Lecture slides to accompany

Engineering Economy, 8th edition

Leland Blank, Anthony Tarquin

©McGraw-Hill Education. All rights reserved. Authorized only for instructor use in the classroom. No reproduction or further distribution permitted without the prior written consent of McGraw-Hill Education.
 Chapter 18

Sensitivity Analysis and Staged

Decisions

©McGraw-Hill Education. All rights reserved. Authorized only for instructor use in the classroom. No reproduction or further distribution permitted without the prior written consent of McGraw-Hill Education.
 LEARNING OBJECTIVES
1. Explain sensitivity to parameter variation
2. Use three estimates for sensitivity analysis
3. Calculate expected value E(X)
4. Determine E(X) of cash flow series
5. Use decision trees for staged decisions
6. Understand real options for staged funding

©McGraw-Hill Education.
 Parameters and Sensitivity Analysis
Parameter -- A variable or factor for which an estimated or stated
value is necessary
Sensitivity analysis – An analysis to determine how a measure
of worth (e.g., PW, AW, ROR, B/C) changes when one or more
parameters vary over a selected range of values.

PROCEDURE:
1. Select parameter to analyze. Assume
independence with other parameters
2. Select probable range and increment
3. Select measure of worth
4. Calculate measure of worth values
5. Interpret results. Graph measure vs.
parameter for better understanding
©McGraw-Hill Education.
 Sensitivity of Several Parameters
When several parameters for one alternative vary
and analysis of each parameter is required …
graph percentage change from the most likely estimate
for each parameter vs. measure of worth

Plots with larger slopes

(positive or negative)
have a higher sensitivity
with parameter variation
(sales price curve)

Plots that are relatively

flat have little sensitivity
to parameter variation
(indirect cost curve)
©McGraw-Hill Education.
 Three Estimate Sensitivity Analysis
Applied when selecting one ME alternative from two or more
For each parameter that warrants analysis, provide three
estimates:
• Pessimistic estimate P
• Most likely estimate ML
• Optimistic estimate O

Calculate measure of worth for each alternative and 3 estimates

and select ‘best’ alternative
Notes -- 1. The pessimistic estimate may be the lowest for some parameters and
the highest for others, e.g., low life estimates and high first cost
estimates are usually pessimistic
2. When calculating the measure of worth, use ML estimate of a
parameter as others varies. This is the independence assumption

©McGraw-Hill Education.
 Expected Value Calculations
Expected Value -- Long-run average observable if a project
or activity is repeated many times

Result is a point estimate based on anticipated outcomes and

estimated probabilities
m
E ( X ) =  X i P( X i )
i =1

Where: Xi = value of variable X for i = 1, …, m different values

P(Xi) = probability that a specific value of X will occur

In all probability statements, the sum is: When E(X) < 0, e.g., E(PW) = $−2550, a
m

 P( X ) = 1.0
i =1
i
cash outflow is expected; the project is not
expected to return the MARR used
©McGraw-Hill Education.
 Example: Probability and Expected Value
Monthly M&O cost records over a 4-year period are shown in
$200 ranges. Determine the expected monthly cost for next year,
if conditions remain constant.
Range,$, X No. of months Range,$, X No. of months
100–300 4 700–900 6
300–500 12 900–1100 10
500–700 14 1100–1300 2

Solution:
𝐏(𝐗) = 𝐧𝐮𝐦𝐛𝐞𝐫 𝐨𝐟 𝐦𝐨𝐧𝐭𝐡𝐬/𝟒𝟖 𝐦𝐨𝐧𝐭𝐡𝐬
𝐄(𝐗) = 𝟐𝟎𝟎(𝟒/𝟒𝟖) + 𝟒𝟎𝟎(𝟏𝟐/𝟒𝟖) + ··· +𝟏𝟐𝟎𝟎(𝟐/𝟒𝟖)
= 𝟏/𝟒𝟖[𝟐𝟎𝟎 × 𝟒 + 𝟒𝟎𝟎 × 𝟏𝟐 + ··· +𝟏𝟐𝟎𝟎 × 𝟐]
= 𝟏/𝟒𝟖[𝟑𝟏, 𝟐𝟎𝟎]
= $𝟔𝟓𝟎/𝐦𝐨𝐧𝐭𝐡
©McGraw-Hill Education.
 Expected Value for Alternative Evaluation
Two applications for Expected Value for estimates:
1. Prepare information for use in an economic analysis
2. Evaluate economic viability of fully formulated alternative

Example: Second use for a complete alternative. Is the investment viable?

