M A N A G E M E N T
HOLISTIC VS. HOLE-ISTIC
E&P STRATEGIES
B.C. Ball Jr., SPE, Massachusetts Inst. of Technology, and S.L. Savage, Stanford U.
A NEW ERA IN PETROLEUM E&P MANAGEMENT
E&P is a risky business because, while a few exploration projects
are tremendously successful, most are total failures. This makes
management of uncertainty crucial. Development of seismic tech-
nology in the 1930’s and 1940’s substantially reduced the risk in
finding petroleum. The resulting geology and geophysics (G&G)
revolutionized petroleum exploration.
Decision analysis traditionally has been applied to rank projects
hole by hole, determining (on an individual basis) whether they
should be explored and developed. Today this “hole-istic” approach
is being challenged by a holistic one that takes into account the
entire portfolio of potential projects as well as current holdings.
This portfolio analysis starts with representations of the local uncer-
tainties of the individual projects provided by geology and geo-
physics. It then takes into account global uncertainties by adding
two additional G’s: geoeconomics and geopolitics, thereby reducing
risks associated with price fluctuations and political events in addi-
tion to those addressed by traditional G&G analysis. The holistic
approach is based on, but not identical to, the Nobel-prize winning
portfolio theory1 that has shaped the financial markets over the past
4 decades. Our approach deals with the following primary differ-
ences between investments in stocks and E&P projects.
• Stock portfolios depend only on uncertain returns. E&P
projects face both local uncertainties, which involve the discovery
and production of oil at a given site, and global uncertainties,
which involve prices, politics, and other such factors. Furthermore,
uncertainties in stock returns usually follow a bell-shaped curve
while E&P uncertainties are highly skewed and stress rare events.
• Historical data exist for stock prices. E&P uncertainties must
be modeled by decision trees, Monte Carlo simulation, or other
computational means.
• Risk in stock portfolios usually is measured in terms of volatil-
ity—i.e., the degree to which the portfolio swings in value. E&P
portfolios must specifically track downside risk.
• The stock market is quite efficient, whereas the market for E&P
projects may be inefficient. (The term “efficient” is used here in its
technical, economic sense. An efficient market is one in which there
are no barriers to each item being priced at its actual value as deter-
mined by all buyers and sellers; i.e., there are no “bargains”).
• Stocks can be bought or sold at will. E&P projects pay out over
long time periods.
• A stock portfolio generally contains a small fraction of the out-
standing shares of any one company. An E&P portfolio, on the
other hand, often contains 100% of its constituent projects, creat-
ing budgetary effects.
Ref. 2 provides a more detailed list.
Portfolio thinking in petroleum E&P is based on understanding
and exploiting the interplay among both existing and potential pro-
jects. It assists in the following areas.
1. Selecting a set of E&P projects to fund.
Copyright 1999 Society of Petroleum Engineers
Original SPEmanuscript received 1 February 1999. Revised manuscript received 12 June
1999. Paper (SPE 57701) has not been peer reviewed.
74
2. Illuminating trade-offs between long- and short-term goals.
3. Dealing with political and environmental risk.
4. Evaluating a project for purchase or sale.
5.Determining the risk-related cost of constraints and policies.
6.Determining the criteria for projects that would reduce the
risk of the portfolio.
7 .I n c reasing the value of the firm.
We show that analytical models can help to address directly
these and similar issues once decision makers become more com-
fortable with the holistic perspective. First, we present a simple
example that shows how to build intuition into portfolio analysis.
RETRAINING OUR INTUITION
You are responsible for investing U.S. $10 million in E&P. Only
two projects are available, and each requires the full U.S. $10 mil-
lion for a 100% interest. One is relatively “safe,” the other relative-
ly “risky.” The chances of success are independent. Table 1 pro-
vides the information. The expected net present value (ENPV),
vpE, of the two projects, respectively, can be shown to be the same.
(vpE)Safe=60%×$50+40%×(−$10)=$26 million . . . . . . . .(1)
and (vpE)Risky=40%×$80+60%×(−$10)=$26 million.
. . . . . . . . . . . . . . . . . . . . . . .(2)
If you lose money, you also lose your job. Thus, you have only a
40% chance of unemployment with the safe project, and a 60%
chance with the risky one. Both have an ENPV of U.S. $26 million,
so you cannot increase ENPV by investing in the risky project.
Therefore, if you had to choose between one or the other, you
should choose the safe project.
D I V E R S I F I CATION EFFECT
Suppose, however, that you could split your investment evenly
between the two. Intuition cautions against taking 50% out of the
safe project and putting it in the risky one. Examine the four pos-
sible joint outcomes. Because the projects are independent, the
probability of any particular joint outcome is the product of the
probabilities of the associated individual outcomes (Table 2). Note
that the sum of the probability column must be 100% because we
have exhausted all possibilities.
This allocation of funds still provides an ENPV of U.S. $26 mil-
lion. Now the only way you can lose money is with two dry holes,
for which the probability is 40%×60%=24%; this cuts your risk of
unemployment nearly in half! You can reduce risk by moving
money from a safe project to a risky one. Intuition misleads.
Although this answer is correct, it is not so obvious.
This is known as the diversification effect, popularly referred to
as not putting all your eggs in one basket. This line of reasoning is
