Decision Analysis - PebblePad

Decision Analysis
On 2014, Steve Blake was proposed the alternative of making a venture to produce electronic components
for home use. Steve Blake has received some information which might help him decide if it profitable or not
to incur in the venture, but he is not sure if he needs more information to make a proper decision. The
following report analyses the expected outcomes that might occur given the available information.
We developed a decision tree and analyzed the proposals and the cost and benefits of obtaining more
information with the following objective:
To advice Steve Blake on the best course of action for his potential business venture.
ASSUMPTIONS





The cost of production are fixed at $500,000
The expected revenue is fixed at $2,000,000
The probabilities of occurrence of each outcome have a high degree of realism
The personal research team is not incurring any cost
The information from Marketing Associates Inc. (MAI) and Iverstine and Kinard (I&K) is sample
information
CALCULATIONS
To analyze the information from MAI and I&K in similar ways, we calculated the probabilities of each outcome
using the Bayesian Theorem. The final probabilities for each outcome are shown in table 1. The calculation
steps to obtain this values can be found on Appendix A.
Personal
Research
Favorable Unfavorable Favorable Unfavorable
Total
MAI
Outcome
I&K
Successful
Venture
70%
40%
87.0968%
15.7895%
60%
Unsuccessful
Venture
30%
60%
12.9032%
84.2105%
40%
Table 1. Probabilities
The data from the personal research team and from I&K and MAI combined with the above probabilities
displayed the payoffs shown in table 2.
PAYOFF TABLE
Alternatives
MAI
I&K
Personal Research
OUTCOMES
Favorable
Make
No Make
Successful
Unsuccessful
$
$
$
1,400,000
(600,000)
(100,000)
$
$
$
1,200,000
(800,000)
(300,000)
$
$
$
1,500,000
(500,000)
0
Table 2. Payoff Table
2
Unfavorable
Make
No Make
Successful Unsuccessful
$
$
$
1,400,000
(600,000)
(100,000)
$
$
$
1,200,000
(800,000)
(800,000)
N/A
N/A
N/A
RESULTS
To make a proper decision on whether which research company to choose (if any) we analyzed the data from
the model by different criteria: Maximax, Maximin, Minimax Regret and Expected Monetary Value.
The expected payoffs, EMV and probabilities can be found in a Decision Tree in the attached Excel File.
Maximax
Balakrishnan et al. (2013) describe Maximax as “the criterion selects the decision alternative that maximizes
the maximum payoff over all alternatives”
Maximax
Decision
(pp.323). The decision that maximizes the
Max Payoff
Choice
maximum payoff is to not incur a survey. The
Survey with MAI
$
1,400,000
maximum payoff is of $ 1.5 Million. Table 3
Survey with I&K
$
1,200,000
shows the maximum payoff for the other No survey/personal research $
1,500,000
Best
alternatives.
Table 3. Maximax Table
Maximin
The Maximin criterion selects the decision alternative that maximizes the
alternatives. The best decision is not incur a
Decision
survey. The $500,000 loss is the maximum of
the minimum payoffs, i.e. is the highest value
Survey with MAI
among the minimum payoffs, therefore it
Survey with I&K
shows the best, worse outcome. Table 4
No survey/personal research
summarizes the results.
minimum payoff over all
Maximin
Min Payoff
Choice
$
(600,000)
$
(800,000)
$
(500,000)
Best
Table 4. Maximin Table
Minimax Regret
The Minimax regret criterion selects the decision alternative that minimizes the maximum opportunity loss
within each alternative. The decision that
Minimax
Decision
minimizes loss is not to incur a survey. This
Max Regret
Choice
regret of $0 is the minimum of the maximum
Survey with MAI
$
100,000
regrets, which means that the opportunity loss
Survey with I&K
$
700,000
is minimized. Table 5 show the maximum
No survey/personal research $
Best
opportunity los for each alternative.
The calculation steps for this table can be found on Appendix B.
3
Table 5. Minimax Table
Expected Monetary Value (EMV)
The EMV is the weighted average of all possible payoffs for each alternative. The best decision is usually the
one with the highest EMV, in this case not to
Decision
EMV
Choice
incur a survey. Table 6 shows the EMV for each
alternative. The $700,000 from this decision
Survey with MAI
$ 500,000.00
means than in the long run this will be the
Survey with I&K
$ 445,161.29
average payoff of this alternative. The actual
No survey/personal research $ 700,000.00
Best
payoffs are the ones displayed on table 2.
Table 6. EMV
INSIGHTS
The four analysis above, provided a clear view of which should be the best decision to make towards hiring a
research company or not. They showed that the best decision is not to hire any company and use the data
from the personal research team. However it is important to note that this results are given assuming sample
information from each of the marketing firms. If for some reason they could provide perfect (or very accurate)
information about the future outcomes the optimal decision for Blake might. However, we did not have
enough information about probability figures so it was not possible to measure the Expected Value with
Perfect Information (EVwPI) nor the Expected Value of Perfect Information (EVPI).
We were provided with sample information from I&K, so we decided to calculate the Expected Value of
Sample Information (EVSI) from I&K. The results showed that their EVSI is $45,161. This means that the
information from I&K is only worth this amount of money, and paying any price above $45,161 is not
economically efficient. The price I&K want for their
Decision
EVSI
information is $ 300,000, six times higher than what it is
Survey with I&K
$ 45,161.29
actually worth, so Steve should not request an analysis
No survey/personal research
N/A
from this company or more information at this cost.
Table 7. EVSI for I&K
4
CONCLUSIONS
The information given about the owner, corresponds to the profile of risk avoider. He tries to make the most
accurate decisions based on the size of the business and the likelihood of successfully delivering his promises.
The current business decision lies in whether to hire a research company or not, to be able to determine an
outcome in the future and whether or not it is viable to make a venture to produce new electronic items.
The previous analysis showed that under optimistic and pessimistic approaches, and measuring the payoff
after risk and calculating the expected values of each alternative, the overall best decision is not to incur a
survey. Only if we were able to obtain perfect information from the surveys and measure the probability of
occurrence of each outcome with accuracy, our decision might change, but with the given information is it
safe to assume that the best decision is not to hire I&K or MAI to do market researches.
The option of not hiring any company provides the highest Expected Monetary Value, and the minimum
opportunity loss, which makes this option very profitable but secure as well. This decision has the potential
to satisfy the two main stakeholders of the company, both the conservative investor and the aggressive one.
5
APPENDIX A – Derivation of probabilities for I&K using Bayes Theorem
Favorable Survey (FS)
Conditional Prior Joint
Successful
Venture
Unsuccessful
Venture
FS
90%
60%
54%
87.0968%
20%
40%
8%
62%
12.9032%
Unfavorable Survey (US)
Conditional Prior Joint
Successful
Venture
Unsuccessful
Venture
US
10%
60%
6%
15.7895%
80%
40%
32%
38%
84.2105%
I&K
Favorable Unfavorable
Successful
Venture
87.0968%
15.7895%
Unsuccessful
Venture
12.9032%
84.2105%
6
APPENDIX B – Minimax Regret Calculation
REGRET
ALTERNATIVES
MAI
I&K
Personal
Research
Outcomes
Favorable
Make
No Make
Successful
Unsuccessful
$
$
$
100,000.00
100,000.00
100,000.00
$
$
$
300,000.00
300,000.00
300,000.00
$
$
$
-
Unfavorable
Make
No Make
Successful
Unsuccessful
$
$
$
$
$
$
200,000.00
200,000.00
700,000.00
N/A
N/A
N/A
Minimax
Max Regret
Choice
$ 100,000.00
$ 700,000.00
$
-
Best