (there are other research
institutions in Sichuan and China and they can be tough competitors). They are more
than willing to incorporate new methodology in their design responsively if they are sure it
works.
 129

Ministry of Water Resources (MWR)

This Ministry is responsible for controlling and coordinating all water resources
development in China. They are interested in more strategic issues, though they would
like to see the technological advances in engineering.

China State Power (CSP)

CSP has been the monopolist of China energy market. Now it is dissolving into several
separate companies according to the plan of China energy market reform. CSP is
concerned about the problem of overcapacity in Sichuan because the grid to transport
electricity from Sichuan to Eastern provinces has not been built yet. Whether to build the
grid itself is a problem given the huge costs to do so. CSP is not particularly interested in
the design issues of hydropower projects.

Target institutions to “sell” real options methodology

The two institutions that might be most willing to accept the results of an analysis of real
options “in” a project are:

• Ertan Hydroelectric Development Corporation (EHDC)

• Sichuan Hydrology and Hydropower Institute (SHHI)

Ertan Hydroelectric Development Corporation (EHDC)

As analyzed above, EHDC is the most active player in pushing hydropower projects on
the Yalongjiang River. They would accept anything that is beneficial to the success of
new projects on Yalongjiang River.
 130

The engineering department might be the most willing part of EHDC to accept the results
of an analysis of real options “in” a project. Their perspective is pretty short-term. They
want to begin construction of whatever project as soon as possible, and the long-term
sequencing and scheduling is not their particular interest.

They may be interested in the real options analysis on tunnel design and selection of
installed capacity. The previous design principle is to minimize total cost. With real
options analysis, we should replace the decision principle with the one of maximizing net
revenue given the uncertainty of electricity price.

To deal with an options analysis for tunnel design and selection of installed capacity, it
might be hard to analyze the China electricity market with the poor data available and
huge uncertainties, while the other part of the analysis can be done relatively easy if
there is a cost model. We also need hydrologic information. Then we can do simulation
to get the expected value of net revenue.

In order to successfully present the method to EHDC building on their existing

understanding, I would stress the inadequacy of previous decision principle of minimizing
the total cost that misses market uncertainty (implicitly assuming that the plant can
produce full capacity all the time). For the difficulty of a credible model for China energy
market, I would ask EHDC to develop a probabilistic model on China energy market and
we use their model to do the rest of analysis.

Sichuan Hydrology and Hydropower Institute (SHHI)

As a research institution and eager to keep their current design projects, SHHI may be
willing to learn and would incorporate new methodology in their design responsively. And
they have a longer and strategic view of the river basin development.

The department that might be most willing to accept the results of an analysis of a real
option “in” a project is the strategic planning department. They are interested in the
 131

strategic plan of developing the river basin, including selecting project location and size,
sequencing, and scheduling.

It should be possible to develop a deeper understanding than what current methods can
offer on selecting project location/size and sequencing, using the current available data,
the knowledge on river basin development optimization, and methodology of real options.

To deal with such a real options analysis, I do not see insurmountable difficulties for now.
Although the model of instantaneous behavior for China energy market is hard to develop,
the long-term equilibrium that the energy with the lowest cost will be consumed first
provides a good basis for solution of the problem.

In order to successfully present the method to SHHI building on their existing

understanding, I would show the optimization model first and then add the real options
part as an modification. Chinese people are unfamiliar with options concept and need
time to understand. By showing real options part as a modification to current methods of
optimization, it might be easier to present the real options methodology. Meanwhile, I
would stress the inadequacy of deterministic analysis all the time.

My strategy

In terms of making the most effective contribution to the efficient implementation of

options thinking “in” hydropower projects, I would target my work toward the strategic
planning section in the Sichuan Hydrology and Hydropower Institution, aiming to
influence the engineering systems planning part of the process of getting real options in
hydropower projects.

With the development of decision science, finance science, and computer technology, it
is now possible to adequately address the issue of building flexibility into systems design.
The water resource systems design and planning has been developed to maturity, and
 132

finance science has developed a systematic theory on uncertainty with options theory as
the most significant contribution. It is now time to explore design large systems with full
awareness of uncertainty.

As a student in Engineering Systems Division (ESD) at MIT, I am more interested in

systems level analysis and planning rather than more detailed design of a specific system.
ESD people has the special expertise to be distinguished form finance people who do not
establish engineering models, while develop a general model for engineers with more
technical knowledge to develop the model with more details.

My background and gained knowledge is adequate in the areas of options and

optimization (both at the application level, I would not pretend to have strong theoretical
capability). Detailed engineering design is not my strength.

The most possible research direction to have the most influence over real options “in”
project is to work on the projects selection, sequencing, and scheduling using
optimization techniques and real options, and present the work to Sichuan Hydrology and
Hydropower Institute.

The general contribution of the work would be providing a general model that water
resources engineers can build on with more details and other large engineering systems
can also use the model for their specific systems. See Figure 6-3.

River Basin Satellite

...
Development Systems
General Our Position

More Details

Figure 6-3 Contribution of my work

 133

6.1.3. Organizational Wisdom to Advocate New Thinking

In this section, the author’s personal experience in specific settings is introduced.

Readers please judge if it works in the specific organization and specific country you are
working in.

During the work in the China Development Bank, I realized the tremendous difficulties in
promoting new thinking, especially in some big institutions that are featured of rigidity of
bureaucracy and highly risk-aversion. After working in the bank for months, I figured out
some tactics to quicken the process of adoption of new ideas. Although any good ideas
will be used in the end, the process is very slow and painful. For example, it takes 20
years for people to use the NPV method widely. However, a smart way will make the
process faster.

People don’t like changes. They would like to follow the old way that they are familiar with.
Because of the efforts and pains to, unless people do not face threats or other pressures,
they won’t take the efforts to change. Change is difficult and we must be patient and
know how to make things happen tactically.