so fundamental that one wonders how the petroleum industry
could do otherwise. However, a recognized authority summarizes
the conventional selection process with “Just rank exploration
projects by expected present worth.”3
SEPTEMBER 1999 •
 TABLE 1—SAFE AND RISKY
Independent
NPV Probability
Outcome (U.S. million $) (%)
Safe Dry hole –10 40
Success 50 60
Risky Dry hole –10 60
Success 80 40
This suggests that one should rank projects from the best to the
worst (known as high-grading) and then select projects from the
top down until the budget is exhausted, ignoring the diversification
effect. In the example presented in the previous section, this strat-
egy would have led to allocating all the funds to the “safe” project,
with nearly twice the risk of the best portfolio.
Our experience with E&P executives confirms that ranking
“exploration projects by expected present worth” is the norm. A
majority of those informally surveyed favor investing 100% in the
safe project. A few are aware that diversifying a portion of the port-
folio into the risky project actually reduces risk, but even they do
not know how to arrive at an optimal mix.
M E ASURES OF RISK
Uncertainty is an inherent characteristic of our universe. Risk, on
the other hand, is in the eye of the beholder. The naive measure of
risk that we used earlier (the probability of getting fired) was cho-
sen for illustrative purposes because it is easy to visualize and cal-
culate. In practice, a more sophisticated measure would be used
(one, for example, that increasingly penalized larger losses). How-
ever, diversification has a similar effect in reducing almost any sen-
sible measure of risk. Ref. 2 discusses an approach that allows for a
broad class of risk measures.
S TAT I S T I C AL DEPENDENCE AND ITS SOURCES
The example assumes that the projects are independent. In gener-
al, the interplay between two projects is more complex because
their economic outcomes are interrelated. This is known as statis-
tical dependence. The simplest type of statistical dependence is
correlation, which comes in two varieties: positive, where a given
outcome for one project increases the chance of an outcome in the
same direction for the other (this diminishes the effect of diversifi-
cation) and negative, where a given outcome for one project
decreases the chance of an outcome in the same direction for the
other (this enhances the effect of diversification).
Consider the effects of positive correlation on a 50/50 split
between safe and risky. We assume that the individual distributions
of safe and risky are as defined in the example. Now, however, if
safe succeeds, risky is more likely to succeed and, if safe fails, risky
is more likely to fail. Thus, there is still a 40% chance that safe will
fail; but if it does, the chance that risky fails is greater than 60%.
Therefore, the probability of losing your job is now greater than
24%. A similar argument shows that the effect of negative correla-
tion reduces the probability of losing your job to less than 24%. It
is important to note that correlation affects only the risks. The
ENPV of U.S. $26 million remains unchanged. In the portfolio
approach, risk is managed by spreading the investments across a
number of opportunities while avoiding positive correlations and
seeking negative correlations. The goal is to do this optimally over
numerous opportunities under diverse constraints.
The importance of correlation is easy to understand if one con-
siders a fire insurance policy on a house. Because the insurance has
a positive ENPV to the insurer, we know that it has a negative ENPV
to the insured and is therefore a bad investment. [Unless, of course,
your portfolio also includes a house (whose value is, by definition,
negatively correlated to the value of your insurance policy).]
Statistical dependence may be due to many sources. The five list-
ed next are not exhaustive, but are widely encountered in E&P
projects. [Although we use “projects” throughout this paper, the
same principles apply equally to E&P sites, exploration prospects,
development projects, and/or acquisition properties. The point is
that E&P companies can add reserves in three ways: exploration,
development (and redevelopment), and acquisition. Each has a
very different risk/reward profile, the integrated analysis of which
could reveal innovative, optimal portfolios. Most E&P companies
isolate these three functions in their respective functional silos and,
at best, suboptimize each.]
• Places.
• Prices.
• Profiles.
• Politics.
• Procedures.
Places. The economic outcomes of two E&P sites in close proxim-
ity (for example, in the same field) are positively correlated through
geological similarities and would not constitute a very diversified
portfolio. On the other hand, two widely distant sites display little
or no geological correlation and, hence, would be more diversified.
Prices. Petroleum projects produce crude oil and natural gas in
various proportions. Crude-oil prices generally track each other
very closely worldwide. Thus, the economic outputs of oil projects
worldwide are positively correlated relative to fluctuations in crude
price. However, this is not true for natural gas. Natural-gas prices
in many parts of the world—notably in the U.S.—do not track
either world crude oil prices or each other very well. Thus, a port-
folio consisting of a gas project and an oil project would tend to be
less positively correlated and therefore better diversified, relative to
price, than a portfolio consisting of two oil projects.
An example of this phenomenon is the economy of Houston,
which suffered during the crude oil price drop of 1986 but weath-
ered a subsequent price drop successfully because it had diversified
between oil and gas.4
Profiles. A frequent concern is timing of the flows of various ele-
ments of projects, which may extend for many years into the

