Utility and multi-criteria decisions
Prepared by Dr. Sami Alzahrani
PMG 621
Risk and Decision Analysis
Lecture 5
Utility and multi-criteria decisions
Lecture 5
PMG 621
Risk and Decision Analysis
Prepared by
Dr. Sami Alzahrani www.midocean.ae
Chapter
Introduction
05
This chapter is for people and organizations that are not content to value
decision outcomes in terms of money or money-equivalence. Even if
monetary wealth is the objective, a decision maker may want a
conservative risk policy.
www.midocean.ae
Chapter
Introduction
05
Exceptions to Expected Monetary Value Decision Policy Most
businesses base their decisions upon monetary value-at least that is the
primary consideration. The cost/benefit/risk analysis focuses on the
calculation of expected monetary value (EMV). Recall that EMVis
expected value (EV) of the present value (PV) net cash flow. (The PV
formula is in Chapter 3.) Maximizing EMVis a suitable decision policy for
most business situations.
www.midocean.ae
Chapter
05
However, not all circumstances are appropriate for EMV-maximizing.
The primary exceptions are:
When there is not enough money, or other resource, to do all of the
worthwhile projects or actions (then most organizations use a ranking
criterion, such as discounted return on investment described in Chapter
3, or return on investment, described in Appendix 8B). When possible
outcome values are relatively large, and the decision maker is
conservative.
www.midocean.ae
Chapter
05
When money is not an appropriate measure. Or, when it's not the sole measure of
value, and no money considerations cannot be converted easily into monetary
equivalents. This chapter shows how utility theory provides a logical, consistent way
to deal with these last two situations.
www.midocean.ae
Conservative Risk Attitude
Logical decision making is based upon appraisal. We appraise or
value the alternatives and implement the best one. In evaluating a
particular alternative, we calculate or judge the value for each
possible outcome. We then weigh these possible outcomes with their
probabilities of occurrence. The probability-weighted average value
is the EV.
In decision analysis, we use probabilities to represent judgments
about
uncertainty.
www.midocean.ae
In a credible analysis, all of the analysis inputs should be judged
as objectively as possible. Properly done, the analysis results in
an unbiased distribution of project value (or cost). Assume that
monetary value, PV, is the measure of value. The PV outcome
distribution shows the range of possible outcomes and each
outcome's likelihood of occurrence. The risk-neutral decision
maker need only know the EV PV, which is the expected monetary
value EMV, for each alternative, in order to choose the best one.
The EMV decision rule would be the organization's decision policy.
www.midocean.ae
Conservative Behavior
Conservative behavior is widespread-almost universal. Should this
be an important part of decision policy? Yes, if it is important. For
small decisions, decision makers can afford to be risk-neutral. When
the outcomes become large compared to the reference net worth,
often decision makers will make substantial value adjustments for
risk attitude. For large, publicly held corporations and governments,
risks are shared by many investors. For these entities, many of us
believe that the EMV decision rule is appropriate.
www.midocean.ae
For individuals and small, closely held companies, a
conservative risk policy is more suitable. Still, small day-
to-day decisions will be almost exactly the same as with
the EMV decision rule. However, decisions with large
potential gains or losses may require conservatism.
Utility theory describes how to mow an EMV decision
analysis for a conservative risk attitude.
www.midocean.ae
Utility is a measure of value reflecting the preferences of the decision maker, based
upon beliefs and values. The EV concept still applies for in utility units instead of PV.
The decision rule is to choose the decision making under uncertainty; outcome
value, however, is measured alternative having the highest expected (value) utility
(EU).
www.midocean.ae
Utility Function for Risk Policy:
In this section, we are using utility theory only for expressing
risk attitude. Later, we will discuss how utility theory applies
also to multi-criteria decisions. Assume that maximizing
monetary value is the objective. The objective value
measure, then, would be PV using an appropriate discount
rate. In the probabilistic sense, the objective measure is
EMV.
www.midocean.ae
Risk aversion is exhibited in what economists call the law of
diminishing marginal utility. As an example, for most persons, winning
a $1 million prize would be a tremendously exciting event. If one
already has $10 mil- lion, however, the added $1 million would be
hardly noticed. Incremental positive amounts add incrementally less
value as wealth is accumulated.
www.midocean.ae
Thus, for conservative persons, value is not a linear function of PV;
twice the positive PV does not represent twice as much value. We
could graph a function relating to how we perceive value versus the
actual monetary value. This graph is a utilityfinction. Economist and
decision theorists use the word utility synonymously with value. Utility
units are often called utils.
www.midocean.ae
Figure 4.1 shows an example utility function. It translates objective
value, PV (x-axis), into perceived value, utility (y-axis). The y-axis
scale is arbitrary, including choice of origin. Utility curves that are
concaved downward rep- resent conservative risk attitudes.