After working in the China Development Bank. I understand the following 4 elements are
the keys for the success of letting people accept the options thinking. They are

- support from the top management;

- identify the needs of the organization;
- from simple to complex;
- perfect communication with grass-root people in the organization.

The support from the top management is key to success. Basically, in most big
organizations, one of the most important motivations of people is promotion. If the top
management would like to promote something, people will have enough motivation to
take the endeavor to study the new ideas. When I did the real options study and
 134

promotion in the China Development Bank, one of the top managers gave me full support
on the project.

Another important tactic is to identify the needs of the organization, and firstly show them
that the method can be used to deal with the questions that they are most interested in.
In the bank, I understand that a bank does not care too much about the value of a project,
because what they can get at most is the principle plus interest. If the project earns more,
or has more upside, it has nothing to do with the bank. On the other hand, the bank
bears the risk of losing the big chunk of the principal (while the interest they can earn is
only a small part compared the principal). So they are more interested in the risk
management rather than valuation. So when I presented the real options to the Bank
during the first several weeks, I did not mention many classical benefits of the options
valuation, such as valuation of the flexibility. Instead, I used simple simulation showing
the idea of right not obligation of options can be designed into a contract and reduce the
default risk. By focusing on the needs of the bank rather than the traditional stressed
benefits of real options, people in the Bank are getting interested in the method and
understand the method gradually.

The real options method is not a simple method, though it looks transparent to experts
who are studying the method everyday. It does take a while for those have never heard
about it to understand the method. So it is important to show the thinking and method
from simple to complex step by step. In the bank, when I showed [people there how to
valuate the options, firstly I did not show them the profound options valuation formula or
even binomial tree. I just showed them the idea of cut probability density function to
valuate an options which is much more intuitive with the presentation of a software like
Crystal ball.

A perfect communication with the bank people is the key to success, especially the
grass-root people. Their trust in me is one of the necessary conditions to success. In the
bank, I had to do a successful case to show the method workable. Though I know the
real options method, I do not know the intricacy of project evaluation. For example, I did
 135

a project valuation on a hydropower project study, but I am not an expert on the China
hydropower industry. So my real options analysis might have made no sense to the
experts in the bank who have studied the industry for decades. So I have to make sure
that my reports were collaborated with experts in the bank. In the bank, I made two very
good friends, one is an expert in China energy industry and another is the people who the
bank designated to cooperate with me. With friendly and perfect communication with
these two grass-root people, I was able to understand the bank faster and avoid many
stupid mistakes.

Please see a journal I took about for my work in the Bank about how I analyzed the
situation to do a successful job in the Bank.

Despite the support from the top management, one of the key of a successful consulting
project, there are two major difficulties to persuade people to try the method: firstly, there
are no options traded in China financial market, the concept of options seems too arcade
for most Chinese people; secondly, the China Development Bank (CDB) is a typical
state-owned enterprise, slow in accepting new things. How to influence CDB to try and
use real options?

I applied my process to make decision to analyze the situation and decide the way to
make impact in CDB. The process is as the following figure:

Gathering Finding possible Choosing

Implementation
Information solution alternatives the solution

Feedback
 136

Gathering Information:
On one hand, I understood the key trait of the Bank is risk aversion; their primary interest
is the default possibility rather than the profitability of the debtor. Options are highly risky
but yielding a fatty premium, so I cannot talk about the options in a traditional sense and
need to customize the useful part of the theory and methodology for the bank. On the
other hand, I tried to understand what every people I worked with in the bank needs and
their relationships with each other. Though relationship is like a maze in a China state-
owned company, so complex that I cannot understand in months, I tried my best to be
careful. Also I identified two people in the Bank were really highly motivated for the
project. One is a chief of a section who was eager to show himself before the top
manager who was pushing the project forward, another is a young man just graduated
from graduate school whose thesis is on real options.

Possible solution alternatives:

I identified several possible solution alternatives:
1, focus on communication to management to overcome the obstacles in the learning
process of the CDB people;
2, gradually involve the CDB people in the method. More specifically, by adapting the real
options method to work for the issues they are caring about most, to show the power of
the method and let them involved in the development of the application of the method to
the bank;
3, let the top management push the project;
4, win over the young people in the bank first, let them spread the method, and finally
make the method a common practice.

Choose the alternative:

Alternative 1, it was not practical in CDB in a short period I worked there, given the facts
that there are no financial options in China, and the theory is not straightforward.
Alternative 2, it was a good approach. As a first step, I could deal with a basic problem of
quantifying default risk (the subject that the Bank are most interested in) in a new way.
 137

Though it is not a traditional real options method meant to be, I gradually unfold the real
options when working with CDB people on quantifying default risks.
Alternative 3, it was not so good at the time I was working in the CDB. I understood the
attention that the top manager could put on the project at the time is small given that time
one other Vice Governor was sick.
Alternative 4, it was a good idea because young people are more willing to learn new
things and faster in learning. If they use the method, at least partly in their work by
themselves, the method will be spread out finally. Although it is a little slow, it is one of
the best ways to influence the Bank in the long run.

Implementation:

I worked together with the 2 CDB people who are highly motivated for the project in the
bank. By attacking the problem of quantifying the default risk, I was gradually presenting
the real options concept. I managed to persuade the division head to accept the real
options method.

I had 3 lectures for around 100 young people in the Bank, and arouse the interest of them
in the method. A number of them contacted me later, and I guided them to learn about
the approach.

Though it is a challenge to make real impact in real business, I have been pretty
successful in presenting the method to CDB. I finished two case reports and one general
report for the Bank, and they are satisfied with the preliminary results and inviting me to
come back again in February 2003 to promote the method further.
 138

6.2. Uncertain Reality and Appropriate Method

The nature is inherently unknown. People can handle the situation with known risks, but
people just cannot deal with the situations where the uncertain factor is unknown, or the
uncertain factor is known but the distribution of the uncertain factor is unknown.
Anticipatable risk should be managed, but other system risks have to be left there. When
people have to enter into the domains that are ultimately unknown, they need to invest in
information to the best capacity possible. If you have done what you can do and still fail
in the end, nothing to blame.