TABLE 2—ALL POSSIBLE OUTCOMES OF INVESTING 50% IN EACH PROJECT WITH STATISTICAL INDEPENDENCE

Probability Return
Scenario Safe Risky (%) (U.S. million $) Result

1 Success Success 60×40=24 50%×$50+50%×$80=$65 Keep job

2 Success Dry hole 60×60=36 50%×$50+50%×(−$10)=$20 Keep job
3 Dry hole Success 40×40=16 50%×(−$10)+50%×$80=$35 Keep job
4 Dry hole Dry hole 40×60=24 50%×(−$10)+50%×(−$10)=(−$10) Lose job
ENPV of portfolio 24%×$65+36%×$20+16%×$35+24%×(−$10)=$26

76 SEPTEMBER 1999 •

ENPV
Fig. 1—An efficient frontier; S=safe project, R=risky project,
and M=50/50 mix.
future. These flows might include such elements as cash flow,
hydrocarbon production, reserve additions, and staff requirements.
Often, the more nearly constant these flows can be, the better. The
correlation among these elements can be taken into consideration
to minimize fluctuations in cash flows. These critical factors, which
can literally make or break a company, can now be considered and
managed explicitly. For example, two projects that required a large
cash outlay in the same year would introduce a source of positive
correlation that would increase risk. On the other hand, a second
project that might produce cash in that same year would introduce
a source of negative correlation that would reduce risk.
Politics. Petroleum investments have always been subject to politi-
cal uncertainties, from the antitrust decision against Standard Oil in
1911, through environmental regulations, to the Gulf War in 1991,
and beyond. Projects subject to disruption in the same direction
because of the same political event are positively correlated. Nega-
tive correlation of projects may also be induced through political
uncertainty. Consider, for example, two politically distinct regions
that supply natural gas through two different pipelines to a single
market. The political disruption of production in either of the two
regions could lead to market shortages and, hence, to increased
prices and/or demands for the nondisrupted region. A portfolio con-
sisting of one project in each negatively correlated region would
thus be protected, or “hedged,” against political risk in either region.
Procedures. Technical and management procedures can introduce
statistical dependence. For example, a firm that was equipped tech-
nically or managerially to explore only a particular type of prospect
will have a positively correlated portfolio by definition.
THE EFFICIENT FRONTIER
We have shown that a combination of a safe project and a risky
project can be less risky than the safe project alone. Fig. 1 displays
this graphically. The original example offered only two potential
projects with identical ENPV’s. However, the choice generally is
among projects of various ENPV’s and risks, as depicted by the
other circles in Fig.1. By calculating the minimum risk (or optimal)
portfolio for each of several levels of expected return, one can
78
TABLE 3—PROJECT ATTRIBUTES
Project
A B C D E
ENPV, U.S thousand $ 820 171 891 1,255 1,563
Expected cost, U.S. thousand $ 373 47 374 370 384
arrive at the solid curve starting at M and curving up to the right.
This curve shows the optimum trade-off between risk and return
and is known as the efficient frontier.
Moving north from the efficient frontier increases risk without
increasing expected return, while moving west decreases expected
return without decreasing risk. Because each point on the efficient
frontier has minimized the risk for that level of expected return, no
portfolios exist to the southeast of the frontier. Thus, the best port-
folios are those on the efficient frontier itself. The concept of the
efficient frontier and the method of finding it were the fundamen-
tal Nobel-prize-winning contributions of Markowitz.1 No rational
person would wish to be at any point above the efficient frontier.
But which point should you pick? That depends on your firm’s
willingness to suffer short-term volatility in the interest of long-
term growth. In E&P, one may also develop additional frontiers
showing optimal trade-offs between reserve additions, budget level,
cash flow shortfall, or other meaningful metrics.
SHOOTING THE MOON
Some E&P executives feel that a major discovery every decade or
so can sustain a large company. They claim that they prefer to
“shoot the moon,” aiming for the highest possible expected return,
regardless of the risk. They argue that they will, in fact, arrive at the
highest-return/highest-risk point (the efficient frontier) by ranking
their projects in order of expected value, then marching down the
list until they run out of money.
This is false. Borrowing cash and adding more projects would
