To: IRWIN JAMES FRANKEL who wrote (5248 ) 2/7/2002 8:25:03 PM From: A.J. Mullen Respond to of 10280 Irwin, Thanks very much for sharing the results of your work. Coincidentally I recently received a J.P. Morgan analyst's report on Celltech using this method. The author, Damian Pethica, makes the same point that you note: when modelling risk explicitly, more modest discount rates should be used. I wrote briefly to Dr. Pethica to say his was the first professional analysis I had seen that seemed worthy of the term. I added that this sort of explicit modelling would allow one to put a range of confidence on one's valuation; one could perform Monte-Carlo simulations. Say there were three drugs with an expected value of $100 million, each with a probability of success of 1/6. The expected value for all three is 3x 100/ 6 = $200 million, but it could range from 0 to $300 million. A simulation would consist of throwing three dice, counting the number of sixes, and multiplying by $100. If that were done many times then the result would be zero rarely; once in 6x6x6 times. The frequency of scoring a hat-trick - all three drugs coming to market would be just as rare. I chose simple numbers. It is easy to perform these simulations with any numbers, but I don't think it's necessary in this case . My point is that if one did this for Sepracor the range would be pretty tight, because Sepracor has many irons in the fire - assuming the probabilities of success for each drug is independent, and that is debatable*. This suggests to me there is less risk to Sepracor than to other biotechs with similarly computed valuations. Ashley *After an FDA disappointment, Peter suggested that we may have systematically overestimated the chances of chirally pure isomers being approved. This implied the chances of success for each molecule were not independent.