Several common quantitative tools for policy evaluation are listed below:

- DCF1
- Decision tree analysis
- Real Options

A useful classification of uncertainty helps people to understand the environment better,

and identify the best quantitative tools for policy evaluation to deal with the specific
uncertainty. And most importantly, people need to understand that the right decision and
method do not necessarily lead to a happy outcome.

Suppose the outcome, denoted Π, is a function of N influence parameters:

Π = Π ( w1 , w2 ,..., wN )

Usually, the project payoff Π cannot be predicted with certainty.

1
Discounted cash flow method.
 139

Complexity: Many of the parameters are not random, and they could be under the
influence of the decision-makers – for example, engineering designs. Interdependencies
among influence parameters may make the optimization a significant task. Thus,
complexity is the centerpiece of much of deterministic analysis tools, such as the NPV
method and Linear Programming. Complexity increases with the number of
nonseparable variables in Π. There is no uncertainty involved in complexity, actually.
Another example of complexity is scheduling classes in a big university.

Risk: Some influence parameters are not deterministic but dynamic. There is payoff
variation of Π associated with those parameters. There are two kinds of situation: the
distributions of those influence parameters are known or unknown. If the distribution is
known, and the influence parameter is market-traded assets, the standard real options
approaches apply, such as the Black-Scholes formula, binomial tree, and simulation
methods. If the distribution is known, and the influence parameters are not marketed
traded, people may use the decision tree analysis and binomial tree to design contingent
plan and value the project. However, if the influence parameter is known while the
distribution is unknown, a decision tree analysis with the probability as an unknown
parameter is an appropriate method. Sensitivity analysis with the unknown parameter
can be done to get important insights into the problem.

Ambiguity: The decision-makers may not be aware of some the influence parameters at
all. The decision-makers explicitly ignore the payoff impact of those influence parameters,
but implicitly takes “default” value for those influence parameters. The past is
extrapolated subject to changes in known parameters in the category of complexity and
risk. For any real decision, it is very hard to avoid ambiguity, or in other words, not to
miss any influence diameter. This is the key reason why forecast is almost certainly to be
wrong. For the case in an ambiguous environment, the best method is to invest in
information carefully, learning by trial and error. It is like entering a room without light to
find a way out,. After bumping onto the table, the pain tells you that there is a table… In
such situation, people should avoid betting too much and do not put all the eggs in a
single basket.
 140

The above analysis of uncertainty types can be summarized in Table 6-1.

Table 6-1 Uncertain Reality and Appropriate Method

Risk
Distribution Known
Complexity Ambiguity
underlyingmarket underlying not market Distribution Unknown
traded traded
Financial Options
DTA DTA (with probability as Trial and error
Deterministic Valuation valuation formula
Methods an unknown to do
(NPV, LP, etc) Binomial Tree Simulation
sensitivity analysis)
Simulation
W hether China energy
Energy Project valuation River basin development New technology
Scheduling classes in a market will be market
Examples with underlying of energy with waterflow distribution aimed at
university allocation or government
price known unkonwn market
allocation regime

With the understanding of uncertainty types, policy-makers can

- decide when different analysis techniques are most suitable, when to use real
options method, when to apply decision tree analysis, and when the mathematical
programming method is the best.;
- understand that there are cases where the analysis techniques do not help much
if a lot ambiguity exists, and the best strategy is to carefully learn by trial and error,
and do not make life-or-death bet;
- because of the ambiguity is ubiquitous in real life, sometimes the policy is
optimized given the information available. The result may not be happy, however.
In such case, policy-makers should not blame the methodology and gather more
lessons for future decisions.
 141

Chapter 7 Conclusions

7.1. ROA with electricity pricing as underlying is

appropriate for Project A

The recommended methodology for the Project A is a ROA with electricity pricing as
underlying. Compared to Ramirez’s study on the Columbian Bogota water supply
expansion plan [Ramirez, 2002] where she concluded that real options method is
imprecise and complicates the process of identifying an optimal strategy, the Project A
case is different from that of Ramirez’s case because of the following observations:

1. The subject of deficit cost that Ramirez was studying is not a market observable
monetary value. She had to change a lot of non-monetary values into a monetary value.
It is not only hard to justify the monetary value assigned convincingly, but also hard to
find data. Moreover, the deficit cost does not look like following a geometric Brownian
motion.

2. She was using the ROA with NPV as underlying, but what this thesis suggests to use
is the ROA with electricity pricing as underlying. The differences between the two
methods please see Table 7-1. She did not recommend a the ROA with NPV as
underlying for Bogota water supply expansion plan; though this thesis does not
recommend a ROA with NPV as underlying for Project A either, it recommends a ROA
with electricity pricing as underlying. If people have to choose between decision tree
analysis and ROA with NPV as underlying Project A, the DTA analysis would be
suggested.
 142

Table 7-1 Comparison of the two ROA methods studied in this Thesis

ROA with NPV as underlying ROA with electricity price as

underlying
Intuitive to
Very unintuitive Medium
Managers?
Identifying
No Yes
strategy?
Data? Very hard to justify convincingly Easier to obtain than ROA with NPV
when underlying is not traded on as underlying *
market
Key issue Justify Brownian Motion (Weiner Justify the stochastic process of the
Process) of the underlying** effective market.
* The three ways to get the volatility for the effective market is historical data, Delphi
method, and simulation.
** I believe this is why Ramirez [2002] got a ROA result way different from DTA. The
deficit cost she put into the RO model is not following the Brownian motion model.