P = $−5000
n = 3 years
MARR = 15%

©McGraw-Hill Education.
 Example: Expected Value for Alternative Evaluation

Solution: Calculate PW value for each condition

𝐏𝐖𝐑 = −𝟓𝟎𝟎𝟎 + 𝟐𝟓𝟎𝟎(𝐏/𝐅, 𝟏𝟓%, 𝟏) + 𝟐𝟎𝟎𝟎(𝐏/𝐅, 𝟏𝟓%, 𝟐) + 𝟏𝟎𝟎𝟎(𝐏/𝐅, 𝟏𝟓%, 𝟑)
= $– 𝟔𝟓𝟔 (cash outflow; not viable)
𝐏𝐖𝐒 = $ + 𝟕𝟎𝟖 (cash inflow; viable)
𝐏𝐖𝐄 = $ + 𝟏𝟑𝟎𝟗 (cash inflow; viable)
Now, calculate expected value of PW estimates
𝐄(𝐏𝐖) = 𝐏𝐖𝐑 × 𝐏(𝐑) + 𝐏𝐖𝐒 × 𝐏(𝐒) + 𝐏𝐖𝐄 × 𝐏(𝐄)
= −𝟔𝟓𝟔 × 𝟎. 𝟒 + 𝟕𝟎𝟖 × 𝟎. 𝟒 + 𝟏𝟑𝟎𝟗 × 𝟎. 𝟐
= $ + 283
On basis of E(PW) > 0 at 15% over 3 years, investment is viable
©McGraw-Hill Education.
 Decision Tree Characteristics
Staged Decision – Alternative has multiple stages; decision at one
stage is important to next stage; risk is an inherent element of the
evaluation
Decision Tree – Helps make risk more explicit for staged decisions

A DECISION TREE INCLUDES:

• More than one stage of selection
• Selection of an alternative at one stage
leads to another stage
• Expected results from a decision at each
stage
• Probability estimates for each outcome
• Estimates of economic value (cost or
revenue) for each outcome
• Measure of worth as the selection criterion
©McGraw-Hill Education.
 Solving a Decision Tree
Once the tree is developed, probabilities and economic information
are estimated for each outcome branch, and the measure of worth is
selected (usually PW), use the following, starting at top right of tree:

PROCEDURE TO SOLVE A DECISION TREE

1. Determine PW for each outcome branch
2. Calculate expected value for each alternative:
𝐄(𝐝𝐞𝐜𝐢𝐬𝐢𝐨𝐧) = ∑(𝐨𝐮𝐭𝐜𝐨𝐦𝐞 𝐞𝐬𝐭𝐢𝐦𝐚𝐭𝐞) × 𝐏(𝐨𝐮𝐭𝐜𝐨𝐦𝐞)
3. At each decision node, select the best E(decision) value
4. Continue moving to left to the tree’s root to select the best
alternative
5. Trace the best decision path through the tree
©McGraw-Hill Education.
 Example: Solving a Decision Tree
1. PW of CFBT is estimated
2. PW for decision nodes
14
𝐄(𝐢𝐧𝐭’𝐥) = 𝟏𝟐(𝟎. 𝟓) + 𝟏𝟔(𝟎. 𝟓) = 𝟏𝟒
D2 𝐄(𝐧𝐚𝐭’𝐥) = 𝟒(𝟎. 𝟒) − 𝟑(𝟎. 𝟒) − 𝟏(𝟎. 𝟐) = 𝟎. 𝟐
D2
𝐄(𝐢𝐧𝐭’𝐥) = 𝟔(𝟎. 𝟖) − 𝟑(𝟎. 𝟐) = 𝟒. 𝟐
D3 𝐄(𝐧𝐚𝐭’𝐥) = 𝟔(𝟎. 𝟒) − 𝟐(𝟎. 𝟒) + 𝟐(𝟎. 𝟐) = 𝟐
4.2
3. Decisions: 14 (int’l) @D2 and 4.3 (int’l)
@D3
D3 4. PW for decision node D1
𝐄(𝐦𝐚𝐫𝐤𝐞𝐭) = 𝟏𝟒(𝟎. 𝟐) + 𝟒. 𝟐(𝟎. 𝟖) = 𝟔. 𝟏𝟔
D1
D1
𝐄(𝐬𝐞𝐥𝐥) = 𝟗(𝟏. 𝟎) = 𝟗

Decision: 9 (sell)
5. Trace through tree; select D1 to sell
at E(PW of CFBT ) = $9 million
©McGraw-Hill Education.
 Real Options
Staged funding ─ Decision to buy or invest can be delayed. There is
usually cost and risk involved to delay the decision