The upward-sloping straight line in Figure 4.1 represents the utility
function of a risk-neutral decision maker, where value is proportional
to PV. This is a risk-neutral decision policy, i.e., an EMV decision
maker.
www.midocean.ae
www.midocean.ae
Many decision analysts scale the y-axis so that the worst
possible out- come has utility = 0, and the best possible
outcome has utility = 1. The curve, regardless of shape or
scale, is usually deduced from the decision maker's
answers to a series of hypothetical decision problems. I
prefer the smooth exponential shape. Rather than
arbitrarily defining the scale, I prefer to scale the y-axis in
somewhat tangible units: risk-neutral (RN) dollars. This
makes the utility measure more meaningful:
www.midocean.ae
-RN$l million cost is one million times worse than a $1 cost.
+RN$l million benefit is one million times better than a $1 benefit.
Note that for positive PVs (benefits), incremental utility value decreases
as PVs get larger. Conversely, costs or losses are amplified for negative
PVs. For small decisions, near zero PV, the straight line and the curve are
nearly coincident. The result is that a value determined by EMV is nearly
identical with the value obtained when recognizing risk-aversion.
www.midocean.ae
Applying the Risk Policy:
In decision analysis, we value the alternatives with either their EUs or CEs.
The EMV decision policy is a special case: CE = EMV when the utility
curve is a straight line, meaning that r = $w . Either the EU or CE criterion
results in the same decisions. However, if any alternative is a fixed-dollar
amount, then conversion to CEs is necessary. When evaluating
alternatives with utility, I like to calculate every node's EMV, EU, and CE
when back-solving decision trees. If there are costs placed along tree
branches, these costs may then be subtracted from the CEs as the tree is
back-solved.
www.midocean.ae
www.midocean.ae
Interesting situations arise,
sometimes, when the highest EMV
alternative is not the same as the
highest EU (or CE) alternative. That
is the point of having a risk policy:
to show which risk versus value Chapter
tradeoffs are appropriate.
05
.
An interesting application arises in a problem of optimizing
participation fraction or "working interest" in a project. If the
project has a positive EMV, the EMV decision rule says, "Take it
all." (We are assuming, here, that not investing is the alternative
and has a $0 value.) However, if the potential downside is serious,
it may be advisable to share the risk with one or more partners. If
the project has a positive EMV, then some fraction ownership is
always desirable. However, you may not want it all.
www.midocean.ae
What is the optimal interest you would want in a project?
A utility function representing risk attitude is the logical way to
solve this problem. By iteration, deduce the shape of the function
of EU (or CE) versus participation fraction. Choose the
participation interest that maximizes EU. Keeping 100 percent of
the project is often desirable. However, if the project cost is high
and there is sufficient risk, a fractional share of the project will
maximize utility.
www.midocean.ae
When running a Monte Carlo simulation (simulation; we'll d&cuss
how in Chapter 7), we calculate the outcome PV, then convert it to
utility units in every trial. EU is the average of this utility-measured
outcome. When using simulation, it is customary to provide the
decision maker with risk-versus-value profile curves. These are
typically reverse cumulative frequency distributions for PV, and are
the natural presentation format for simulation results. Most
decision tree software provides similar, although sometimes more
stairstep-looking, graphs. Figure 4.4 shows example reverse
cumulative distribution curves produced by simulation.
www.midocean.ae
www.midocean.ae
if the curves cross, and when the best-EMV alternative is riskier
(wider), having a risk policy is useful. This is the typical and
unfortunate reality: Alter- natives with better EMVs are usually
associated with greater risk, as illustrated in Figure 4.4. Alternative B
has a higher EMV, and has greater uncertainty. Which alternative has
the best risk-versus-value profile? That depends upon the decision
maker, who may visually compare the curves, allowing his intuition to
make the decision. The more consistent-and recommended-way isto
convert each curve into its EU (or CE); then choose the best
alternative based upon the EU decision rule.
www.midocean.ae
Chapter
05
The utility function is a succinct, complete risk policy for an organization
or individual decision maker. This is a powerful concept for a
conservative company that wants to delegate decision making
downward. It enables decision makers at all levels in the organization to
make consistent risk-versus- value tradeoffs. Decision analysis keeps
separate: objective, time value of money, risk attitude, and judgments
about uncertainty. This decomposition facilitates more logical and more
consistent evaluations.
www.midocean.ae
Chapter
05
Multi-Criteria Decisions:
Multi-criteria decision making (MCDM) is an approach for problems
where value is multi-dimensioned. Usually these arise because of
multiple objectives or goals. A city government, for example, faces
tradeoffs across many dimensions such as schools, roads, and police
protection. It can apply MCDM to solve such problems in a reasonably
consistent way. There are three principal ways to deal with MCDM
problems:
www.midocean.ae
Chapter
05
- Recast the problem solely as an economic issue, and maximize
EMV, EU, or CE.