have the effect of moving them to an even higher-risk/return point
than that achieved through ranking. Thus, high-grading does not
deliver the highest point on the efficient frontier at all but results in
an arbitrary and unexamined trade-off between risk and return.
We are not arguing that a company should explore with bor-
rowed money; we are merely observing that high-grading ignores a
universe of other portfolios, both riskier and safer, that can be illu-
minated only through a holistic approach.
E&P PORTFOLIO OPTIMIZATION MODEL
We have developed a simple E&P portfolio optimization (EPPO)
model, which, unlike financial portfolio models, can accommodate
the characteristics of petroleum projects listed earlier. An Excel
spreadsheet version of EPPO may be downloaded from
http://www.stanford.edu/~savage or http://www.ziplink.net/~benball.
Ref. 2 provides a full description of EPPO.
As discussed earlier, traditional financial portfolio models were
intended for stocks (Ref. 5 includes an Excel version of the
Markowitz model. See also http://www.AnalyCorp.com.) and are
not suited to portfolios of petroleum projects. EPPO is based on
two technologies already in wide use individually in the petroleum
industry: Monte Carlo simulation and linear programming. These
technologies have been combined to create a single period sto-
chastic linear program.6 This model has the following advantages
for our purposes.
• It allows for arbitrary realistic probability distributions of
project outcomes as opposed to multivariate normal or log-normal.
• It supports a wide variety of risk measures.
SEPTEMBER 1999 •
 TABLE 4—MANAGEMENT IMPLICATIONS

Current Practice Asset-Interplay Management

Selecting set of E&P projects Rank (high-grade) projects. Start at top Select the set of projects that achieves optimal
to fund of the list and go down until budget trade-offs of various risks and economic
is exhausted. factors.
Illuminating trade-offs between Establish long-term return on investment (or Make informed trade-offs between long-term
long-and short-term goals return on capital employed or other similar goals for growth and the risks of short-
metric) and growth goals independent term volatility, failure to meet reserve
of their implications for concomitant- requirements, or other measures of risk.*
long-term risk or short-term volatility.
Dealing with political and Political and environmental risks are Political and environmental risk of entire
environmental risk considered subjectively on a project-by- portfolio is managed by taking interplay
project basis. among projectsinto account.
Evaluating a project for Determine project’s “market” value (what Determine project’s worth in context of firm’s
purchase of sale other firms might pay for it) or base current holdings. (Remember, I won’t spend a
its value on an estimate of its ENPV. a dime to insure your house, but an identical
policy on my house is invaluable to me even
though it has a negative ENPV.)
Determining risk-related cost Costs of company policies and external Implications of constraints and company
of constraints and policies constraints are appraised subjectively. policies can be evaluated on a risk/return
basis.**
Determining strategic criteria Missing ingredients or strategic weaknesses Missing ingredients or strategic thrusts that
for future exploration are difficult to identify. would contribute to portfolio robustness are
identified [e.g., investigating oil projects,
(instead of gas projects) that reduce the
impact of price uncertainty). This provides
a list of desirable qualities for additional
projects or programs, given a firms current
portfolio and situation. †
Increasing value of the firm Sale focus is on ENPV. Focus of real E&P companies is never solely on
ENPV but includes reserves replacement,
debt ratios, cash flows, and other such
factors. At one extreme, risk is ignored
almost entirely. However, Froot et al. 7
contend that situations exist where risk
reduction can increase the value of a firm
by ensuring that cash flow is available
when needed for critical investment.
*Note that this approach captures the insights underlying utility theory, preference curves, and other such parameters, while avoiding the practical difficulties that arise from their
explicit application.
**Constraints
often reflect "strategic" issues of concern to top management (e.g., reserves replacement, cash flow for debt repayment). Relative to stock market valuation, these
may at times be as important or more important than ENPV. These constraints can be incorporated into the portfolio analysis and evaluated for their effects on risk and return.