Ramirez’s three key arguments for the real options method might not be valid for Project
A, though reasonable in her Bogota case:

• Data not available. On the electricity market, data on electricity price is abundant.
For the deficit cost that Ramirez was studying, the data is harder to obtain.
• Not intuitive for manager. The ROA with NPV as underlying that Ramirez was
using is hard to explain to mangers because the vague and hard-to-define
underlying value of the project. The ROA with electricity pricing as underlying for
Project A is using electricity price as underlying. The event tree built to display
the different scenarios of electricity price is much more transparent to mangers.
• Cannot identify the optimal strategy. The ROA with electricity pricing as
underlying solves this problem well and offers an optimal contingent strategy,
 143

while the ROA with NPV as underlying based on a vague underlying of the value
of the project is hard to provide a clear optimal strategy.

7.2. Comparison of Methods

The methods used in this thesis can be divided into two categories: the first are point
valuation methods, including NPV, NPV with simulation, IRR, and IRR with simulation;
and the second are dynamic analysis methods, including decision tree analysis, ROA
with electricity price as underlying, and ROA with NPV as underlying.

The point valuation methods provide a point measurement of the project such as NPV or
IRR. They are simple yet power tools to decide the acceptance of projects or compare
between projects, but they do not take into account the flexible decisions people can
make after the project begins. In a word, the point valuation methods miss the value of
flexibility, and they cannot provide a contingency strategy for projects. If resources
permit, a dynamic analysis method is preferred to a point valuation method.

This paragraph compares the four point valuation methods. Traditional NPV method is
the simplest model that gives a primitive financial evaluation. NPV with simulation is a
better method than traditional NPV method because 1, E[ f ( X )] ≠ f [ E ( X )] and 2, the
expected value of the inputs for the traditional NPV method is not easy to observe. And
both NPV methods have the problem of choice of discount rate. IRR method and IRR
with simulation method overcome the dilemma of choice of discount rate. The difference
between IRR method and IRR with simulation method is similar to the difference between
NPV method and NPV with simulation method.
 144

This paragraph compares the three dynamic analysis models. Table 7-2 shows the pros
and cons about the decision tree analysis, ROA with NPV as underlying, and ROA with
electricity pricing as underlying is established.

Table 7-2 Comparison between Methods

Decision Tree Analysis ROA with NPV as ROA with electricity price
underlying as underlying
Pros - Can readily identify - Better dealing with - Readily identify optimal
optimal strategy. discount rate problem strategy
- Intuitive to managers by using risk-neutral - Better dealing with
discount rate discount rate problem
by using risk-neutral
discount rate
Cons - Very easy to become - Data is not readily - Need to decide the
extremely bushy available because the stochastic process of
- Problem with choice of underlying value (NPV) the electricity price
discount rate is not traded on market - Data is sometimes not
- Hard to do sensitivity - Need to justify the readily available
analysis assumptions for Options
- Hard to assign valuation, such as
convincing probability if Brownian motion and no
lack of data. arbitrage.
- Can not readily identify
the optimal strategy
- Not intuitive for
managers
 145

7.3. Future Work

As the next step following this thesis, engineering options will be explored in the
development of Yalongjiang River Basin. In the process of applying options thinking in
engineering projects evaluation, 3 key components are identified. They are identifying
options, modeling, and obtaining insights into the project.

7.3.1. Identifying Options

The possible engineering options built into the design include the road building as growth
options and the tunnel design for Project B project as scaling options.

Roads as growth options. For the projects in the Yalongjiang River Basin, people will
have to build roads into the remote sites of dams. These cost a lot. Traditionally, people
decide the design of the dam and the specifications for roads almost the same time;
actually, people can, at the time the roads are finished, decide the height of the dams
given the conditions of the energy market then. In real options literature, it can be
regarded as growth options, or early investments to get information. Incorporating growth
options, important engineering decisions regarding roads and dams are different from
traditional way of designing.

Tunnel design as scaling options. The Project B project would dig tunnel(s) to shortcut a
part of Yalongjiang River to gain significant head. People can build one big tunnel, two
small tunnels, or three smaller tunnels. Traditionally, the decision rule is based on
hydraulics. But if decision-makers take into account the market uncertainty and financing
uncertainty, they may get the additional value of three tunnels over two tunnels or two
tunnels over one tunnel because of the options to postpone building additional tunnels if
the market is unfavorable and the financing is very expensive. This value can be
significant and reverse the decisions based only on hydraulics considerations.
 146

7.3.2. Modeling

The modeling has two parts: traditional models and real options models.

Traditional Models

Although options thinking is added into the framework of systems design, the foundation
of systems design is nevertheless traditional model, such as engineering technical
models and mathematical programming models.

The researcher has all the pieces for the technical cost models because he has built a
financial analysis model for the China Development Bank in summer 2002 that has a very
important part of cost analysis. The mathematical programming screening models in
river basin planning needs more work. The objective function is to maximize the profit
from power sales, because the Yalongjiang River Basin development is mainly for
hydroelectric power, and all the other considerations are secondary.

Real Options Models

The first question for a real options model is what the underlying1 is. Effective market is
introduced as underlying. Effective market is effective demand multiplies price. It is in
terms of money. If the effective demand for the company is bigger than the capacity of
the company, then the company will produce to its maximum; if the effective demand for
the company is smaller than the capacity of the company, the company can only produce
up to the effective demand.

1
In this study, a generalized concept of underlying is used. An underlying is the agent that
determines the value of a project or an investment. An underlying can be assets, but it also can
be other agents such as market size or utility price.
 147

A binomial tree is one of the most easy-to-use options valuation models. It is a scenario
analysis in essence, because a binomial tree can be thought as an event tree that
establishes different scenario on each node. See a binomial tree as Figure 7-1.

pu Muuu
pu Muu pd
pu Mu pd pu Muud
M pu Mud pd
pd Md pd pu Mudd
Mdd pd
Mddd

(where M denotes the effective market value)

Figure 7-1 A binomial Tree

On each node of the binomial tree in Figure 7-1, the project is analyzed based on the
traditional models developed, i.e., the mathematical programming screening models and
technical cost models. Under each scenario specified by the node, the exercise value
given a specific option exercised is obtained by the traditional models. For example, at
a specific node, traditional models calculate the project value if the deferral option is
exercised at this node.

