ESE204 (Fall 2017) Decision Models

Rakesh Vohra

This version: September 27, 2017

Description
We all use models to arrive at decisions. They range in sophistication from the
informal “everyone but me is a fool” to the full rationality assumed in financial
markets. The informal models we use often lead us astray. This course is about how
to make good models as well as how to distinguish a good model from a bad one.
In the study of models and how they relate to decision making we seek the middle
ground between the perspective of the parachutist and that of the truffle hunter.
We focus on two challenges. The first with the allocation of scarce resources (labor,
material, time) between a variety of competing activities. The second is evaluating
choices in the presence of uncertainty and limited information. These challenges
are best handled quantitatively, and so it will be useful to summarize and dispose
of the usual objections against doing so.
• Objection #1: Not everything can be reduced to numbers.

True. But a great deal of importance can.

• Objection #2: Workable Quantitative models cannot capture the complexi-
ties of real life.

So what? The question is not whether a particular quantitative model ac-

curately represents reality but whether there is an alternative model that is
more accurate. How do we know that a decision arrived at by what we are
pleased to call instinct (or intuition , experience, etc.) has acknowledged all
the complexities of reality? Indeed, one of the beauties of quantitative models
is their explicitness. Like Cromwell’s portrait they appear warts and all.
• Objection #3: The data requirements of quantitative models are prohibitive.

This objection had some merit when dinosaurs roamed the earth. Given
the state of modern computing, it is an excuse for laziness. One of the use-
ful features of a quantitative model is that it tells the decision maker what
information s/he should be collecting.

1
 • Objection #4: Quantitative models ignore the context.

Yes. However, one is not obliged to follow the recommendations of a quanti-

tative model that is not appropriate to the context. Second, ignoring the con-
text can be a good thing when the context is irrelevant. A quantitative model
often captures the essence of a decision, say, the choice between a gamble and
a sure thing. It does not matter whether the context is the stock market, a
medical diagnosis or wildcatting. This allows for the transfer of insight from
one setting to another. An example of this is the use by hotels of the same
models and principles that airlines use for yield management. A more reveal-
ing example comes from experiments on auction behavior. The subjects were
managers responsible for submitting bids on construction projects. The goal
was to see whether these managers made bids in accordance with the opti-
mal bids of the standard (quantitative models) auction model. As long as the
auctions were couched in the context of the construction industry, essentially,
yes. However, when they were presented with same auction scenarios, but
the context changed (to say oil leases) they did remarkable badly. Why? The
bids they made were arrived at through the use of rules of thumb that were
specific to their industry. When the context switched, their rules of thumb
became useless.
Perhaps the most surprising thing that you will learn from this course is that
simple quantitative models frequently provide profound qualitative insights into
the process being modeled. Thus, one may wish to build a model not so much to
decide what to do, but to test and refine one’s intuition about what is going on.

Grading
Most homework assignments are to be done individually. There is one group
assignment and it is labeled accordingly. The class will be divided into groups,
and you should do the group homework assignments with your assigned group (one
submission per group).
The final grade will depend on
• Homework (Individual & Group) (15 %)
• 3 Exams (15%, 15% 15%)
• Final (40 %)
Homework assignments to be submitted at the beginning of class on the due
date. Submissions must follow the following format: the answer is clearly dis-

2
played first, demarcated, followed by a justification written in clear, concise En-
glish. Mangled telegrams and streams of consciousness should be avoided. Marks
are deducted for submissions that don’t follow this format, are hard to read or
incomprehensible. No late work will be accepted. If you cannot attend class,
you may email me your homework (but I prefer hard copy). Some homework will
require you to download data files from CANVAS.

Review Sessions
Held each Friday and used to go over problems in the review packet. The
problems to be covered each week will be posted on CANVAS. For maximum
benefit, you should attempt these problems before attending the review sessions.

Course Material
This course will make use of three pieces of software: an EXCEL add-in called
SOLVER (so familiarity with the basics of spreadsheets is assumed), MATLAB
and @Risk. See CANVAS for instructions on how to access @Risk.
There is no textbook for this course. However, there are a wealth of on-line
resources that are good (and possibly better) substitutes. You will find links to
them in the various handouts for the class located on CANVAS.
Required for the course is a case packet available from Harvard Business
School Press that needs to be purchased (around $12). To purchase this case
packet you must go to:
http://cb.hbsp.harvard.edu/cbmp/access/65685217
CANVAS will be used for posting announcements, homework assignments,
homework solutions, additional handouts, and other important materials. Note, I
do not post slides ahead of time. They frequently contain questions and answers
that I wished discussed in class rather than digested before hand. You are respon-
sible for regularly checking, downloading, and reading materials posted on the site,
as they form an integral part of the class.

Laptop Use
Since we employ software in most sessions, you may use your laptop in class.
However, this is not a carte blanche to employ it to distract classmates by tracking
your portfolio, instant messaging, tweeting, blogging, booking a flight or updating
yourself on the whereabouts of Brangelina.

3
Office Hours and Contact Info
The TA for the class will hold one office hour a week. The day and time will be
posted on CANVAS. My office hours will be Wednesday, 10-11 am in 408B, 3410
Walnut Street.

Academic Integrity
Students are expected to abide by the Code of Academic Integrity in the com-
pletion of assignments, papers and exams.

