DK5739 CH10

10
Sensitivity and Uncertainty Analysis
In Chapter 9, it was assumed that all values of costs, expenses, revenues,

economic lives, and acceptable rates of return were known with certainty. This
was done to focus on the methodology of estimating profitability and ultimately
decision making. This high degree of confidence is referred to as analysis under
assumed certainty; therefore, decisions based on this analysis are decisions under
certainty [1].
Whenever a project is proposed, market surveys, market price
projections, market share, and ultimately revenues are estimated. Capital
costs and operating expenses are estimated as in Chapters 4 and 5,
respectively. From these data, profitability of a project is calculated. All these
estimates are based on what is believed to be the best available data. Errors
inherent in all estimates, and the effect of these errors on profitability, will be
considered in this chapter.
Sensitivity analysis is used to determine the effect of technical and
economic parameters on the profitability of a project. The potential error of each
parametric variable is examined, as well as its effect on the project. Questions
such as “What if the capital investment is 15% greater than the estimated value?”
or “What if the market is 20% less than the best estimate?” can be resolved by
determining the effect of percentage changes in these variables on profitability
and will be addressed subsequently in this chapter.
In uncertainty analysis, probabilistic distributions are assigned to each
variable to be considered and a methodology is used to determine the resulting
profitability measure expressed as either the “probability of achieving more than
or at least a specified return on investment or a net present worth.”
TM
Copyright © 2003 by Marcel Dekker, Inc. All Rights Reserved.
246 Chapter 10
Sensitivity analysis is easy to use but the results have definite limitations.
Uncertainty analysis requires much more sophistication on the part of the person
performing the analysis. The executives who will use the results need to be
educated in the method used and the meaning of the results; otherwise wrong
decisions could be made concerning the fate of a future venture.
10.1 SENSITIVITY ANALYSIS

Sensitivity analysis is concerned with the extent of change in a cost analysis
resulting from variations in one or more elements of a cost study. It shows the
influence of possible changes of significant variables upon profitability. From this
analysis, those variables that have a critical effect are identified. Especially
important are those variables that might alter a decision when only small changes
occur.
The ordinary practice is to make a number of computations of profitability,
varying each significant cost element over the most likely range of values. This
process can be tedious unless a computer is used. A visual aid, e.g., a plot or
graph, depicts the most sensitive variables in a cost study. Managers prefer a
graphic illustration rather than tables of numerical values because significant
results can be overlooked in tables.
10.1.1 Break-Even Analysis

Break-even analysis is a simple form of sensitivity analysis, and is a useful
concept that can be of value to managers when a certain level of uncertainty
exists. Break-even refers to the point at which operations break even, where
income just equals expenses (see Fig. 10.1 for a typical break-even plot). It is
widely used for financial studies because it is simple and extracts useful insights
from a modest amount of data. In Figure 10.1, the revenue or expenses are plotted
as a function of the production rate or production capacity. The production
capacity at which the revenue line intersects the total expense line is the break-
even point. Management, of course, wants to do better than break even; therefore,
the break-even plots may be used as a profit planning tool, for product pricing,
production capacity, incremental costing of equipment, etc. The shutdown point
occurs where the revenue line crosses the fixed expense line. This says that if a
company cannot make fixed expenses, it should shut down operations. For a short
period of time a company may operate between the break-even point and the
shutdown point to maintain customers.
Break-even analysis can be used to show, for example, the effect of various
selling prices of a product on profit. Figure 10.2 is such a plot. Other possible uses
may be to study fixed and variable expenses and production level scenarios.
TM
Sensitivity and Uncertainty Analysis 247
FIGURE 10.1 Typical break-even plot.
The slopes at any given point of the total expense and total revenue curves
measure the marginal cost (MC) and the marginal revenue (MR), respectively. A
line drawn from the origin of a break-even plot to any point on the revenue curve
will have a slope equal to the average selling price (total $/units sold). A line
drawn from the origin to any point on the expense curve will have a slope equal to
the average unit expense (total $/unit produced). If a vertical line intersects both
curves at points where the slopes of the two curves are equal, MR equals MC and
profits are at a maximum.
TM
248 Chapter 10
FIGURE 10.2 Break-even plot for pricing alternatives.
In Figures 10.1 and 10.2, straight lines represent revenue and expenses,
whereas in actual practice, these might be curves. Figure 10.3 is an example of a
more realistic break-even plot.
10.1.2 The Strauss Plot