A simulation model of the internal rate of return (IRR) of the project is developed to get
the lifetime risk profiles of the project, more specifically, the standard deviation of the IRR.
The standard deviation of the IRR, or σ, is the measure of the average risk per year of
the project with the design. With this σ, a risk-adjusted discount rate is obtained.

Having the exercise values, risk-adjusted discount rate, the move-up probability (pu), and
move-down probability (pd), the binomial tree is rolled back from the rightmost side to get
the option value.
 148

7.3.3. Obtaining Insights

With the binomial tree established, important strategic and operational insights into the
project are developed. Overall, the thesis helps provide a preliminary methodology to
incorporate options thinking into river basin development and engineering systems
analysis.
 149

Appendix I Ito’s Lemma

Ito’s lemma was discovered by a mathematician, K. Ito, in 1951. Suppose that the value
of a variable x follows an Ito process:

dx = a ( x, t )dt + b( x, t )dz

where dz is a Wiener process. A function G(x,t) follows the process

∂G ∂G 1 ∂ 2 G 2 ∂G
dG = ( a+ + b )dt + bdz
∂x ∂t 2 ∂x 2
∂x

where the dz is the same Wiener process. Thus, G also follows an Ito process.
 150

Appendix II China Steps Up Reform of

Electric Power System

- China will intensify reforms to break up the State-monopoly of the electricity supply
system in order to cut costs and raise efficiency.

The reform will include restructuring State-owned assets, implementing a new pricing
mechanism and bettering the regulation of the electricity industry, Xie Songlin, vice-
president of the State Power Corporation, said Wednesday at the Annual Meeting of the
Asian Development Bank (ADB) in Shanghai.

At present, the State Power Corporation owns 46 percent of the country's total electricity
generation assets, and 90 percent of the total electricity supply assets. The reform will
divide most of the electricity generation sector from the corporation and set up five
independent national power plants, according to Xie.

After the restructuring, a national power network corporation will be established, and five
regional subsidiaries will be affiliated to the national corporation in North, Northeast,
Northwest, East and Central China. A separate power network company for south China
will be set up, said Xie. He added that a more appropriate electric pricing system will
come into being in the foreseeable future.

Since the founding of the State Power Corporation five years ago, the sales volume of
electricity has increased 26 percent, and the net profit has grown 40 percent, according
to Xie. In 2001, the corporation ranked the 77th among the world’s top 500 companies.

(People’s Daily, Thursday, May 09, 2002)

 151

References

• Aberdein, D. A. (1994) “Incorporating risk into power station investment

decisions in South Africa,” Master of Science Thesis, Sloan School of
Management, MIT, Cambridge, MA.
• Alvarado, F. L. and Rajaraman R. (2000) “Understanding price volatility in
rd
electricity markets,” proceedings of the 33 Hawaii International Conference on
System Sciences, Hawaii.
• Amram, M. and Kulatilaka, N. (1999) Real Options - Managing Strategic
Investment in an Uncertain World, Harvard Business School Press, Boston, MA.
• Baldwin C.Y. and Clark K.B. (2000) Design Rules, Vol. 1: The Power of
Modularity, MIT Press, Cambridge, MA.
• Baldwin C.Y. and Clark K.B. (2002) “Institutional Forms, Part 1: The Technology
of Design and its Problems,” Real Options Symposium, University of Maryland,
College Park, MD.
• Baldwin C.Y. and Clark K.B. (2002). “The Fundamental Theorem of Design
Economics,” Real Options Symposium, University of Maryland, College Park,
MD.
• Baldwin, C.Y. and Clark, K.B. (2001) “Modularity after the Crash,” HBS working
paper, May.
• Baxter, M and Rennie, A. (1996) Financial Calculus - An Introduction to Derivative
Pricing, Cambridge University Press, Cambridge, UK.
• Bertsimas, D. and Tsitsiklis, J.N. (1997) Introduction to Linear Programming,
Athena Scientific, Belmont, MA.
• Bjerksund, P. and Ekern, S. (1990) “Managing Investment Opportunities Under
Price Uncertainty: from ‘Last Chance’ to ‘Wait and See’ Strategies,” Financial
Management, Autumn, pp. 65 - 83.
• Boer, F. P. (2002) The Real Options Solution – Finding Total Value in a High-Risk
World, John Wiley & Sons, New York, NY.
 152

• Borison, A. (2003) “Real Options Analysis: Where are the Emperor’s Clothes?”
7th Annual Conference on Real Options, Washington DC.
• Brealey, B. A. (1983) An Introduction to Risk and Return from Common
Stocks, The MIT Press, Cambridge, MA.
• Brennan, M. and Schwartz, E. (1985) “A New Approach to Evaluating Natural
Resource Investments,” Midland Corporate Finance Journal, Spring, pp. 37 - 47.
• Brach, M.A. (2003) Real Options in Practice, John Wiley & Sons, Hoboken, NJ.
• Cherian, J.A. Patel, J., and Khripko, I. (2000) “Optimal Extraction of
Nonrenewable Resources When Costs Cumulate,” in Project Flexibility, Agency,
and Competition (2000) pp. 224 - 253, edited by Brennan, M.J. and Trigeorgis, L.,
Oxford University Press, New York, NY.
• Child, P.D., Riddiough, T.J., and Triantis, A.J. (1996) “Mixed Use and the
Redevelopment Option,” Real Estate Economics, 24, 3, pp. 317 - 339.
• China Development Bank (2002) “Sichuan Yalongjiang River Hydropower
Projects Study,” Beijing, China.
• China State Power (2001) Guideline for Financial Evaluation, Beijing, China.
• Clarke, C. (2000) “Cross-Check Survey,” World Commission on Dams Work
Program Component, November. http://www.dams.org/kbase/survey/.
• Cohon, J.L. and Marks, D.H. (1973) “Chapter 21: Multiobjective Analysis in Water
Resource Planning,” adapted from Water Resources Research, 9, 4, August, pp.
333 – 340. in Systems Planning and Design (1974) edited by de Neufville, R. and
Marks, D.H. Prentice Hall, Englewood Cliffs, NJ.
• Cortazar, G. and Casassus, J. (1997) “A Compound Option Model for Evaluating
Multistage Natural Resource Investment,” in Project Flexibility, Agency, and
Competition (2000) pp. 205 - 223, edited by Brennan, M.J. and Trigeorgis, L.,
Oxford University Press, New York, NY.
• Cooper, R.G., Edgett, S.J., and Kleinschmidt, E.J. (1997) “Portfolio Management
in New Product Development: Lessons from Leading Firms,” Working Paper, Feb,
SOB, McMaster University, Ontario, Canada.
 153