Assignments
Write-ups must be your original work. You may not use materials containing
solutions or partial solutions to the assignments (including solutions prepared by
current or former students). If your analysis contains information from outside
sources, then you must properly cite the sources. You may not discuss the group
assignments with anyone not a member of your group, other than me.
As in many classes, you will be asked to complete the assignments individually.
Yet, one does not wish to discourage encourage learning from one’s peers. This
leaves room for ambiguity, so I will try here to make expectations as clear as pos-
sible. In brief:

1. Discussing the general ideas behind the problems is permitted.

2. Writing formal solutions should be completely individual, done in the equivalent

of separate rooms.

These guidelines leave a gray area as discussions of general ideas gradually be-
come more specific. Some judgment calls are unavoidable, but here’s the kind of
interaction I have in mind: If a peer conveys an idea which seems to be central to
the solution, do not write it down.....immediately. Approach the problem again on
your own as if afresh, influenced by however much of their idea you remember. If
you can re-create it without notes, you have mastered it, and I’m happy to give
you credit. In this way we can let everyone help each other learn, while steering a
wide berth around simple copying.

Exams
No assistance may be given or received during an exam. Regardless of when you
take the exam, you may not discuss the exam with any other person before the

4
graded exams are returned (in case there are students who have not yet taken
the exam). Even casual statements, such as it was easy or it was hard are not
permitted. Exams must be completed within the allotted time and only approved
materials may be used.
Exams are open book. This means they will not be memory tests. Exam at-
tendance is mandatory. Students who miss an exam for an allowable reason must
report their absence on the Course Absence Reporting (CAR) System. There will
be no make-up exams; students who are excused from an exam will see the weights
on the remaining exams and final adjusted upwards to account for the absence.

Course Outline
Week 1 (August 29 & 31): Introduction to Linear Program-
ming
Read the document morpheus.pdf before the August 29 class. Come to class
prepared to discuss possible solutions to the problem posed.
Prepare the Environmental Protection Case for class discussion for August 31.
The document fomulhandout.pdf summarizes the weeks discussion.
Week 2 (Sept 5 & 7): Use of EXCEL SOLVER, Interpreting
Output & the Dual
Prepare the Ricardo-Bond Co. case for discussion for Sept 5.
The document excelTutorial.pdf summarizes the use of the EXCEL SOLVER. The
document EXCEL Sensitivity Analysis as well as sensanalysis.pdf summarize how
to interpret the SOLVER output.
Week 3 (Sept 12 & 14): Matrix-Vector Notation & Exam
#1
Bring the document formulationexercises.pdf to class.
Week 4 (Sept 19 & 21): MATLAB & Integer Programming
For help with MATLAB consult the relevant portion of online.pdf.
Bring the document formulationexercises.pdf to class.
Week 5 (Sept 26 & 28): Integer Programming & Network
Optimization
Bring the document formulationexercises.pdf to class.
Week 6 (Oct 3): Network Optimization and Non-linear Pro-

5
gramming
Bring the document formulationexercises.pdf to class.
Week 7 (Oct 10 & 12): Introduction to Probability
Read sections 1-3 of Prob1.pdf before the Oct 12 class.
Week 8 (Oct 17 & 19): Introduction to Probability & Exam
#2
This week we start using the package @RISK. The document @Risk Tutorial.doc
provides a guide to @RISK. Instructions for accessing it can be found in this
document. You can also download it directly onto your device (Windows only).
Instructions on CANVAS. It will also be available in the computer lab.
Week 9 (Oct 24 & 26): Random Variables, Distributions,
Expectation, Variance
Prepare Walton book case (WaltonBook.xlsx) for Oct 24
Read the remainder of Prob1.pdf after the Oct 24 class.
Week 10 (Oct 31 & Nov 2): Monte Carlo Simulation &
Central Limit Theorem
The handout Prob2.pdf summarizes the technical content of this week. The
document ExcelHandout.pdf describes in detail the various EXCEL commands in-
troduced in Prob2.pdf.
The document CLT.pdf summarizes the Central Limit Theorem.
Week 11 (Nov 7 & 9): Mean-Variance Portfolios & Decision
Trees
The document Prob3.pdf summarizes Mean-Variance Portfolios.
Week 12 (Nov 14 & 16): Decision Trees & Exam #3
Prepare the Weston Manufacturing Case for discussion on Nov. 14.
For an extensive discussion of decision trees see the folder DecisionTreePrimer.
Week 13 (Nov 21): Bayesian Updating & Value of Informa-
tion
Prepare the Sellers Razor exercise for discussion for Nov. 22.
The handout Prob4.pdf has a brief summary of utility functions.
Handout Prob5.pdf covers Bayesian updating.
Week 14 (Nov 28 & 30): Options & Stochastic Programming
Prepare the Cost Uncertainty case for discussion for Nov 28.

6
Week 15 (Dec 5 & 7): Stochastic Programming

7
Course Calendar
These dates are not fixed in stone. I reserve the right to change them to adjust
to the pace of the class.
• Sept 7: Homework 1 due
• Sept 14: Exam #1 in class
• Sept 28: Homework 2 due
• Oct 5: No class (fall break)
• Oct 9: Drop period ends
• Oct 12: Homework 3 due
• Oct 19: Exam #2 in class
• Nov 2: Homework 4 due, Group Assignment (Omega Oil) due.
• Nov 9: Homework 5 due
• Nov 10: Last day to withdraw
• Nov 16: Exam #3 in class
• Nov 23: No class (thanksgiving)
• Nov 30: Homework 6 due
• Dec 7: Homework 7 due
• Dec 19, 12-2 pm: Final Exam.