04 Decision Analysis Part2-2 PDF
04 Decision Analysis Part2-2 PDF
04 Decision Analysis Part2-2 PDF
Topics Outline
Developing a Decision Model
Using the Precision Tree Add-In
Developing a Decision Model
SciTools Bidding
SciTools Incorporated, a company that specializes in scientific instruments, has been invited to
make a bid on a government contract. The contract calls for a specific number of these
instruments to be delivered during the coming year. The bids must be sealed, so that no company
knows what the others are bidding, and the low bid wins the contract.
SciTools estimates that it will cost $5000 to prepare a bid and $95,000 to supply the instruments
if it wins the contract. On the basis of past contracts of this type, SciTools believes that the
possible low bids from the competition, if there is any competition, and the associated
probabilities are those shown in following table.
Low Bid
Probability
0.2
0.4
0.3
0.1
In addition, SciTools believes there is a 30% chance that there will be no competing bids.
Objective: To develop a decision model that finds the EMV for various bidding strategies and
indicates the best bidding strategy.
Solution:
1. Describe the decisions available to SciTools.
SciTools has two basic strategies: submit a bid or do not submit a bid. If SciTools submits a bid,
then it must decide how much to bid. Based on the cost to SciTools to prepare the bid and supply
the instruments, there is clearly no point in bidding less than $100,000 SciTools wouldnt make
a profit even if it won the bid. (Actually, this isnt totally true. Looking ahead to future contracts,
SciTools might make a low bid just to get in the game and gain experience. However, we
wont consider such a possibility here.) Although any bid amount over $100,000 might be
considered, the data in the table above suggest that SciTools might limit its choices to $115,000,
$120,000, and $125,000.
-1-
Monetary Value
SciTools Wins
NA
15K
20K
25K
SciTools Loses
0
5
5
5
1.00
0.14
0.42
0.63
The rightmost columns show the probabilities that SciTools wins or loses the bid for each possible
decision. For example, if SciTools bids $120,000, then it wins the bid if there are no competing bids
(probability 0.3) or if there are competing bids (probability 0.7) and the lowest of these is greater
than $120,000 (probability 0.3 + 0.1). Therefore, the total probability that SciTools wins the bid is
0.3 + 0.7(0.3 + 0.1) = 0.58
Then, the probability that SciTools loses the bid is 1 0.58 = 0.42.
4. Develop a risk profile for SciToolss decision to bid $120,000.
If SciTools bids $120,000, there are two monetary values possible, a profit of $20,000 and a loss
of $5000, and their probabilities are 0.58 and 0.42, respectively. The corresponding risk profile is
a spike chart with two spikes, one above $5000 with height 0.42 and one above $20,000 with
height 0.58.
-2-
EMV Calculation
EMV
No bid
$115,000
$120,000
0(1)
15,000(0.86) + (5,000)(0.14)
$0
$12,200
20,000(0.58) + (5,000)(0.42)
25,000(0.37) + (5,000)(0.63)
$9,500
$6,100
$125,000
The company first makes a bidding decision, then observes what the competition bids, if anything,
and finally receives a payoff. The folding-back process is equivalent to the calculations of EMVs in (6).
There are often equivalent ways to structure a decision tree. One alternative for this example is
shown below.
Otherwise, the company then decides how much to bid. Note that if the company decides to bid,
it incurs a sure cost of $5000, so this cost is placed under the Bid branch.
Once the company decides how much to bid, it then observes whether there is any competition.
If there isn't any, the company wins the bid for sure and makes a corresponding profit.
Otherwise, if there is competition, the company eventually discovers whether it wins or loses the
bid, with the corresponding probabilities and payoffs.
Note that these payoffs are placed below the branches where they occur in time. Also, the
cumulative payoffs are placed at the ends of the branches. Each cumulative payoff is the sum of all
payoffs on branches that lead to that end node.
The folding-back procedure is somewhat more complex than it was for the smaller tree.
To illustrate, the EMVs above a selected few of these nodes are calculated as follows:
Node 7: EMV = 20000(0.40) + (5000)(0.60) = $5000 (uses monetary values from end nodes)
Node 4: EMV = 20000(0.30) + (5000)(0.70) = $9500 (uses monetary value from an end node
and the EMV from node 7)
Node 2: EMV = max(12200, 9500, 6100) = $12,200 (uses EMVs from nodes 3, 4, and 5)
Node 1: EMV = max(0, 12200) = $12,200 (uses monetary value from an end node and EMV
from node 2)
The results are the same, regardless of whether you use the table of EMVs developed in (6), the
smaller decision tree, or the larger one, because they all calculate the same EMVs in equivalent ways.
10. Does the decision made with the EMV criterion guarantee a good outcome for the company?
Acoording to the EMV criterion, the company should bid $115,000, with a resulting EMV of
$12,200. This decision is not guaranteed to produce a good outcome for the company.
For example, the competition could bid less than $115,000, in which case SciTools would lose $5000.
Alternatively, the competition could bid more than $120,000, in which case SciTools would be
kicking itself for not bidding $120,000 and gaining an extra $5000 in profit.
Unfortunately, in problems with uncertainty, there is virtually never a guarantee that the optimal
decision will produce the best result. The only guarantee is that the EMV-maximizing decision is
the most rational decision, given the information known when the decision must be made.
-5-
11. Use the following spreadsheet model to describe how the optimal EMV changes as the
probability of no competing bid changes. (See SciTools_Bidding_Spreadsheet_Model.xlsx.)