Richard Strauss [2] developed a method for plotting the results of a sensitivity
analysis in which the ordinate is the measure of profitability and the abscissa is
TM
FIGURE 10.3 A realistic break-even plot.
the change in a variable greater or less than the base case. Where the abscissa
crosses the ordinate is the base case value of the return, net present worth, net
annual worth, etc. (see Fig. 10.4). This plot is also known as the “spider” plot due
to its shape.
The slope of the line on this plot is the degree of change in the profitability
resulting from a percentage change in a variable such as selling price or
investment. The length of the line represents the sensitivity of the variable and its
degree of uncertainty. The plot has been designed so that positive slopes are
associated with variable related to income like selling price, sales volume, yields,
market share. Negative slopes are related to cost or expense items such as
investment, fixed and variable expenses.
In preparing a sensitivity plot, it is advisable to have the base case on a
spreadsheet. Then one can determine the effect on profitability by varying the
items in the study. One should be aware that variables, such as sales volume
are reflected in operating expenses as raw material requirements, general
TM
250 Chapter 10
FIGURE 10.4 A typical Strauss plot.
overhead expenses, etc. Numerous scenarios may be prepared from the basic
spreadsheet. A plot then may be developed from the spreadsheet results. A
sensitivity analysis of the purchase of wire-line trucks and the resulting sales
scenarios for a petroleum production application are found in Table 10.1 and
Figure 10.5.
10.1.3 Relative Profitability Plot

Agarwal and Klumpar [3] developed a method of graphically presenting the
results of a sensitivity study. The first step is to solve a base case to determine the
profitability. Each variable’s affect on profitability is calculated and indexed to
the same variable and the profitability of the base case. Figure 10.6 is an example
TM
TABLE 10.1 Results of Sensitivity Analysis—Wire-Line Trucksa
% Change Decrease in DCFROR

Variable percentage points
Sales þ 10 þ 2.4
volume 2 10 2 2.6
Sales þ 10 þ 10.2
price 2 10 2 10.0
Capital þ 10 2 3.7
investment 2 10 þ 3.7
a
If decision is delayed 1 year, there will be a 6.3% decrease in the DCFROR.
Base Case DCFROR ¼ 28.2%.
FIGURE 10.5 Strauss plot for wire-line truck study.
TM
252 Chapter 10
FIGURE 10.6 Relative profitability plot.
of a relative profitability plot. A 10% decrease in investment cost produces a 1.10

relative profitability or approximately a 10% increase. A 15% increase in selling
price will yield a 25% in relative profitability.
The relative profitability plot clearly represents the affect of variables on
the measure of profitability, but the method does require an additional step of
indexing variables to a base case.
10.1.4 Tornado Plot

A popular graphical presentation of profitability is the so-called “tornado” or
“Christmas tree” plot. This method derives its name from the shape of the
resulting plot (see Fig. 10.7). A commercial computer program, @ Riskw, may be
used to develop a tornado plot [4]. As in the previous methods, a base case must
be solved first. Usually the net present worth method of expressing profitability is
used. The base case is represented as a vertical line and variations in the elements
that affect the study above and below the base case are solved and plotted. From
Figure 10.7 the changes in sales volume, selling price, raw materials, capital
investment, fixed and variable expenses are plotted. It is easy to observe that
the sales volume and selling price have the greatest effect on the profitability of
project. Note the shape of the resulting curve.
10.1.5 Discussion of Sensitivity Analyses

Although sensitivity analyses are easy to prepare and they yield some useful
information for managers, there is a serious disadvantage to their use. Only one
TM
FIGURE 10.7 Typical tornado plot; interest shown at the base case.
variable at a time can be studied. Interrelated variables such as fixed capital

investment, maintenance, and other investment items in operating expenses
and their effect upon one another cannot be represented correctly. Synergistic
effects particularly with respect to marketing variables such as sale volume,
selling price, market share cannot be taken onto account.
10.2 UNCERTAINTY ANALYSIS

Investment in new ventures involves an element of risk no matter how well
defined costs and markets may be. What makes the process of proceeding with
TM
254 Chapter 10
a venture difficult is not the problem of projecting the profitability based upon a
given set of assumptions; the problem lies in the assumptions made and their
impact on profitability. Each assumption has its own degree of uncertainty, and
when taken together these uncertainties can result in large potential errors.
The objective of performing an uncertainty analysis is to obtain better, more
reliable decisions.
There is a need to clarify two terms, risk and uncertainty. Uncertainty is
exactly what the work means—not certain. An example is the price of crude oil. If
crude oil price provides a basis for a project forecast, this subjective, geopolitical
variable is indeed uncertain. Most input data to a study bear a degree of uncertainty.
Risk seems to imply that the probability of achieving a specific outcome is known
within certain confidence limits. Mortality rates, coin tosses, business interruptions,
and fire insurance premiums all fall into this category of events.
Uncertainty analysis, if understood and performed with reasonable limits,
can help the executive to sharpen his or her decisions by providing some
measurement of risks. Uncertainty analyses can be performed with varying
degrees of sophistication, but there is a trade-off of cost versus benefits obtained
that must be carefully weighed.
10.2.1 Analysis of Risk

