I've already written about (and provided some support for) the shortcomings of DDM and its extremely limited ability of pegging anything close to a reasonable approximation of a valuation range.
The simplest reason I can think of for chucking DDM is that there are too many moving parts for DDM to be of any real use. For a one stage DDM, you need to precisely estimate Re (impossible, sorry), and you need to reasonably estimate g. You could estimate g by solving for 1/payout x ROE, but again, g changes over time because ROE and retention change over time.
If you thought estimating a one stage DDM was difficult, try estimating a two stage DDM! You need two Re's, two g's and who knows what else. And on top of all of the above, you have to extrapolate the dividends out perpetually...
I recall seeing one attempted analysis where a blogger used current annual dividend, grew this at X% for the first n years, .35 x X% terminally, and discounted the entire enchilada at 10% to come up with a DDM valuation in excess of 50% above the stock price at the time.
Hypothetically, if you know that a patent cliff is coming in six years and cash flows are going to get cut in half, would you still discount the perpetual post-patent cash flows at 10%? These are riskier cash flows.
I started thinking about Gilead, because it's sort of an enigma wrapped in a puzzle. The growth has been phenomenal on Sovaldi and Harvoni (which together account for in excess of 50% of consolidated sales). So why then, does Gilead seem so cheap (it commands a fwd PE of about 8.5x 2016 estimated earnings)?
Well, when something looks too cheap to be true, and everyone in the market is pounding the same table, it's probably cheap for good reason.
I think that any well thought analysis has to consider the following questions:
- What is the estimated market for Sovaldi and Harvoni? How big is the market, now, and how big is the market expected to be? Obviously, there is a limit to how big the market is. It's equivalent to patients with (or expected to contract) liver disease / HEPc, etc. Maybe one way of estimating the market is to look at historical liver disease/Hepc stats over recent decades as a %ge of population globally. Once you estimate the market, you can forecast future expected growth
- What is the estimated market for the remaining drugs?
- Take results of 1. and 2. above and grow expected cash flows as profit margin per drug/treatment x a conservative estimate of target patient growth between now and patent expiry (GILD lists all key patent expiry dates in the 10K). Beyond patent expiry, generics compete and the cash flows change dramatically
- Consider whether the recent growth in revenue is a result of "limitless demand" (as most likely extrapolated by Wall Street) vs. "pent-up demand". If due to pent-up demand, do near term cash flows grow at the same rate?
- Do a sensitivity of 3. at different pricing tiers (it's no secret that GILD's pricing has come under scrutiny). What happens to free cash flow estimates between now and patent expiry if margins come under pressure due to exogenous or competitive factors. See this recent WSJ article on current pricing pressure.
I was reading an article from 2010 published by Greg Speicher on Michael Price, who valued pharmaceutical companies along the following lines:
"How He Values Pharmaceuticals
This is exactly how one has to think along the lines of Gilead. No DDM necessary.