• Copeland, T.E. and Antikarov, V. (2001) Real Options - A Practitioner's

Guide, TEXERE, New York, NY.
• Copeland T. E. and Keenan, P. T. (1998a) “How much is flexibility worth?” The
McKinsey Quarterly, 2, pp. 38 - 49.
• Copeland, T.E. and Keenan P.T. (1998b) “Making Real Options Real,” The
McKinsey Quarterly, 3, pp. 128 - 141.
• Cox, J., Ross, S., and Rubinstein, M. (1979) “Option Pricing: A Simplified
Approach,” Journal of Financial Economics, 7, pp. 263 - 384.
• de Neufville, R. (2002) Class notes for Engineering Systems Analysis for Design,
MIT engineering school-wide elective, Cambridge, MA.
• de Neufville, R. (1990) Applied Systems Analysis, McGraw-Hill, New York, NY.
• Dias, M. (2002) “Overview of Real Options in Petroleum,” Real Options Seminar
Series, Spring, Sloan School of Management, MIT, Cambridge, MA.
• Dias, M. “Real Options in Petroleum Investments: Uncertainty, Irreversibility,
Timing.” at http://www.puc-rio.br/marco.ind/.
• Dixit, A.K. and Pindyck, R.S. (1994) Investment under Uncertainty, Princeton
University Press, Princeton, NJ.
• Dixit, A.K. and Pindyck, R.S. (1995) “The Options Approach to Capital
Investment,” Harvard Business Review, May-June, pp.105 - 115.
• Faulker, T. (1996) “Appling ‘Options Thinking’ to R&D valuation Research,”
Technology Management, May-June, pp. 50 - 56.
• Fuggle, R. and Smith, W.T. (2000) “Experience with Dams in Water and Energy
Resource Development in the People’s Republic of China,” World Commission on
Dams Country Review Paper, November. http://www.dams.org/kbase/studies/cn/.
• Gallager, R.G. (1996) Discrete Stochastic Process, Kluwer Academic Publishers,
Boston, MA.
• Geske, R. (1979) “The valuation of compound options,” Journal of Financial
Economics, March, pp. 63 - 81.
• Grenadier, S. and Weiss, A.M. (1997) “Investment in Technological Innovations:
An Options Pricing Approach,” Journal of Financial Economics, 44, pp. 397 – 416.
 154

• Grinblatt, M. and Titman, S. (1998) Options of Financial Market and Corporate

Strategy. Irwin/McGraw-Hill, Boston, MA.
• Hlouskova, J., Kossmeier, S., Obersteiner, M., and Schnabl, A. (2002) “Real
Options Models and Electricity Portfolio Management,” 6th Annual Conference on
Real Options, Paphos, Cyprus. http://www.realoptions.org/abstracts/.
• Howell, S., Stark, A., Newton, D., Paxson, D., Cavus, M., Pereira, J., and Patel, K.
(2001) Real Options – Evaluating Corporate Investment Opportunities in a
Dynamic World, Financial Times Prentice Hall, London, UK.
• Hreisson, E.B. (2000) “Economies of Scale and Optimal Selection of Hydroelectric
Projects,” presented at the IEEE/IEE DRPT2000 Conference, April, City
University, London, UK.
• Hreisson, E.B. (1990) “Optimal Sizing of Projects in a Hydro-based Power
System,” IEEE Transactions on Energy Conversion, 5, 1, March, pp. 32 - 38.
• Huchzermerier, A. and Loch, C.H. (1999) “Project Management Under Risk Using
the Real Options Approach to Evaluate Flexibility in R&D,” INSEAD Working
Paper, March, Cedex, France.
• Hufschmidt, M.M. and Fiering, B. (1966) Simulation Techniques for the Design of
Water-Resource Systems, Harvard University Press, Cambridge, MA.
• Hull, J.C. (1999) Options, Futures, and Other Derivatives, Prentice Hall, Upper
Saddle River, NJ.
• Ito, K. (1951) “On Stochastic Differential Equations,” Memoirs, American
Mathematical Society, 4, pp. 1 - 51.
• Jacoby, H.D. and Loucks, D. (1972) “Chapter 17: The Combined Use of
Optimization and Simulation Models in River Basin Planning,” adapted from Water
Resources Research, 18, 6, Dec, pp. 1401 – 1414. in Systems Planning and
Design (1974) edited by de Neufville, R. and Marks, D.H. Prentice-Hall,
Englewood Cliffs, NJ.
• Johnson, R.A. and Wichern, D.W. (1997) Business Statistics: Decision Making
with Data, John Wiley & Sons, New York, NY.
 155

