Business Decision Analytics Under Uncertainty (33:136:400) 
Professor Eckstein
A Bayesian Decision Tree with Three "States of the World"
(Former Midterm Problem)

Your firm currently holds the mining rights to a plot of land in southern Alaska , which is suspected to contain valuable cobalt deposits.  You estimate there is 25% chance the plot of land will be an “excellent” source of cobalt, in which case you would make a $24 million profit from developing it.  There is a 50% chance it will be “good” source of cobalt, in which case you would earn a profit of $9 million from developing it.  Otherwise, it will be a “poor” source of cobalt, and developing it would yield a loss of $12 million.  If you do not develop the plot, you can sell it for $4 million.

 Before deciding whether to develop or sell the plot, you may conduct a test drilling program, which will cost $2 million.  The results of the program may be either “promising” or “unpromising”.  In the past, test drilling has yielded a “promising” result for 96% of the tested plots that turned out to be “excellent”, 60% of the tested plots that turned out to be “good”, and 24% of the tested plots that turned out to be “poor”. 

 By law, the results of the test drilling program must be made public.  A “promising” result would increase the sales value of the plot, if you don’t develop it yourself, to $10 million.  An “unpromising” result would lower the sales value of the plot to $1 million. 

What should you do to maximize the expected value of the plot of land? What is the EVSI of the test drilling program?