Up to this point, all values in calculations such as investment, operating expenses,
cash flows have been treated as if they were correct or deterministic. From these
data, decisions under certainty can be made. Economic analysis, however, requires
the estimation of values sometime in the future. The farther into the future we
venture, the greater the chance of error. Any number is selected with the best
chance of being correct, but there is always the probability that the estimate will be
above ore below the actual value. Although we prefer to deal with certainty, it
seldom exists. To allow for a range of values around the average or best value,
statistical decision theory is employed but that is beyond the scope of this text.
Although important, only concepts useful in an applied manner will be treated.
10.2.2 Methods
10.2.2.1 Best Guess
One of the most common methods of evaluating projects involves the use of a
single point, or best guess, estimate of a value for each important factor that
might affect an investment decision. This is the starting point for any uncertainty
analysis. Ross [5] wrote a classic article that still provides the foundation for the
best guess, range, and Monte Carlo analysis.
What information can be gained from such an analysis? A project based
upon the best guess may show a 20% return, but how is that figure to be
TM
interpreted? Does this mean the company can expect a 20% return? Perhaps that
might be the case if each factor is exactly correct. What are the chances that the
return might be better than 20%? What are the chances that money might be lost
on this project? There is no way to answer these questions because the
uncertainty of each factor has not been considered. This brings us to the range
approach.
10.2.2.2 Range Approach

To remedy these shortcomings, a range of values may be tried. One estimates
the most optimistic and the most pessimistic values for each factor. Returns are
calculated for each scenario but the important questions posed in the previous
section still remain unanswered. Ranges by themselves are not effective in all
but the simplest cases. When combining a set of estimates based upon
range values, the financial yardstick computed becomes very wide and is
uninformative.
What is needed in the decision-making process is some way to quantify
uncertainties in each assumption made, along with a calculation technique that
will yield a curve for each return and its uncertainty. This approach will give the
manager a far more reliable estimate of risks involved in a project investment and
will provide a realistic basis for decisions. Therefore, this leads to the Monte
Carlo technique.
10.2.2.3 Monte Carlo Method

From the previous methods in this chapter, it is apparent that they have
shortcomings. In the Monte Carlo simulation, each variable is represented by a
probability distribution model. This requires the gathering of enough data to
develop a reasonable probability model. There are probability models of various
shapes, for example, normal, Gaussian, Poisson, beta. Not all variables in the
simulation follow the normal distribution curve. If that were the case, the final
profitability model would also be a normal distribution. The normal distribution
curve may be used to represent sales-related items, and it has been found that the
beta distribution is a good model for capital investment estimates. Conceivably,
several people might be involved in gathering the data and this could introduce
further biases and inconsistencies.
Figure 10.8 is schematic of the Monte Carlo procedure [5]. In the figure,
the steps are numbered, the first step being to determine the probability
distribution for each of the variables in the simulation. The next step is by
means of a random number generator, random values are selected from the
probability models and a profitability measure for each set of values is
TM
256 Chapter 10
FIGURE 10.8 Schematic diagram of the Monte Carlo procedure. (From Ref. 5.)
determined. This process is repeated until a plot of return versus the probability
of that return being achieved is obtained (see Fig. 10.9). In simple terms, this is
the way the Monte Carlo simulation operates. After the analysis has been
performed, the next task is to interpret the results. The management must
clearly understand clearly what the results mean and the reliability of the results.
Experience can only be obtained by performing uncertainty analyses. One
or two attempts will not permit management to gain confidence in this
decision-making tool [6]. The bottom line is: Are the stakes high enough and
TM
FIGURE 10.9 Typical uncertainty analysis plot.
the benefits great enough to spend the time and money on this sophisticated
method?
Software companies market programs that permit the user to perform
probability distribution models and Monte Carlo simulations, like SAS and
@RISK.
An example of a simplified Monte Carlo simulation is found in the
following example. This will be a three-point simulation of a capital invest-
ment study for capital investment. C. A. Miller [7] proposed a cost estimation
TM
258 Chapter 10
method using the highest, lowest, and most likely values of various costs for
the simulation.
Example 10.1
A company is involved in specialty chemical manufacturing. A new chemical
is to be produced and flowsheets and an equipment list has been prepared for
a pilot plant. A deterministic estimate of capital requirements based upon
equipment cost estimates and best guesses for the factors in the Miller method [7]
predicts a total investment of about $7.8MM.
To use the probabilistic approach, low, best guess, and high values are
specified for the input variables. The high and low values were selected based
upon the estimator’s judgment and from scatter diagrams that correlate the
factored estimates with actual plant construction. The appropriate ranges for
equipment costs, buildings, auxiliary equipment depend upon the reliability of
the cost correlations, the precision of equipment sizing, and many other
considerations. Figure 10.10 illustrates the input distribution for f1 an overall
contingency factor. Using computer graphics software, plotting output
distribution curves for the estimated investment is possible. Simple
FIGURE 10.10 Six-point contingency distribution for Example 10.1.
TM
FIGURE 10.11 Cumulative probability for Example 10.1.
assumptions concerning the location of endpoint values and cumulative