†"Current holdings" represents the mass of most portfolios. "Hold and produce" takes little or no capital, requires no overt decisions, and yields great returns, especially if opportu-
nity costs are ignored. However, serious consideration of the trading of current holdings (i.e., the sale of current assets in exchange for the purchase of new ones) usually is not
seriously considered in any systemic way. Such a study, however, could significantly enhance the efficient frontier. A crucial point is that both current holdings and new project
opportunities must be evaluated together because the portfolio’s risk depends on the ways in which all of its constituent parts interact.

• It can be based on simulations, decision trees, or other types of

input instead of requiring historical data.
• It estimates the entire distribution of outcomes, not just the
mean and variance of the portfolio.
EPPO works by generating Monte Carlo trials representing the
uncertainties of the various projects, including the interplay among
projects. The trials are then fed into a linear program, which finds
a portfolio of minimum risk for a given ENPV. This process is
repeated for a range of desired ENPV’s, thereby determining the
efficient frontier and the portfolios that make it up.

R E S U LT S
In the example presented here, the efficient frontier was found for
five exploration projects, A through E. Table 3 lists some of the
attributes for these projects but does not fully describe their
underlying uncertainties, which were modeled by Monte Carlo
Fig. 2—Example of efficient frontier and budget compositions
for a U.S. $600,000 budget. simulation. Fig. 2 shows the results, which were produced by
running the EPPO model with a budget of U.S. $600,000. This

80 SEPTEMBER 1999 •
 graph has two elements, both of which have ENPV on the hori-
zontal axes.
The first element is an efficient frontier similar to that in Fig. 1,
with ENPV running from U.S. $1.9 million to $2.3 million, and risk
running from U.S. $100,000 to just under U.S. $300,000 on the
right-hand vertical axis. This curve is the internal efficient frontier
because it represents the best the firm can do with investments
among its own projects. As mentioned earlier, variance (the measure
of risk used in the financial markets) is unsuitable here because it
treats upside and downside risks identically. The risk measure in this
example is mean loss, one of a broad class of risk measures that the
EPPO model can accommodate. (Ref. 2 provides a detailed discus-
sion of risk measures.) A firm that does not apply portfolio analysis
(position denoted by X in the figure) would be unlikely to be on
such an internal efficient frontier. Such a firm could increase its
expected return at constant risk by moving to Z, decrease its risk at
constant return by moving to Y, or use some intermediate strategy.
The second element of the graph consists of the stacked bars,
which show the makeup of the efficient portfolio for various levels
of ENPV. The vertical bars show the percentage of the budget (on
the left-hand vertical axis) that make up each project in each effi-
cient portfolio.
Once management chooses a point on the frontier specifying a
desired risk/return trade-off, the underlying portfolio associated
with that point is revealed. For example, if management is satisfied
with an ENPV of U.S. $2.2 million and a risk value of U.S.
$170,000, then the appropriate portfolio would allocate 6% in
Project A, none in Project B, 8% in Project C, 22% in Project D, and
64% in Project E. [These “results” are no more than a first iteration.
For example, if 6% working interest in Project A is not practical,
then the program should be rerun after placing the appropriate
constraints on the working interest in Project A (e.g., ≥10%). Such
iterations should be continued until all results are within the realm
of practicality. This, then, represents the optimum practical portfo-
lio, which, of course, is the only one that matters in the real world.]
By contrast, current practice would be to decide separately on an
investment level in each of the five projects largely on the basis of
the intrinsic merits of each. The resulting portfolio would likely be
northwest of the internal efficient frontier, like the Point X. The
total risk would be higher than necessary; the expected value
would be lower than necessary; or, most likely, both.
In the hands of talented petroleum industry analysts, EPPO
served as the foundation for important new diagnostics and visual-
ization of managerial options.8,9

