Well I'm back from a short trip, and pleasantly surprised to see the strength in the biotechs compared with Nasdaq, especially after last week's unpleasantness.
Chiron's good-looking results might help the sector today if the overall market isn't too weak. The first-tier profitable biotechs ultimately set the overall pricing for the sector (except perhaps for the genomics stocks). For the last few years the market has been prepared to pay a (perhaps overly-generous) premium for consistent growth and no unpleasant quarterly "surprises." This provided the pharmas and the first-tier biotechs (like AMGN) with their huge gains in the 90's.
Early in this thread I discussed the issue of how to value earlier-stage biotechs. My view is that the straight discounted cash flow analysis that is the standard basis for rational valuation doesn't work very well with early-stage biotechs. The trouble is that it conflates two issues - the timing of future payment streams and the riskiness of those streams.
As a good example, consider a new Sepracor drug candidate. Let's take say (R,R)-formoterol, their quick-acting and long-acting asthma drug candidate. Now this is scheduled for 2003 rollout, but to be even more dramatic let us assume we are doing this analysis in 1998. So at that point we are at least 5 years from revenue-flow, which typical analysts discount back at a hefty discount rate (say 75%) to account for the "riskiness" of drug development. Now that values $100 in revenue in the first year (2003) at a mere $6.09 in 1998. By the time the drug is hitting its peak (after say 2007) you are discounting back 10 years and more at a high rate. This makes the net present value extremely sensitive to changes in the discount rate you use.
In my view, the better way to look at it is to assign probabilities to various outcomes - here say 15% to failure (because it's a simple variant of a known European-approved widely used drug), 25% to moderate success (peak sales of $200 million), 40% to considerable success (peak sales of $400 million) and 20% to blockbuster status (peak sales of $800 million or more). One then looks at what the market is presently capitalizing such drugs at (perhaps 15x to 20x sales) and figure out what the market cap attributable to the drug will be at peak. You then discount this back at a low rate (say 10%) to account for the time value of money.
Obviously one can quarrel with the particular numbers I've guessed here, but I think this methodology is the correct one to use.
You have to adapt this methodology to use it for a cyclical stock (like a semiconductor company). The market applies a very different multiple to earnings at the peak versus in the trough, and of course you have to decide where in the cycle you currently are. (Not easy to do of course, but probably easier than trying to estimate the probability of a drug candidate succeeding).
Now the one saving grace is that when you make your estimates you don't actually have to be right - you just have to be more right than the market. <g>
(Like the old bear story that they used to trot out to incoming Harvard Law students:
A bear is attacking a tent with two guys in it and is clearly just about to break in. The one guy takes time out to put on his shoes and his companion scoffs at him saying "That's not going to help you outrun the bear."
The reply comes quickly: "I don't have to outrun the bear - I just have to outrun you!"
Peter |
