Due January 19, 2010
1. Analyze the Hawthorne Plastics case. Give the complete decision tree and carry out
any extra analyses you think may be relevant such as computing the value of information
and exploring the sensitivity of decisions to changes in key model parameters. Use the
questions following the case for guidance. Summarize your findings in a brief
(maximum one page) report.
2. Consider the ball and urn problem in lecture 2 notes. Evaluate the other two options
on slide 48.
a. What is the optimal course of action and what is the most you would pay to draw two
b. Is it better to sample with or without replacement? Why?
3. Consider the following variants of the newsvendor problem. It might be convenient it
set up a simple spreadsheet like the one I used for the Alberta case.
a. Suppose the newsvendor has 5 potential customers and each week, each week, each
customer, independently decides whether or not to purchase one item with probability p.
Consequently the demand has a Binomial distribution with n=5 and success probability p.
Suppose you do not know p and instead assume that it has a beta distribution with
parameters α = 2, β = 2. How much should you order? Assume that the selling price for
the item is $5000, the cost to the newsvendor of the item is $2000 and the salvage value
b. How will your order quantity change if α increases?
c. What is the expected value of perfect information and what does it mean?
d. Suppose you can observe one week’s demand prior to choosing the order quantity.
Give the optimal decision under each realization of demand. What is worth to observe
this demand before choosing your order quantity?
e. Risk averse version. Suppose the newsvendor’s utility function is 1-e-c(x + 7500) with c
equal 1/1000 and 1/5000. Repeat your analysis in parts a., c. and d. and compare you
results to those above with respect to the optimal policy and the EVPI.
Version January 10, 2010