probability data are obtained from Monte Carlo trials.
Cumulative probability and probability distribution curves are shown in
Figures 10.11 and 10.12. A number of quantitative statements can be made; for
example, from Figure 10.12, the most probable value for the pilot plant
investment is $7.9MM and the mean value is $8.8MM. The 10 and 90%
probabilities are $6.3MM and $13.0MM, respectively, in Figure 10.11. Further,
there is a 95% probability will be between $5MM and $15MM and a 35%
probability that the investment will exceed $10MM [8].
The advantages of probabilistic analysis for factored capital investment
estimates are maximum utilization of available information and the ability to
quantify uncertainty.
The incentives for performing an uncertainty analysis are greatest when
the size of the investment is large and when the decision maker has no way
to modify or reverse his or her decisions once construction has started.
Judgment is not replaced by uncertainty analysis. The decision maker is,
TM
260 Chapter 10
FIGURE 10.12 Probability distribution for Example 10.1.
however, supplied with a means to determine the probability of the risk the
manager is taking.
One major problem is the understanding of how uncertainty analysis is
to be used. There is an educational problem. People have to be trained in
the development of probabilistic models and managers would have to learn
to interpret the results. Uncertainty analysis is another tool to help managers
become more effective in their role of guiding the fortunes of a company [5].
REFERENCES
1. JR Couper, WH Rader. Applied Finance and Economic Analysis for Scientists and
Engineers. New York: Van Nostrand Reinhold, 1986.
2. R Strauss. Chem Eng 112 – 116, March 25, 1968.
3. JC Agarwal, IV Klumpar. Chem Eng 66 – 72, September 29, 1975.
4. @RISKw, Palisade Corporation, Newfield, NY, 2002.
5. RC Ross. Chem Eng 149 – 215, September 20, 1971.
TM
6. DB Hertz. Harv Business Rev 46(1):96 – 108, January –February, 1968.

7. CA Miller. Chem Eng 226 – 237, September 13, 1965.
8. Private communication, 1997.
PROBLEMS
10.1 Rustic Wood Products manufactures a line of children’s toys. The
following is an analysis of their accounting data:
Fixed costs $90,000 per year

Variable costs $5.00 per toy
Capacity 25,000 toys per year
Selling price $18.50 per toy
a. Compute the break-even point in number of toys.

b. Find the number of toys the company must sell to show a profit of
$40,000 per year.
c. At 75% of capacity, what is the fixed cost per toy? At that capacity what
are the variable cost per toy?
d. What is the shutdown point?
10.2 A producer of electronic equipment is considering the installation of one

of two types of manufacturing machinery. A long-term sales forecast indicates
that the sales will not fall below 8200 units per year for the next 5 years, the
expected life of each machine. Machine 1 will increase fixed costs by $20,000 per
year but reduce variable costs by $6 per unit produced. Machine 2 will increase
the fixed costs by $4000 per year but will reduce the variable costs by $4 per unit.
Variable costs now amount to $20 per unit. At what point are you indifferent as to
which machine to purchase? Which machine should be bought?
10.3 Plastics Projects, Inc. considered the manufacture of a plasticizer
for potential customers in the plastics industry. At that time, a market survey was
made but the results did not appear promising. The project was shelved.
Four years later, the company was acquired by Fusible Plastics Corporation,
a large plastics manufacturer which wanted to integrate backwards. Since the
acquisition, the surviving company has been reviewing the profitability of all
projects of the former company. The marketing department of Fusible Plastics has
prepared a 10-year market survey for the product:
TM
262 Chapter 10
Potential sales, Estimated sales price,

