PART ONE
1) What is Monte-Carlo Simulation?
Monte-Carlo
Simulation is an analytical method meant to imitate a real-life system. It
answers “what-if” questions through deterministic analyses. Monte-Carlo
Simulation randomly generates values for uncertain variables repeatedly in
order to simulate a model. For variables that have a known range of values but
an uncertain value for any particular time or event, Monte-Carlo Simulation is
particularly useful.
Monte-Carlo
Simulation can be related to sensitivity and scenario analysis. With one-way
sensitivity analysis, the most sensitive variable is at the top and variables
then descend in sensitivity. This type of analysis measures the
sensitivity/changes in the outcome due to changes in the input variable. This
shows a sensitivity range, producing a “tornado-like” appearance. N-way
sensitivity analysis makes changes in several inputs, helping with scenario
evaluation. Controllable and uncontrollable variables are input into the model
in order to predict certain outcomes. Sensitivity and scenario analyses are
limited by the fact that probability distributions are not utilized.
Monte-Carlo Simulation overcomes this obstacle by modeling uncertain variables
using probability distributions.
2) How is Monte-Carlo Simulation
performed?
Monte-Carlo
Simulation is performed by calculating multiple scenarios of a model by repeatedly
sampling values from the probability distributions for the uncertain variables
and using those values for the cell. Simply stated, a Monte-Carlo simulation
allows for hypothetical values to be input into a model. These variables allow
for the calculation of various outcomes.
For each
uncertain variable with a range of possible values, the possible values are
defined with a probability distribution. The type of distribution that is
selected is based on the conditions surrounding that variable. Distributions
can be normal, triangular, uniform, or lognormal.
I am using @Risk to perform
Monte-Carlo Simulation, an add-in for Microsoft Excel. Monte-Carlo Simulation
selects a value for an input, recalculates the outputs for the DSS, plots the
forecast, and then repeats for as many iterations as
are selected by the user.
The user selects an input
in the support system. The distribution of values must be selected by clicking
on the "Define Distributions" tab. There are a number of options
to choose from. The most common distribution is the "normal"
distribution, which looks like a bell-shaped curve. This means that when
the simulation begins, it will select a range of random values centered around the peak of the bell-shaped curve. Minimum and
maximum values can also be changed if needed. Triangle and uniform
distributions are also common.
Outputs must be
defined. First, select the output of interest in the
spreadsheet. Then, click on the "Add Output" tab. A dialogue box
appears asking for a name for the output. Usually the one given is correct.
Click on the "@Risk
Model Window" and verify that all inputs and outputs are accounted for
before simulating.
Click on
the "Simulation Settings" tab. The most important item here is the number of
iterations. Select the number of iterations desired. The more iterations that are performed, the longer Monte-Carlo
Simulation takes.
Click on "Start
Simulation"! When it is finished, you can then view multiple graph types
and statistics that show the results of the simulation.
3) Why is Monte-Carlo Simulation
appropriate for managing uncertainty in complex situations?
Monte-Carlo
Simulation takes the guessing out of business situations, while allowing
decision-makers to choose the most promising calculated risk. Monte-Carlo
Simulation emphasizes optimization as a process that finds the best solution.
When variables exist that can be controlled, and you want a maximum or minimum
goal that relies on those variables, Monte-Carlo Simulation is ideal.
PART TWO
Inputs
Price
- input variable has normal distribution with a mean of $377.78 and SD of
6.1032
Advertising - cost input variable has normal distribution with a mean of
$93325.78 and SD of 9828.045
Outputs
Market Share
Relative Demand
Summary Statistics
In
this simulation model I used #iterations as 4000 and #simulation as 1. The
table below summarizes the quality of data that was used and the output
generated off of the simulation model. The minimum, maximum,
range and the kind of distribution of the inputs can be controlled in the
simulation.
|
Variables |
Cell |
Minimum |
Mean |
Maximum |
|
Relative Demand |
G7 |
-5.35E-02 |
0.9980822 |
1.993445 |
|
Market Share |
G9 |
-5.35E-03 |
9.98E-02 |
0.1993445 |
|
Price |
B12 |
356.0646 |
377.7847 |
401.882 |
|
Advertising |
B13 |
58808.89 |
93329.42 |
142045.1 |
Diagrams
Graphs below show the normal distribution of the outputs and
inputs in the simulation. They include both the histogram and fitting
distribution diagrams.









The distribution of the outcome can be
very critical to businesses when making important decisions like launching a
new product or for increasing market share. Monte-Carlo simulation helps
companies get a feel of a near real-life situation before making any capital
investments.
Application
We have seen above how Monte-Carlo
simulation can be effectively used to study the effect of uncertainty in the
inputs on the distribution of the output variables. Businesses can plan their
entire operations based on the expected demand and plan their revenue targets
and strive to achieve them.
Conclusion
One of the limitations of the human
mind is that it can process only one piece of information at a time. As models
get increasingly complex, it is difficult to manually factor in the uncertainty
of inputs and their effect on the distribution of the outcomes. Monte-Carlo
simulation helps us factor in the uncertainty in the inputs using their
probability distributions and repeat it for a large number of iterations.