Decision Making II

Quiz 2
Maximum time allowed 45 minutes Problem: [3+3+2+3+3+3+3= 20 points] The Mektek Company has decided to introduce a smartphone as well. They have two choices, either to develop the user interface in-house, or to buy the user interface. They also have two options to find out about the market for their smartphone; they can conduct an independent market survey at a cost of Rs.15 lakhs, or obtain a confidential market survey report for Rs.10 lakhs on the acceptability of similar smartphones. The independent market survey predicts one of three demand scenarios: high, medium or low. In a high demand situation, If Mektek develops the user interface in-house, they make a profit of Rs.500 lakhs if the demand is high, a profit of Rs.100 lakhs if the demand is medium, a loss of Rs.200 lakhs if the demand is low. If they buy the user interface, then they make a profit of Rs.200 lakhs if the demand is high, a profit of Rs.120 lakhs if the demand is medium, a loss of Rs.100 lakhs if the demand is low. Historically such surveys have been inaccurate. The following table describes the inaccuracies observed historically given the survey predictions. Demand scenario actually observed High demand Medium demand Low demand 40% of the situations 30% 30% 20% 60% 20% 10% 10% 80%

Prediction High demand Medium demand Low demand

Based on Mekteks beliefs about the probability of high, medium, and low demand scenarios, and their knowledge about the accuracy of the market survey results, Mektek estimates that the independent market survey will predict high, medium or low demand scenarios with probabilities 0.4, 0.2, and 0.4 respectively. The company producing the confidential market survey report predicts one of two scenarios: success and failure. A success prediction corresponds to either high or medium demand scenarios, and a failure prediction corresponds to a low demand scenario. It is known that this company predicts success 80% of the times when the actual demand scenario is either high or medium, but it predicts failure only 70% of situations in which the demand is actually low. a) Based on MekTeks predictions about the results from the independent market survey, what are the probabilities that Mektek assigns for high, medium and low demand scenarios for the smartphone? In this question, we are required to find the prior probabilities. Note that the marginal

probabilities for the three demand scenarios have been given, along with the posterior probabilities. From these, we can obtain the joint probabilities by multiplying each of the posterior probabilities with the corresponding marginal. Given joint probabilities, the prior probabilities can be found by summing the columns. The computations are tabulated below. Posterior prob. Pred Hi Pred Med Pred Lo Hi Med Lo Marginal 0.4 0.3 0.3 0.4 0.2 0.6 0.2 0.2 0.1 0.1 0.8 0.4 Hi Med Lo 0.16 0.12 0.12 0.04 0.12 0.04 0.04 0.04 0.32 0.24 0.28 0.48 <-- Priors

b) Assuming that Mektek takes a decision without the aid of any market survey, independent or confidential, and assuming that they use the expected value criterion, should MekTek develop the user interface in-house or should they buy the user interface? What is the expected profit of their optimal decision strategy? In this question, given prioi probabilities, we need to compute the expected profit for each decision, and choose the best. The expected profit for In-house development is 500*0.24+100*0.28-200*0.48 = 52 lakhs while that of buying is 33.6 lakhs. So they should develop the user interface in-house, and the expected profit is 52 lakhs. c) What is the expected value of perfect information in part (b)? With perfect info, expected profit = max{500,200}*0.24 + max{100,120}*0.28 + max{-200, -100}*0.48 = 105.6 lakhs. So EVPI = 105.6 52 = 53.6 lakhs. d) Assume for this part that MekTek uses the independent market survey, and the survey predicts a medium demand. Should they develop the user interface in-house or should they buy the user interface? What is MekTeks expected profit for their optimal decision under this circumstance? Under this condition, the probabilities of the three scenarios are 0.2, 0.6 and 0.2 respectively. So the expected profits for both choices have to be computed using these probabilities. The expected profit for developing in-house becomes 120 lakhs and that of buying the interface is 92 lakhs. So they should develop the user interface in-house for an expected profit of 120 lakhs. e) Should Mektek pay Rs.15 lakhs for the independent market survey? What is the maximum amount that MekTek can consider spending for the independent market survey? In the situation that the survey says that the demand is High (prob = 0.4) we have to use the corresponding posterior probabilities (0.4,0.3,0.3) to compute the best decision. With these probabilities, the best decision would be to develop the interface in-house yielding an expected

profit of 170 lakhs. Using the same argument, the best expected profit if the survey says medium demand (prob = 0.2) is 120 lakhs and if the survey says low demand (prob = 0.4) is -48 lakhs. So the EVSI is (170*0.4 + 120*0.2 -48*0.4)-52 = 20.8 lakhs. Since this is more than 15 lakhs, they should spend the amount for the survey. The maximum amount that they can consider spending on the survey is 20.8 lakhs. f) Assume for this part that MekTek uses the confidential market survey report, and the report predicts failure. Should they develop the user interface in-house or should they buy the user interface? What is MekTeks expected profit for their optimal decision under this circumstance? The answer to this part depends on the assumption taken. We have to reduce the three scenarios into two corresponding to the success and failure conditions for the confidential survey report. A natural way to do this is to take the expected profits for the high and medium demand scenarios as the profits corresponding to the success condition and the profits for the low demand scenario for the failure condition. [There are other ways to merge these profit values, we will see your assumptions to check if they are valid.] Under these conditions, the payoff table looks like: Payoffs Inhouse Bought Success Failure 284.62 -200 156.92 -100

If the report predicts failure, then the expected profit if you develop the interface in-house is 0.3*284.62-0.7*200 = -54.62 lakhs while that if the interface is bought is -22.92 lakhs. So the optimal decision here is to buy the interface and the expected payoff is -22.92 lakhs. g) Should Mektek pay Rs.10 lakhs for the confidential market survey report? What is the maximum amount Mektek can consider spending for the confidential market survey report? The answer to this part depends on the assumption taken. We use the same argument as in the previous part. Here we first have to compute the probability with which the report will say success. Let that probability be p. Then the joint probability of actual success and the report saying success is 0.8p, while that of actual success where the report says failure is 0.3(1-p). The probability of actual success is 0.24, so that 0.8p + 0.3(1-p) = 0.24, solving which we find that p = 0.44. The EVSI computation for this part is identical to that in part (e) and we find that the EVSI is 17.74 lakhs. So they should pay 10 lakhs for the report. The maximum that they should be prepared to pay is 17.74 lakhs.