Year MM lb/yr $/lb
1 20.0 0.52
2 20.5 0.52
3 21.0 0.51
4 22.0 0.50
5 23.0 0.50
6 –10 25.0 0.48
Four years before the acquisition, the total fixed capital investment for the
hydrocarbon was estimated to be $5MM for a 20 MM lb/yr plant. The cost index
to update this investment is 12% more over the 4-year period. Working capital
may be taken at 15% of the total capital investment. For purpose of this
evaluation, land may be ignored.
To manufacture the plasticizer, raw material A is required and it may be
obtained from Plastics Products, Inc. plant. The requirements of this raw material
are 5% of PP’s capacity and the total capital investment of that plant is $10MM.
The operating expenses for the proposed plasticizer plant have been
estimated as follows:
Raw materials $0.12/lb product

$0.05/lb product up through
Labor and supervision 21 MM lb/yr and $0.04/lb product
above this rate
Maintenance 6%/yr of the fixed capital investment
Utilities $0.03/lb product up through 21 MM b/yr
and $0.02/lb product above this rate
Other direct expenses $0.01/lb product
Depreciation 7-year straight-line
Other indirect expenses $0.01/lb product
General overhead expenses $0.02/lb product
At the screening stage, unless a proposed project yields a rate of return of

20% calculated by the DCF method, the venture will not be considered further. A
federal income tax rate of 35% may be assumed. The fixed capital investment
will be purchased and installed uniformly over a 2-year period prior to start-up
that is expected to be from 2 2 years to 0 before sales begin.
TM
In order to conform to the standard evaluation procedure of Fusible

Plastics, please prepare the following:
a. An estimate of the total capital requirements for a 25 MM lb/yr plant
updating the costs to the start of year 1
b. An annual operating expense sheet using the data above and assuming
that the potential sales in each year is manufactured and sold
c. An estimate of the profitability using the DCFROR Method
d. Prepare a Strauss sensitivity plot to determine the effect of the following
variables on the return:
1. ^ 10%, ^ 20% variation in sales price.
2. ^ 10%, ^ 20% variation in sales volume.
3. ^ 5%, ^ 15% variation in raw material expenses.
4. ^ 10%, ^ 20% in fixed capital investment.
5. Based upon your calculations, what recommendations would you
make to management about this proposed venture?
10.4 Using the data in Problem 10.3, prepare an uncertainty analysis for the
proposed venture. Probabilities for sales related items may be by normal
distribution, capital investment by beta distribution, and raw material costs by
Poisson distribution. What is the probability of achieving a 20% DCF return?
TM

DK5739 CH10

Uploaded by

Copyright:

Available Formats

DK5739 CH10

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

DK5739 CH10

Uploaded by

Copyright:

Available Formats

10

Sensitivity and Uncertainty Analysis

In Chapter 9, it was assumed that all values of costs, expenses, revenues,

10.1 SENSITIVITY ANALYSIS

10.1.1 Break-Even Analysis

FIGURE 10.1 Typical break-even plot.

FIGURE 10.2 Break-even plot for pricing alternatives.

10.1.2 The Strauss Plot

FIGURE 10.3 A realistic break-even plot.

FIGURE 10.4 A typical Strauss plot.

10.1.3 Relative Proﬁtability Plot

TABLE 10.1 Results of Sensitivity Analysis—Wire-Line Trucksa

% Change Decrease in DCFROR

FIGURE 10.5 Strauss plot for wire-line truck study.

FIGURE 10.6 Relative proﬁtability plot.

of a relative proﬁtability plot. A 10% decrease in investment cost produces a 1.10

10.1.4 Tornado Plot

10.1.5 Discussion of Sensitivity Analyses

variable at a time can be studied. Interrelated variables such as ﬁxed capital

10.2 UNCERTAINTY ANALYSIS

10.2.1 Analysis of Risk

10.2.2.2 Range Approach

10.2.2.3 Monte Carlo Method

FIGURE 10.9 Typical uncertainty analysis plot.

FIGURE 10.10 Six-point contingency distribution for Example 10.1.

FIGURE 10.11 Cumulative probability for Example 10.1.

assumptions concerning the location of endpoint values and cumulative

FIGURE 10.12 Probability distribution for Example 10.1.

6. DB Hertz. Harv Business Rev 46(1):96 – 108, January –February, 1968.

Fixed costs $90,000 per year

a. Compute the break-even point in number of toys.

10.2 A producer of electronic equipment is considering the installation of one

Potential sales, Estimated sales price,

Raw materials $0.12/lb product

At the screening stage, unless a proposed project yields a rate of return of

In order to conform to the standard evaluation procedure of Fusible

You might also like