• Juan, C., Olmos, F., Perez, J.C., and Casasus, T. (2002) “Optimal Investment
Management of Harbor Infrastructure – A Options Viewpoint,” 6th Annual
Conference on Real Options, Paphos, Cyprus.
http://www.realoptions.org/abstracts/.
• Kasanen, E. (1993) “Creating Value by Spawning Investment Opportunities,”
Financial Management, Autumn, pp. 251 - 258.
• Keefer, D.L and Bodly, S.E. (1983) “Three-Point Approximation for Continuous
Random Variables,” Management Science, 29, 5, pp. 595 - 609.
• Kemna, A. G.Z. (1993) “Case Studies on Real Options,” Financial Management,
Autumn, pp. 259 - 270.
• Koekebakker, S. and Sodal, S. (2002) “The Value of an Operating Electricity
Production Unit,” 6th Annual Conference on Real Options, Paphos, Cyprus.
http://www.realoptions.org/abstracts/.
• Kogut, B. and Kulatilaka, N. (1994) “Operating Flexibility, Global Manufacturing,
and the Option Value of a Multinational Network,” Management Science, 40, 1,
pp. 123 - 139.
• Kulatilaka, N. (1993) “The Value of Flexibility: The Case of a Dual-Fuel Industrial
Steam Boiler,” Financial Management, Autumn, pp. 271 - 280.
• Kulatilaka, N. (1988) “Valuing the Flexibility of Flexible Manufacturing Systems,”
IEEE Transactions in Engineering Management, pp. 250 - 257.
• Lambrecht, B.M. (2002) “The Timing of Takeovers under Uncertainty: A Real
options Approach,” Real Options Symposium, University of Maryland, College
Park, MD.
• Laughton, D.G. and Jacoby, H.D. (1993) “Reversion, Timing Options, and Long-
term Decision-Making,” Financial Management, Autumn, pp. 225 - 240.
• Leslie, K.J. and Michaels M.P. (1997) “The Real Power of Real Options,” The
McKinsey Quarterly, 3, pp. 97 - 108.
• Loch, C.H., Pich, M.T., and de Meyer, A. (2000) “Project Uncertainty and
Management Styles,” INSEAD Working Paper, April, Fontainebleau, France.
 156

• Loucks, D., Stedinger, J.R. and Haith, D.A. (1981) Water Resource Systems
Planning and Analysis, Prentice-Hall, Englewood Cliffs, NJ.
• Luehrman, T.A. (1997) “What's it Worth? A General Manager's Guide to
Valuation,” Harvard Business Review, May-June, pp. 132 - 142.
• Luehrman, T.A. (1998a) “Investment Opportunities as Real Options: Getting
Started on the Numbers,” Harvard Business Review, July-August, pp. 51 - 67.
• Luehrman, T.A. (1998b) “Strategy as a Portfolio of Real Options,” Harvard
Business Review, Sep-Oct, pp. 89 - 99.
• Luenberger, D. G. (1998) Investment Science, Oxford University Press, New
York, NY.
• Maass, A., Hufschmidt, M.M., Dorfman, R., Thomas, H.A. Jr., Marglin, S.A., and
Fair, G.M. (1962) Design of Water-Resource Systems, Harvard University Press,
Cambridge, MA.
• McDonald, R. and Siegel, D. (1986) “The Value of Waiting to Invest,” Quarterly
Journal of Economics, 10, November, pp. 707 – 727.
• Major, D.C. and Lenton, L. (1979) Applied Water Resource Systems Planning,
Prentice-Hall, Englewood Cliffs, NJ.
• Major, D.C. (1971) “Chapter 22: Multiobjective Redesign of the Big Walnut
Project,” adapted from Multiple Objective Redesign of the Big Walnut project by
D. C. Major and Associates, Water Resources Council-MIT Cooperative
Agreement WRC-69-3 and Multiobjective Resources Planning presented at the
Fourth Australian Conference on Hydraulics and Fluid Mechanics, Monash
University, Melbourne, Australia, Nov – Dec 1971 by D. C. Major and W. T.
O’Brien. in Systems Planning and Design, edited by de Neufville, R. and Marks,
D.H., Prentice-Hall, Englewood Cliffs, NJ.
• Major, D.C. and Lenton, R.L. (1979) Applied Water Resource Systems
Planning, Prentice-Hall, Englewood Cliffs, NJ.
• Manne, A. S. (1961) “Capacity Expansion and Probabilistic Growth,”
Econometrica, 29, 4, Oct, pp. 632 - 649.
 157

• Mason, S.P. and Merton, R.C. (1985) “The Role of Contingent Claims Analysis
in Corporate Finance,” Recent Advances in Corporate Finance, pp. 7 –54, Edited
by Altman, E. and Subrahmanyam, M. Richard Irwin, Homewood, IL.
• Mauboussin, M.J. (1999) Get Real - Using Real Options in Security Analysis,
Credit Suisse First Boston Corporation, Boston, MA.
• Micalizzi, A. and Trigeorgis, L. (2001) “Eli Lilly - Marketing Strategy of Launching
A New Drug with Interacting Expansion Option,” Real Options Group, Nicosia,
Cyprus.
• Miltersen, K.R. (1997) “Valuation of Natural Resource Investments with Stochastic
Convenience Yield and Interest Rates,” in Project Flexibility, Agency, and
Competition (2000) pp. 183 - 204, edited by Brennan, M.J. and Trigeorgis, L.,
Oxford University Press, New York, NY.
• Mock, R., Schwartz, E., and Stangeland, D. (1989) “The valuation of Forestry
Resources under Stochastic Prices and Inventories,” Journal of Financial and
Quantitative Analysis, December, pp. 473 - 487.
• Montgomery, D.C., Peck, E.A., and Vining, G.G. (2001) Introduction to Linear
Regression Analysis, John Wiley & Sons, New York, NY.
• Morimoto, R. and Hope, C. (2001) “An Extended CBA Model of Hydro Projects in
Sri Lanka,” Research Papers in Management Studies, Judge Institute of
Management, University of Cambridge, UK.
http://www.jims.cam.ac.uk/research/working_papers/, WP 15/01.
• Morimoto, R. and Hope, C. (2002) “An Empirical Application of Probabilistic CBA:
Three Case Studies on Dams in Malaysia, Nepal, and Turkey,” Research Papers
in Management Studies, Judge Institute of Management, University of Cambridge,
UK. http://www.jims.cam.ac.uk/research/working_papers/, WP 19/02.
• Mun, J. (2003) Real Options Analysis: Tools and Techniques for Valuing Strategic
Investments and Decisions, John Wiley & Sons, Hoboken, NJ.
• Myers S.C. and Howe, C.D. (1997) “A Life-Cycle Financial Model of
Pharmaceutical R&D,” Working paper, Sloan School of Management, MIT,
Cambridge, MA.
 158