Convert no money considerations into monetary equivalents, then
max- amaze EMV, EU, or CE.
- Use a multi-criteria value function (which can have risk aversion
embedded), and maximize EU.
www.midocean.ae
Chapter
05
Monetary Value Objective Approach:
In project management, goals commonly relate to cost, time, and
performance. How are we to make tradeoffs between these three
dimensions? Understanding and quantdymg tradeoffs is difficult
unless a value model is used. Suppose a project plan is in trouble.
The main alternatives are: Spend an additional $1 million to
complete on time.
www.midocean.ae
Chapter
05
Finish three months late but within budget.
Which alternative is better?
Figure 4.5 maps project outcome value according to two
dimensions, cost and schedule, for illustration. We
measure project outcome value in PV dollars. (We the
uncertainties are unresolved, however, we would be
measuring project value as EMV.) The project feasibility
model determines the shape and position of the is value
contours. In this case, the contour undulations reflect the
seasonality of the business. Interpolating between the
curves:
www.midocean.ae
Chapter
05
www.midocean.ae
Chapter
05
Alternative 1: Cost overrun EMV = $2.4 million.
Alternative 2: Late EMV = $1.5 million.
Both alternatives have unattractive outcomes compared to
the original feasibility study. If maximizing EMV is the
corporate decision policy, then Alternative 1 is the clear and
logical choice. The contours (or the equivalent value
formula) represent decision policy.
www.midocean.ae
Chapter
05
Monetary-Equivalents Approach
Dollars are a convenient and standard measure for
comparing decision alternatives, especially in business.
Increasingly, even government and not-for- profit
organizations are making decisions based primarily upon
monetary value.
www.midocean.ae
Chapter
05
Sometimes, developing a comprehensive value model is
impractical or
inappropriate. However, costs or monetary value often
represent 80 percent or more of the decision basis. Minor
additional considerations can be included in the analysis by
factoring in monetary-equivalent values for the side factors.
www.midocean.ae
Chapter
05
Consider the situation of a national oil company that located a
new re- finery. One site was near an urban area that already
had most of the needed infrastructure and good availability of
labor. An alternative was to locate the refinery in an
impoverished, rural area where jobs were greatly needed. Most
of the decision was driven by the refinery's value in the nation's
economy. However, because the company was government-
owned, it was important to consider and balance other
secondary objectives, such as improving employment
opportunities.
www.midocean.ae
Chapter
05
all that we need for the substitution approach are ways to
translate non- monetary criteria into money equivalents. For
example, it may cost $5,000 to create a new job in the local
economy. Economists can assess the value, cost, and timing of
job creation, and we can then add cost equivalents into the
cash flow projections. Then we can make the decision based
upon EMV or CE, calculated in the usual way.
www.midocean.ae
Chapter
05
Multi-Criteria Value Functions:
For true MCDM problems, the principles of decision analysis
still apply. A key challenge is devising a way to measure value.
Often, multiple objectives are involved, usually the result of
having multiple stakeholder groups. Different stakeholders, of
course, have different preferences. EMV-maximizing
businesses also use MCDM on occasion. It is sometimes
impractical to develop a model to characterize different
alternatives' impacts on corporate value, such as hiring
decisions and employee evaluations.
www.midocean.ae
Chapter
05
Although possible, it is usually cumbersome to attribute
incremental cash flow to an employee because cosmos benefits
are too hard to quantify. In project management, most decisions
are made considering impacts on cost, schedule, and
performance. These form a hierarchy, as shown in Figure 4.6.
This example structure has only two levels, but sometimes
objectives hierarchies have several more levels.
www.midocean.ae
Chapter
05
www.midocean.ae
Chapter
05
Three Pillars of Decision Analysis:
Let me restate and amplify something important in Chapter 1.
Decision
analysis has three key characteristics:
1. Judgments as Probabilities. Usually we ask the most
knowledgeable people available to express their expert
judgments about risks and un-certainties as probability
distributions.
www.midocean.ae
Chapter
05
2. Single Value Measure. The company or decision maker has a
way of measuring outcome value as a single parameter. This is
the "value Function" or "objective function" that we seek to
optimize. For example, many for-profit companies use PV of net
cash flow, PV, for this purpose. This value function may be
composed of several criteria that represent measures for each
of several objectives.
.
www.midocean.ae
Chapter
05
3. Expected Value. The EV calculation is the basis for
calculating value under uncertainty. We condense the
distribution of possible outcome values into a single measure-
EV, the probability-weighted average-for valuing alternatives.
www.midocean.ae
Chapter
05
Recourse:
RISK and DECISION ANALYSIS in Projects
Second Edition
By: John Schuyler
Chapter 4
www.midocean.ae
Thank You