In the data table for sensitivity analysis, the probability of no competing bid is allowed to vary
from 0.2 to 0.7. The data table shows that the optimal EMV increases over this range from
$12,200 to $16,900.
The third column of the data table shows that the $115,000 bid is optimal for small values of the
input, but that a $125,000 bid becomes optimal for larger values.
-6-
If you want to unload PrecisionTree without closing Excel, you can do so from its Utilities
dropdown list in the Tools group.
2. Build the Decision Tree
Use the file SciTools_Bidding.xlsx to build your tree.
Use the file SciTools_Bidding_Finished.xlsx as a reference.
Click on the Decision Tree button on the PrecisionTree ribbon, and then select cell A14 below
the input section to start a new tree. In the dialog box, enter a name for the tree, e.g. SciTools
Bidding, and click on OK. The beginnings of a tree appears:
To obtain decision nodes and branches, select the (only) triangle end node to open the dialog box.
Click on the green square to indicate that you want a decision node, and fill in the dialog box as
shown below.
-7-
Then click on the Branches (2) tab and supply labels for the branches under Name
(No for Branch #1 and Yes for Branch #2).
Click OK. The tree expands. Under the Yes branch, enter the following link to the bid cost cell:
= B4 (Note: It is negative to reflect a cost.)
The top branch is completed; if SciTools does not bid, there is nothing left to do.
So click on the bottom end node (the triangle), following SciTools's decision to bid, and proceed
as in the previous step to add and label the decision node and three decision branches for the
amount to bid. (By default, you get two branches. Click on Add to get additional branches.)
The tree to this point should appear as follows.
-8-
Note that there are no monetary values below these decision branches because no immediate
payoffs or costs are associated with the bid amount decision.
Using the same procedure (and keeping in mind the finished tree you eventually want),
create probability (chance) nodes extending from the bid $115,000 decision.
You should have the skeleton shown below.
-9-
The Policy Suggestion shows only the subtree corresponding to the optimal decision strategy:
- 10 -
As the risk profile indicates, there are only two possible monetary outcomes if SciTools bids
$115,000. It either wins $15,000 or loses $5000, and the former is much more likely.
(The associated probabilities are 0.86 and 0.14, respectively.)
- 11 -
It has a lot of options and it takes some practice and experimenting to get used to them.
Here are the main options and how to use them.
The Analysis Type dropdown list allows you to vary one input (One-Way Sensitivity)
or two inputs (Two-Way Sensitivity) simultaneously.
The Starting Node dropdown list lets you choose any node in the tree, and the sensitivity
analysis is then performed for the EMV from that node to the right. In other words, it assumes
you have gotten to that node and are now interested in what will happen from then on.
You add inputs to vary in the Inputs section. When you add an input to this section, you can
specify the range over which you want it to vary. For example, you can vary it by plus or minus 10%
in 10 steps from a selected base value, as we did for the production cost in cell B5, or you can
vary it from 0 to 0.6 in 12 steps, as we did for the probability of no competing bids in cell B7.
The Include Results checkboxes allow you to select up to four types of charts, depending on the
type of sensitivity analysis. (The bottom two options are disabled for a two-way sensitivity analysis.)
By default, the results are presented in a new workbook. If you would rather have them in the
same workbook as the model, click on the PrecisionTree Utilities dropdown arrow, select
Application Settings, and select Active Workbook from the Place Reports In option.
(This is a global setting. It will take effect for all future PrecisionTree analyses.)
- 12 -
The following chart was constructed by choosing cell C29 as the cell to analyze. This is the optimal
EMV for the problem, given that the company has decided to place a bid. The sensitivity chart
indicates how the EMV varies with the probability of no competing bid for each of the three bid
amount decisions. The $115,000 bid is best for most of the range, but when the probability of no
competing bid is sufficiently large (about 0.55), the $120,000 bid becomes best (by a small margin.)
- 13 -
For each of the possible values of production cost and the probability of no competitor bid,
this chart indicates which bid amount is optimal.
(By choosing cell C29, we are assuming SciTools will bid; the only question is how much.)
As you can see, the optimal bid amount remains $115,000 unless the production cost and the
probability of no competing bid are both large.
Then it becomes optimal to bid $120,000 or $125,000. This makes sense intuitively.
As the probability of no competing bid increases and a larger production cost must be recovered,
it seems reasonable that SciTools should increase its bid.
- 14 -
- 15 -
Notes:
1. Allowable Entries
You should enter your own probabilities and monetary values only in the cells with black font.
2. Sum of probabilities
PrecisionTree does not enforce the rule that probabilities on branches leading out of a node must
sum to 1. You must enforce this rule with appropriate formulas.
3. Entering monetary values and probabilities
A good practice is to calculate all of the monetary values and probabilities that will be needed in
the decision tree in some other area of the spreadsheet. Then the values needed next to the tree
branches can be created with simple linking formulas.
4. Folding-back procedure
Note that you do not have to perform the folding-back procedure manually. In addition, if you
change any of the inputs, the tree reacts automatically. For example, try changing the bid cost in
cell B4 from $5000 to some large value such as $20,000. You will see that the tree calculations
update automatically, and the best decision is then not to bid, with an associated EMV of $0.
5. Values at end nodes
There are two values following each triangle end node.
The bottom value is the sum of all monetary values on branches leading to this end node.
The top value is the probability of getting to this end node when the optimal strategy is used.
This explains why many of these probabilities are 0; the optimal strategy will never lead to these
end nodes.
- 16 -