Option ─ Contractual Real option ─ In engineering

agreement to take a specified economy, the option can involve
action at some stated future time. physical assets (thus the title real
In other words, pay some amount option), leases, subcontracts, etc.
now to reserve the right to accept Risk analysis is always involved
or reject an offer in the future for the predictable future events

Real option example: An airline purchases 3 commercial planes now and pays
$2 million to the manufacturer for the option to buy up to 5 more within the
next 3 years at today’s price.
If accepted, the $2 million is 25% credited toward the delayed purchase; if the
option is not exercised within 3 years, the entire $2 million is forfeited.
©McGraw-Hill Education.
 Real Options Analysis
Real options analysis ─ Determine the economic consequences
of delaying the funding decision, that is, analyze staged funding

PRIMARY CHARACTERISTICS OF REAL OPTIONS ANALYSIS

• Cost of option to delay, i.e., PW of investment/payment required now
• Future options and cash flow estimates (staged funding)
• Time period for follow-on decision (staged decision)
• Market and risk-free interest rates (MARR and estimated inflation)
• Estimates of risk, i.e., probabilities for each option’s cash flows
• Economic criterion (PW, ROR) to make a decision now on the real option
• If needed, decision tree of options, probabilities, and cash flow
estimates to assist with the analysis

©McGraw-Hill Education.
 Example: Real Options Analysis (1)

A real estate developer has the option to buy prime property 2 years from now for $35M, if a $3.5M
option is purchased now. In 2 years, the economy can be ‘up’ or ‘down’ and the decisions then are:
(1) exercise the option (buy at $35M) and hold; (2) exercise and sell immediately; or (3) forfeit. PW of
eventual net cash flows for further development are predicted in year 2 depending upon the
economy (up or down). Selling price (high or low) 2 years hence is estimated. At MARR = 12%, what
is better economically now ; to accept or to decline the option? Assume probabilities and cash flows
are estimated as follows:

PW of CF, Estimated
year 2, Selling P(selling selling price,
Economy P(economy) if held, $M environment environment) $M
Up 0.3 50 High 0.4 50
Up 0.3 50 Low 0.6 40
Down 0.7 30 High 0.4 30
Down 0.7 30 Low 0.6 25

Solution: Construct the 2-stage decision tree for time now and 2 years from
now. Probabilities and PW of cash flows are shown on the tree.
©McGraw-Hill Education.
 Example: Real Options Analysis (2)

YEAR NOW 2 Future outcome

Exercise/hold -35 + 50 = $15M
15M
High
(-35 +50)(0.4) = $6M
Exercise P = 0.4
D2 /sell

Accept Low (-35+40)(0.6) = $3M

UP
option P = 0.6
P = 0.3
0.1M -$3.5M Forfeit $0
D1
DOWN Exercise/hold -35 + 30 = $-5M
Decline P = 0.7 0 High
(-35+30)(0.4) = $-2M
option Exercise P = 0.4
0$ D3 /sell

Low (-35+25)(0.6) = $-6M

P = 0.6

Forfeit $0
0 Largest E(X)
of decision
$0 branches
P = 1.0
©McGraw-Hill Education.
 Example: Real Options Analysis (3)

D2 analysis: PW in year 2, PW2

Tree from previous slide Exercise/hold: PW2 = purchase + PW of future
cash flows = −35 + 50 = $15M
Exercise/sell; high: PW2 = (− 35 + 50)(0.4) = $6M
Exercise/sell; low: PW2 = (− 35 + 40)(0.6) = $3M
Forfeit: 0
D2 decision: Select exercise/hold at $15M

D3 analysis: PW in year 2, PW2

D3 decision: Select forfeit at 0

D1 analysis: PW in year 0 (NOW), PW0

Accept option; up: PW0 = option $ + PW of D2
= − 3.5 + 15(P/F,12%,2)(0.3) = $0.1M
Accept option; down: PW0 = − 3.5+0(0.7) = $−3.5M
Decline option: PW0 = 0
D1 decision: Select accept option at $0.1M

CONCLUSION: Accept option now; buy property in two years and hold it
©McGraw-Hill Education.
 Summary of Important Points
Sensitivity analysis evaluates variation in parameters using a specific
measure of worth (PW, ROR, B/C, etc.)

Independence of parameters is assumed in sensitivity analysis

If E(PW) < 0, an alternative is not expected to return the stated

MARR, given the estimated probabilities

Decision trees assist in making staged decisions when risk is

explicitly considered

Real options analysis determines the economic consequences of

delayed funding with risk accounted for; however, an up-front price is
usually imposed to have the option of staged funding

©McGraw-Hill Education.