• Myers, S.C. (1987) “Finance Theory and Financial Strategy,” Midland Corporate
Finance Journal , Spring, pp. 6 - 13.
• Myers, S.C. (1984) “Finance theory and financial strategy,” Interfaces, 14, Jan-
Feb, pp. 126-137.
• Myers, S.C. (1972) “A Note on Linear Programming and Capital Budgeting,” The
Journal of Finance, March, pp. 89 - 92.
• Neely, III, J. E. and de Neufville, R. (2001) “Hybrid Real Options Valuation of
Risky Product Development Projects,” International Journal of Technology, Policy
and Management , 1, 1, pp. 29 - 46.
• Nichols, N.A. (1994) “Scientific Management at Merck: An Interview with CFO
Judy Lewent,” Harvard Business Review, Jan-Feb, pp. 89 -99.
• Oriani, R. and Sobrero, M. (2001) “Market Valuation of Firm's Technological
Knowledge: A Real Options Perspective,” 21st Annual International Conference of
Strategic Management Society, San Francisco, CA.
• Oueslati, S.K. (1999) “Evaluation of Nested and Parallel Real Options: Case
Study of Ford's investment in Fuel Cell Technology,” Master of Science Thesis,
Technology and Policy Program, MIT, Cambridge, MA.
• People’s Daily (2002) “China Steps up Reform of Electric Power System,” May 9,
Beijing, China.
• Pindyck R.S. and Rubinfeld, D.L. (2001) Microeconomics, Prentice-Hall, Upper
Saddle River, NJ.
• Pindyck, R.S. (1993) “Investment of Uncertain Cost,” Journal of Financial
Economics, 34, Aug, pp. 53 – 76.
• Quigg, L. (1993) “Empirical Testing of Real options Pricing Models,” Journal of
Finance, June, pp. 621 - 640.
• Ramirez, N. (2002) “Valuing Flexibility in Infrastructure Developments: The
Bogota Water Supply Expansion Plan,” Master of Science Thesis, Technology
and Policy Program, MIT, Cambridge, MA.
• Rocha, K., Moreira, A., and David, P. (2002) “Investment in Thermopower
Generation: A Real Options Approach for the New Brazilian Electrical Power
 159

Regulation,” 6th Annual Conference on Real Options, Paphos, Cyprus.

http://www.realoptions.org/abstracts/.
• Rogers, J. (2002) Strategy, Value and Risk – The Real Options Approach,
Palgrave Macmillan, New York, NY.
• Sagi, J.S. and Seasholes, M.S. (2002) “Firm-Level Momentum: Theory and
Evidence,” Real Options Symposium, University of Maryland, College Park, MD.
• Savage S. (2000) “The Flaw of Averages,” San José Mercury News, October,
San José, CA.
• Schoenberger, C. R. (2000) “Sidelines,” Forbes, Dec 25, pp. 32-36.
• Schwartz, E. (2002) Patents and R&D as Real Options, UCLA working paper,
April, Los Angles, CA.
• Schwartz, E.S. and Trigeorgis, L. (2001) Real Options and Investment under
Uncertainty, MIT Press, Cambridge, MA.
• Sichuan Hydrology and Hydropower Institute (2002) “The Means of Development
of Hydropower Stations on Yalongjiang River”.
• Siegel, D.R., Smith, J.L., and Paddock, J.L. (1987) “Valuing Offshore Oil
Properties with Option Pricing Models,” Midland Corporate Finance Journal,
Spring, pp. 22 - 30.
• Smit, H. T.J. and Ankum, L.A. (1993) ”A Real Options and Game-Theoretic
Approach to Corporate Investment Strategy Under Competition,” Financial
Management, Autumn, pp. 241 - 250.
• Smith, J.E. and Nau, R.F. (1995) “Valuing Risky Projects: Option Pricing Theory
and Decision Analysis,” Management Science , 41, 5, pp. 795 - 816.
• Smith, K.W. and Triantis, A. (1993) “The Value of Options in Strategic
Acquisition,” Real Options in Capital Investment: New Contributions, Edited by
Trigeorgis, L. Praeger, New York, NY, pp. 405 - 418.
• Titman, S. (1985) “Urban Land Prices under Uncertainty,” American Economic
Review, June, pp. 505 - 514.
• Triantis, A. Portal real options at
http://www.rhsmith.umd.edu/finance/atriantis/RealOptionsportal.html
 160

• Trigeorgis, L. (1996) Real Options: Managerial Flexibility and Strategy in

Resource Allocation, MIT Press, Cambridge, MA.
• Trigeorgis, L. (1993) “The Nature of Options Interactions and the Valuation of
Investments with Multiple Options,” Journal of Financial and Quantitative Analysis,
Spring, pp. 1 - 20.
• Trigeorgis, L. (1993) “Real Options and Interactions with Financial Flexibility,”
Financial Management, Autumn.
• Trigeorgis, L. and Mason, S.P. (1987) “Valuing Managerial Flexibility,” Midland
Corporate Finance Journal , Spring, pp. 202 - 224.
• Tufano, P., and Moel, A. (2000) “Bidding for the Antamina Mine – Valuation and
Incentives in a Real Option Context,” pp. 128 – 150. in Project Flexibility, Agency,
and Competition edited by Brennan, M., Trigeorgis, L., Oxford University Press,
Oxford, UK.
• Williams, J. (1991) “Real Estate Development as an Option,” Journal of Real
Estate Finance and Economics, June, pp. 91 - 208.
• World Energy Council (1999) Hydropower, Energy Info Center.
http://www.worldenergy.org/wec-geis/publications/reports/ser/hydro/hydro.asp