Freemark Abbey:

The following are short possible answers to the questions posed in the case. Obviously your answers may differ from these and hopefully discuss the case more fully than just to answer the questions. The payoffs are revenues associated with various decisons and eventualities --since if storm hits and mold forms there is a loss of yield. The payoffs are calculated based on 1000 cases of wine.
  1. At the outset, there are two alternatives: (1) wait (W) for the storm to hit, if it will; (2) Harvest (H) now. If waited, the storm may or may not hit. If it hits, the mold may or may not form, greatly affecting the revenue. If the storm hits and no mold forms, the eventual sugar level, which determines the quality of the wine and hence the revenue, is uncertain. Likewise the uncertainty  about the mold is quite important as it determines the difference between an exquisite wine versus a thin wine that the winery may be unwilling to risk their reputation by selling it under company label.  This calls for another decision (if storm hits and the mold does not form): whether to sell the grapes in bulk (B) or make the thin wine and sell it under label (T). This decision however is different in that, the EV criterion will always favor making the wine since EV maximizes the current returns but cannot consider the long term marketing impact of selling the thin wine now. Thus this decision must be made outside the framework of the sequential decision based on monetary EV.



    2. The blue values are based on the assumption of selling thin wine in bulk, while the red numbers assume the thin wine will be bottles and sold for $2.00/bottle.
    According to the above decision tree, solved on the assumption of (B), waiting for the storm has an EV of $35,640 versus (H) with an inferior EV of $34,200. Admittedly the difference is rather small (about 4.2% higher). If (T) is used the EV of (W) goes up to $39,240 which is substantially more. However one has to consider the negative long term strategic effects of (T).
     

  2. If the probability of the  mold is only 20%
        a) On the assumption that the thin wine will be bottled:  the (W)'s  EV drops  from $39,240 to $34,920  leaving (W) still better than $34,200. However, since the difference is now quite minimal, the sure alternative (H)
            may be given consideration.
          b) On the assumption that the thin wine will be sold in bulk  :  (W)'s EV drops  from $35,640 to $30,120 making (H), the better alternative, not only in higher EV as well as lower risk.
  1. Again assuming (B) is used, EVPI (storm) can be calculated as follows: EV with prior knowledge of storm minus EV without such knowledge. If it was known that the storm would hit, the best alternative is (H) with expected revenue of $34,200 (versus $34,080); however if it were clear the storm would not hit then it is better to wait with EV of $37,200 (vs. $34,200). Thus with the prior knowledge, the EV = .5 * 34,200 + .5 * 37200 =  $35,700. Without prior knowledge however the EV (from part 2 above) is $35,640. Thus the EVPI (storm) = $35,700-$35,640 = $60.  A similar calculation can be made with the assumption of (T)-- interestingly the EVPI (storm) is nil. With this assumption regardless of wether the storm is known to come or not the better alternative is to wait (41,280 vs. 34,200 and 37,200 vs 34,200) yielding the same EVof 39,240.