BUSINESS IMPLICATIONS:
ASSET INTERPL AY MANAGEMENT
We contend that project-by-project or “hole-istic” analysis misses
many important insights provided by the holistic perspective. The
implications for the E&P business are that management must
place emphasis on the interplay among projects as well as on the
individual merits of the projects themselves. This is asset interplay
management, a tool that we believe is not adequately exploited by
the industry. Table 4 shows how various types of corporate deci-
sions might be transformed by asset interplay management.
It is tempting to think of asset interplay management as just
another analytical tool or computer program. The danger in this
view is that the tool or program will be adopted and won’t deliver,
thereby “inoculating” the company against a successful case of port-
folio optimization for a generation of management. If the portfolio
approach is to meet the expectations it is generating, top manage-
ment must understand that it represents a fundamentally new way
of thinking about the business of E&P. This requires the following.
• Re-educating management to develop intuition into the holis-
tic approach.

82 SEPTEMBER 1999 •
 • Restructuring corporate systems to collect and interpret sto-
chastic data from a global as well as local perspective.
• Revising reward programs to provide incentives for overall
risk/reward positioning of the firm.
Each of these implications of asset-interplay management is
valuable in itself and offers insights not available through project-
by-project analysis. They also avoid some of the subtle but systemic
errors to which the industry currently is vulnerable. They are, how-
ever, just first steps as E&P enters the dawn of this new style of
management.
C O N C LU S I O N S
• Portfolio principles developed for the financial arena must be
modified before application to E&P.
• The holistic approach empowers decision makers to manage as
well as measure risk by focusing on sources of interactions, such as
places, prices, profiles, politics, and procedures.
• Changes in perspective, intuition, and culture will do more to
promote asset-interplay management than new computer programs.
Modern portfolio theory provided the conceptual underpinning
for the financial engineering that now dominates Wall Street. How
this will play out in the area of E&P remains to be seen. However,
one thing is certain. The holistic approach deals head-on with one
of the elements that distinguishes the upstream business but that
has been handled only subjectively for too long—risk.
REFERENCES
1. Markowitz, H.: Portfolio Selection—Efficient Diversification of Investments,
second edition, Blackwell Publishers Inc., Malden, Massachusetts (1997).
2. Ball, B.C. and Savage, S.L.: “Notes on Exploration and Production Port-
folio Optimization,” available for download from http://www.
stanford.edu/~savage or http://www.ziplink.net/~benball (1999).
84
3.Downey, M.: “Business Side of Geology,” AAPG Explorer (December
1997) 28.
4. “Regional Resilience—Tumbling Oil Prices Won’t Batter Texas The Way
‘86 Crash Did,” Wall Street Journal (6 December 1993) 1.
5. Savage, S.L.: “Insight.xl—Business Analysis Software for Microsoft®
Excel,” Duxbury Press, Pacific Grove, California (1998).
6.Infanger, G.: Planning Under Uncertainty, Boyd & Fraser Publishing,
Danvers, Massachusetts (1994).
7 .F root, K.A., Scharfstein, D.S., and Stein, J.C..: “A Framework for Risk
Management,” Harvard Business Rev. (November–December 1994) 91.
8. Howell, J.H. III, Anderson, R.N., and Bentz, B.: “Managing E&P Assets
From a Portfolio Perspective,” Oil & Gas J. (30 November 1998) 56.
9. Anderson, R.N. et al.: “Quantitative Tools Link Portfolio Management
With Use of Technology,” Oil & Gas J. (30 November 1998) 46.
Ben C. Ball Jr. is a petroleum consultant and for 23 years has
held appointments at Massachusetts Inst. of Technology (MIT),
including that of Adjunct Professor of Management and Engi-
neering. He also serves as an expert witness. He retired from
Gulf Oil Corp. as Corporate Vice President after 30 years in
operations and planning. He is the coauthor of Energy After-
math. Ball holds BS and MS degrees in chemical engineering
from MIT and completed Harvard Graduate Business School’s
Advanced Management Program. Sam L. Savage is Director
of Industrial Affiliates for Stanford U.’s Dept. of Engineering
Economic Systems and Operations in Stanford, California, and
was a member of the faculty of the U. of Chicago Graduate
School of Business. He also consults widely, serves as an expert
witness, and offers executive seminars. He led the develop-
ment of a software package that couples linear programming
to Lotus 1-2-3 in 1985 and has published analysis software.
Savage holds a PhD degree in computer science from Yale U.
SEPTEMBER 1999 •